Builder ffmpeg64-solaris10-i386 Build #14066

Results:

Failed shell_2 shell_3 shell_4 shell_5

SourceStamp:

Project	ffmpeg
Repository	https://git.ffmpeg.org/ffmpeg.git
Branch	master
Revision	9fc8c785e2747c87121ec28f8f10ceab0562384b
Got Revision	9fc8c785e2747c87121ec28f8f10ceab0562384b
Changes	1072 changes

BuildSlave:

unstable10x

Reason:

The SingleBranchScheduler scheduler named 'schedule-ffmpeg64-solaris10-i386' triggered this build

Steps and Logfiles:

git update ( 35 secs )
1. stdio
shell 'gsed -i ...' ( 0 secs )
1. stdio
shell_1 'gsed -i ...' ( 0 secs )
1. stdio
shell_2 'gsed -i ...' failed ( 0 secs )
1. stdio
shell_3 './configure --samples="../../../ffmpeg/fate-suite" ...' failed ( 7 secs )
1. stdio
2. config.log
shell_4 'gmake fate-rsync' failed ( 0 secs )
1. stdio
shell_5 '../../../ffmpeg/fate.sh ../../../ffmpeg/fate_config_64.sh' failed ( 0 secs )

Build Properties:

Name	Value	Source
branch	master	Build
builddir	/export/home/buildbot/slave/ffmpeg64-solaris10-i386	slave
buildername	ffmpeg64-solaris10-i386	Builder
buildnumber	14066	Build
codebase		Build
got_revision	9fc8c785e2747c87121ec28f8f10ceab0562384b	Git
project	ffmpeg	Build
repository	https://git.ffmpeg.org/ffmpeg.git	Build
revision	9fc8c785e2747c87121ec28f8f10ceab0562384b	Build
scheduler	schedule-ffmpeg64-solaris10-i386	Scheduler
slavename	unstable10x	BuildSlave
workdir	/export/home/buildbot/slave/ffmpeg64-solaris10-i386	slave (deprecated)

Forced Build Properties:

Name	Label	Value

Responsible Users:

Abdessamie
try70457ohnoyoudont@gmail.com
Ackanir
git.ackanirohnoyoudont@gmail.com
Aleksander Shveikin
alex.svekohnoyoudont@gmail.com
Andreas Menze
andreas.menzeohnoyoudont@hr.de
Andreas Rheinhardt
andreas.rheinhardtohnoyoudont@outlook.com
Anxious Koisi
ddoaoingohnoyoudont@gmail.com
Araz Iusubov
Primeadviceohnoyoudont@gmail.com
Ayoub Nabil
237098474+ayoubnabilohnoyoudont@users.noreply.github.com
Ayoub Nabil Boubagrat
237098474+ayoubnabilohnoyoudont@users.noreply.github.com
Brad Smith
bradohnoyoudont@comstyle.com
Brian Leake
phydohnoyoudont@phyd.org
Charly Morgand-Poyac
charly.morgand-poyacohnoyoudont@bbright.com
Christopher Decker
chris.decker08ohnoyoudont@gmail.com
Cloud-LHY
securityohnoyoudont@clouditera.com
Dan Dennedy
danohnoyoudont@dennedy.org
Daniel Verkamp
danielohnoyoudont@drv.nu
David Rosca
david.roscaohnoyoudont@amd.com
Diego de Souza
ddesouzaohnoyoudont@nvidia.com
Dmitrii Gershenkop
bortengineerohnoyoudont@protonmail.com
Frank Plowman
postohnoyoudont@frankplowman.com
Gabriel Balaich
ffmpegohnoyoudont@ninbura.com
Guo Yejun
yejun.guoohnoyoudont@intel.com
Hans-Kristian Arntzen
postohnoyoudont@arntzen-software.no
Hao Chen
haochencohnoyoudont@google.com
Haochen Chen
haochencohnoyoudont@google.com
Hawthorn
rv.help23ohnoyoudont@gmail.com
Hunter Kvalevog
hunterohnoyoudont@kvog.sh
J. Brandt Buckley
brandtohnoyoudont@runlevel1.com
Jack Lau
jacklau1222gmohnoyoudont@gmail.com
Jamaika1
lukaszcz18ohnoyoudont@wp.pl
James Almer
jamrialohnoyoudont@gmail.com
Jeongkeun Kim
variety0724ohnoyoudont@gmail.com
Jiale Yao
yaojiale02ohnoyoudont@163.com
Joshua Rogers
MegaManSecohnoyoudont@users.noreply.github.com
Julius Bairaktaris
juliusohnoyoudont@bairaktaris.de
Jun Zhao
barryjzhaoohnoyoudont@tencent.com
KO Myung-Hun
komh78ohnoyoudont@gmail.com
Kacper Michajłow
kasper93ohnoyoudont@gmail.com
KawaiiEngine (Sashimiso)
209421934+KawaiiEngineohnoyoudont@users.noreply.github.com
Kenan Alghythee
kalghy2ohnoyoudont@uic.edu
Kirill A. Korinsky
kirillohnoyoudont@korins.ky
Kostazol
kostazolohnoyoudont@users.noreply.github.com
Leo Izen
leo.izenohnoyoudont@gmail.com
Lynne
devohnoyoudont@lynne.ee
Marcos Ashton Iglesias
marcosashiglesiasohnoyoudont@gmail.com
Martin Storsjö
martinohnoyoudont@martin.st
Marvin Scholz
epirat07ohnoyoudont@gmail.com
Michael Niedermayer
michaelohnoyoudont@niedermayer.cc
Michael Partridge
michael.partridgeohnoyoudont@petrosys.com.au
Mirkó Visontai
mirkovohnoyoudont@google.com
Nathan Lucas
nlucasgitohnoyoudont@gmail.com
Nicolas Gaullier
nicolas.gaullierohnoyoudont@cji.paris
Nicolas George
georgeohnoyoudont@nsup.org
Niklas Haas
gitohnoyoudont@haasn.dev
Nil Fons Miret
nilfohnoyoudont@netflix.com
Omkhar Arasaratnam
omkharohnoyoudont@linkedin.com
Paulo Cabral Sanz
paulosanzohnoyoudont@poli.ufrj.br
Pavel Kohout
disclosureohnoyoudont@aisle.com
Philip Langdale
philiplohnoyoudont@overt.org
Raja-89
imraja729ohnoyoudont@gmail.com
Ramiro Polla
ramiro.pollaohnoyoudont@gmail.com
Reda DRISSI
reda.drissiohnoyoudont@yahoo.fr
Romain Beauxis
romain.beauxisohnoyoudont@gmail.com
Scott Kidder
scottohnoyoudont@kidder.io
Shiyou Yin
yinshiyou-hfohnoyoudont@loongson.cn
Shreesh Adiga
16567adigashreeshohnoyoudont@gmail.com
StachowiakDawid
dawidos234ohnoyoudont@gmail.com
Steve Lhomme
robux4ohnoyoudont@ycbcr.xyz
Steven Liu
lqohnoyoudont@chinaffmpeg.org
Sylvestre Ledru
sylvestreohnoyoudont@debian.org
Thomas Guilbert
tguilbertohnoyoudont@chromium.org
Théo Valette
315040668+theovaletteohnoyoudont@users.noreply.github.com
Timo Rothenpieler
timoohnoyoudont@rothenpieler.org
Tobias Rapp
t.rappohnoyoudont@noa-archive.com
Tristan Matthews
tmatthohnoyoudont@videolan.org
Tymur Boiko
tboikoohnoyoudont@nvidia.com
Umar Pathan
helloohnoyoudont@umar.ac
Vignesh Venkat
vigneshvohnoyoudont@google.com
Vladimir Panteleev
gitohnoyoudont@cy.md
Vladimir Panteleev
gitohnoyoudont@thecybershadow.net
Wu Jianhua
toqsxwohnoyoudont@outlook.com
WyattBlue
wyattblueohnoyoudont@auto-editor.com
Zhao Zhili
quinkblackohnoyoudont@foxmail.com
Zhao Zhili
zhilizhaoohnoyoudont@tencent.com
Zsolt Vadasz
zsolt_vadaszohnoyoudont@protonmail.com
Zuxy Meng
zuxy.mengohnoyoudont@gmail.com
adorn
adornohnoyoudont@hr.de
almogyalin
almog.yalinohnoyoudont@gmail.com
deng.zewen
deng.zewenohnoyoudont@zte.com.cn
guy-with-a-why
aaronohnoyoudont@iterationlab.io
iSold Leo
sohnoyoudont@qwqlog.com
jiangjie
jiangjieohnoyoudont@agora.io
lck
luckyskohnoyoudont@foxmail.com
marcos ashton
marcosashiglesiasohnoyoudont@gmail.com
ninbura
gabrielohnoyoudont@ninbura.com
panboxiaosa
panboxiaosaohnoyoudont@gmail.com
qaq03101
1440810736ohnoyoudont@qq.com
sfan5
sfan5ohnoyoudont@live.de
shadowcaster3
vasyl.samoilovohnoyoudont@gmail.com
softworkz
softworkzohnoyoudont@hotmail.com
tangsha
tangshaohnoyoudont@kylinos.cn
tknaveen
unbornboron69ohnoyoudont@gmail.com
unfunnyatearug
yalisib2222ohnoyoudont@gmail.com
wang-bin
wbsecg1ohnoyoudont@gmail.com
wangbin
wbsecg1ohnoyoudont@gmail.com
wangnov
wangnov6ohnoyoudont@gmail.com
yongdev
yongdevohnoyoudont@google.com
younengxiao
steven.xiaoohnoyoudont@amd.com
yuanhecai
yuanhecaiohnoyoudont@loongson.cn
zhanghongyuan
zhanghongyuanohnoyoudont@uniontech.com
zhorzhetta1404-ux
zhorzhetta1404ohnoyoudont@gmail.com
张永鹏
roc63ohnoyoudont@outlook.com
罗乾林
luoqianlinohnoyoudont@sansi.com

Timing:

Start	Thu Sep 3 14:50:27 2026
End	Thu Sep 3 14:51:14 2026
Elapsed	47 secs

All Changes:

:

Change #273322

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Tue 30 Jun 2026 10:23:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ae4314e2f4618b8b4d36864fdacc03ea0e85a1f6

Comments

tests: Implicitly set TAG=GEN for data/*.m3u8
This avoids having to declare it for every rule that generates
test input data, and adds such a missing tag for
tests/data/hls-seek.m3u8.

This fixes a case of a missing TAG for data/hls-seek.m3u8, which
caused it to be printed without a tag, like
"        tests/data/hls-seek.m3u8", now printed as
"GEN     tests/data/hls-seek.m3u8" as expected.

Changed files

tests/Makefile
tests/fate/filter-audio.mak
tests/fate/hlsenc.mak
tests/fate/segment.mak

Change #273348

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 30 Jun 2026 13:05:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	59a5090070d094029cc328f6160b87a9f214c01d

Comments

avfilter/af_adeclick: free transform contexts and buffers on error
Found-by: Ao Xijie
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/af_adeclick.c

Change #273349

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 30 Jun 2026 13:15:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9a01c1cb6a4cf87529fe9898b66ec55c5b032639

Comments

avfilter/vf_removelogo: free the whole mask on allocation failure
Found-by: Ao Xijie
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_removelogo.c

Change #273391

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 30 Jun 2026 19:29:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b553b810c460e484f70a7368ed7abbdcecbe845b

Comments

avformat/movenc: free sgpd_entries on the roll-distance error path
Found-by: Ao Xijie
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/movenc.c

Change #273407

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 30 Jun 2026 22:50:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f0bfebc9adce90d7c3d44f1a081a705389995b34

Comments

avcodec/x86/hevc/add_res: Port ff_hevc_add_residual_4_8_mmxext to SSE2
No change in performance here.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/hevc/add_res.asm
libavcodec/x86/hevc/dsp.h
libavcodec/x86/hevc/dsp_init.c

Change #273433

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 01 Jul 2026 03:19:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b1327568b6cd5b244e4ed0488ee1ed645ac4a552

Comments

avformat/movenc: unify remainder samples signaling
When writing edit lists and the last packet has skip samples side data to
report discard padding, for codecs that have a fixed frame size it's best to
always use the full packet duration to calculate track duration (and thus
bitrate in some cases) and let the edit list be the sole source of triming
information.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/movenc.c
tests/ref/fate/autorotate
tests/ref/fate/mov-mp4-iamf-5_1_4
tests/ref/fate/mov-mp4-iamf-7_1_4-video-first
tests/ref/fate/mov-mp4-iamf-7_1_4-video-first-2
tests/ref/fate/mov-mp4-iamf-7_1_4-video-first-3
tests/ref/fate/mov-mp4-iamf-7_1_4-video-last
tests/ref/fate/mov-mp4-iamf-ambisonic_1
tests/ref/fate/mov-mp4-iamf-stereo

Change #273434

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 01 Jul 2026 03:48:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d5a2ab9c08980b74dce93f55da9d95a7eefef11d

Comments

checkasm/sw_rgb: fix too small source stride in uyvytoyuv422 test
The planes table stored a source stride smaller than the 2*width bytes a
packed UYVY line occupies (e.g. width 12 with stride 12), and the correct
stride for width 128 would be 256, which does not even fit the uint8_t
field. The test passed only because the oversized source buffer absorbed
the resulting out-of-bounds reads.

Derive srcStride from the width (2*width) instead of storing it, so each
line is passed its true size.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

tests/checkasm/sw_rgb.c

Change #273435

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 01 Jul 2026 03:48:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ef18ece3f0cc7297b08918d3d242ad3eca3021ce

Comments

checkasm/sw_rgb: test uyvytoyuv422 with odd widths
The SIMD uyvytoyuv422 implementations only handled even widths correctly.
Add odd width and 1x1 entries so the trailing column handling is exercised
against the C reference.

An odd width reads one source byte more than 2*width, the V sample at
src[2w], so the stride is extended by one for odd widths.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

tests/checkasm/sw_rgb.c

Change #273436

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 01 Jul 2026 03:48:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	48f30aa08c473fb034f0a43313ad566a1ea7a3ed

Comments

checkasm/sw_rgb: also test yuyvtoyuv422
yuyvtoyuv422 reads the trailing odd V sample at src[2w+1], one byte further
than uyvytoyuv422, so the number of extra source bytes an odd width needs is
passed in per format (1 for UYVY, 2 for YUYV) and added to the stride.

uyvytoyuv420 and yuyvtoyuv420 are intentionally not added: their x86 mmxext
chroma averaging uses PAVGB rounding and so is not bit-exact with the C
reference, which truncates, so they cannot be verified this way.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

tests/checkasm/sw_rgb.c

Change #273438

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 05:03:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fad1db043d45d4d3cc482934ac6cd579f556a6e8

Comments

swscale/format: Build ff_sws_chroma_pos() even without CONFIG_UNSTABLE
It is used in add_legacy_sws_pass() in graph.c.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/format.c

Change #273439

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 05:03:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0002cf3ba1ee55e70381b9fa0b445c9a356573cc

Comments

swscale/Makefile: Mark sws_ops* test tools as requiring CONFIG_UNSTABLE
Fixes "make testprogs" with --disable-unstable

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/Makefile

Change #273440

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 05:03:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fc7a984aae0b44c722d3977bb04c1d5037a3892c

Comments

swscale/graph: Remove ff_sws_graph_create()
Unused since 9fe0ff3d56618c24d95fa4e832148bab54928f18.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

doc/swscale-v2.txt
libswscale/graph.c
libswscale/graph.h

Change #273441

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 05:03:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	66d9b8e483738cd6b4c5ec3e4a5b4687512a5274

Comments

swscale/ops_chain: Disable ff_sws_setup_{clear,clamp,scale}
They are only used on aarch64.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/ops_chain.c

Change #273494

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1298ce9eda515b718a77e9a77bbe1f5d434b1408

Comments

aacenc_tns: reengineer coefficient derivation
The TNS tool needed a full rewrite.

1.) PNS cannot be used with M/S or I/S or TNS. The coding tools
interfere with one another.

2.) The coeffs were windowed, which was wrong.

3.) The applied filter did not match what the spec required.

Changed files

libavcodec/aacenc_tns.c
libavcodec/lpc.c
libavcodec/lpc.h

Change #273495

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d1943afbb787955db996ae7f2254f2d9f22d35f9

Comments

aaccoder: add NMR-based coder

Changed files

libavcodec/aaccoder.c
libavcodec/aaccoder_nmr.h
libavcodec/aaccoder_twoloop.h
libavcodec/aacenc.c
libavcodec/aacenc.h
libavcodec/aacencdsp.c
libavcodec/aacencdsp.h
libavcodec/aacpsy.c
libavcodec/psymodel.c
libavcodec/psymodel.h

Change #273496

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2fe4a916691a22990a7168d5587e6224b2051469

Comments

aacenc: report coding tool stats at uninit
Useful for debugging.

Changed files

libavcodec/aacenc.c
libavcodec/aacenc.h

Change #273497

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	06e949e45de3dd8f8680e7a290a0e947808f8bd2

Comments

aacpsy: pre-echo-aware short-block detection for isolated onsets
The LAME attack detector high-pass filters at fs/4 before measuring attack
intensity, so the gentler attacks of tonal material (piano, plucked strings)
fall below the threshold and stay in long blocks, smearing pre-echo across the
quiet gap before the note. Relax the attack threshold for an isolated onset,
preceded by PSY_LAME_PE_GAP long frames with a near-silent pre-onset sub-block
(< PSY_LAME_PE_QUIET of the frame peak), so it switches to short blocks.

Changed files

libavcodec/aacpsy.c

Change #273498

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0efac66e7e75aabd26ee63c16c6a198337334b0e

Comments

aacenc: use the NMR coder by default
The old coders will soon be removed.

Changed files

libavcodec/aaccoder.c
libavcodec/aacenc.c

Change #273499

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	38161819e7ab6fea011aa0c310f863ba2ba4f457

Comments

fate: update AAC encoder tests

Changed files

tests/fate/aac.mak
tests/ref/fate/generic-tags-remux-mov

Change #273500

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Wed 01 Jul 2026 14:12:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a974a54486c865b4482139ec55df8a04861ae9e3

Comments

avfilter: add less confusing error on frame queue overflow
Currently, such filter graphs just fail with a nebulous:

  [fc#0 @ 0x2a7b3c0] [error] Error while filtering: Cannot allocate memory

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/avfilter.c

Change #273502

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Wed 01 Jul 2026 14:24:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1dd71edca3127c38bdd70cbaf0cb63abd5b7d2a

Comments

configure: Add a dependency on user32 to the gfxcapture filter
This filter uses a bunch of user32 functions, like GetMessage,
DispatchMessage, PeekMessage, GetWindowRect, etc.

This makes it clear that if the gfxcapture filter is built, we
need to link user32, if other components that depend on user32
would happen to be disabled.

Separately, if targeting UWP (with -DWINAPI_FAMILY=WINAPI_FAMILY_APP)
then these user32 APIs are unavailable. We have an existing
configure check for user32, checking for the GetShellWindow
function. This function isn't used by gfxfilter, but its
availability on UWP matches the other functions that gfxfilter
uses.

This fixes builds for UWP, by excluding the gfxcapture filter.
(This is somewhat ironic, as the gfxcapture filter uses
"winrt" APIs - but the implementation of the filter also relies
on lots of APIs that are unavailable in UWP/"winrt".)

Changed files

configure

Change #273512

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 14:38:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d936a51b270c8a9125469527f59c211a9b116fe0

Comments

libavcodec/Makefile: add aaccoder_nmr.h to SKIPHEADERS

Changed files

libavcodec/Makefile

Change #273542

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 01 Jul 2026 18:14:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e2e889d9da7c065009198e3cc7b97cc1c47895f8

Comments

aacenc: reduce automatic cutoffs at high rates by 2Khz
We were spending too many bits coding useless junk.

Changed files

libavcodec/aacenc.c

Change #273545

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 01 Jul 2026 18:52:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d7d2d9d539e469c64f4a386d73dacc0c2371c501

Comments

tests/fate/generic-tags: use ac3_fixed encoder instead of float aac
There's no guarantee the aac encoder will be bitexact in its output across platforms.
Use ac3_fixed instead of aac_fixed while at it, so the aac encoder can get improvements
without affecting this test.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/generic-tags.mak
tests/ref/fate/generic-tags-remux-mov

Change #273563

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 01 Jul 2026 19:35:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a693a4cd802d33d9994761a8c7c8b69d9f913d17

Comments

tests/fate/generic-tags: use codec copy instead of encoding
Using ac3_fixed output was not enough as there's float to int conversion due to
the fact the mp3float decoder is used.
Instead of playing with codec combinations, just remove encoding from the test
altogheter. Mov supports muxing mp3 just fine.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/generic-tags.mak
tests/ref/fate/generic-tags-remux-mov

Change #273589

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	296d86c8bb7ce097cbfd8151bcf713de00b8a6a5

Comments

tests/checkasm/vc1dsp: Improve mspel test
Up until now, only the fullpel test (i.e. the test without pixel
interpolation) has been tested at all.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/vc1dsp.c

Change #273590

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b122b918eb8a7b0d1966da33e5f340f5281b8d89

Comments

avcodec/x86/vc1dsp_mc: Remove unnecessary movsxdifnidn
Forgotten in 5a49097b42cbc3eab888d15a91eeaf5520b5c381
merging Libav commit 2ec9fa5ec60dcd10e1cb10d8b4e4437e634ea428
(which changed the stride to ptrdiff_t).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_mc.asm

Change #273591

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	45c41d1d1b13db78ea3fee515bb5aafa9e3d7526

Comments

tests/checkasm/vc1dsp: Fix shadowing
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/vc1dsp.c

Change #273592

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f5b8a15b61bdf5cd18202a45010fcc2b67ee8d6e

Comments

avcodec/x86/vc1dsp_mc: Move inverse transform into a file of its own
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/Makefile
libavcodec/x86/vc1dsp_inv_trans.asm
libavcodec/x86/vc1dsp_mc.asm

Change #273593

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fd711adcb61e2b2e9c6a8bef2006f9e00e84ab35

Comments

avcodec/x86/vc1dsp_mc: Add size 8 horizontal SSSE3 mc functions
pmaddubsw strikes again.

vc1dsp.avg_vc1_mspel_pixels_tab_mc10_8_c:              150.2 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_8_mmxext:          44.5 ( 3.38x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_8_ssse3:           18.5 ( 8.12x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_8_c:              288.2 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_8_mmxext:          37.7 ( 7.64x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_8_ssse3:           18.1 (15.97x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_8_c:              155.4 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_8_mmxext:          46.5 ( 3.34x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_8_ssse3:           18.1 ( 8.60x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_8_c:              282.2 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_8_mmx:             42.7 ( 6.61x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_8_ssse3:           16.4 (17.16x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_8_c:              223.4 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_8_mmx:             36.3 ( 6.15x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_8_ssse3:           16.4 (13.59x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_8_c:              255.2 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_8_mmx:             43.6 ( 5.85x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_8_ssse3:           16.4 (15.52x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_mc.asm

Change #273594

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3354cbf1b76ff687cae4852fc15b6174c4e36b48

Comments

avcodec/x86/vc1dsp_mc: Add size 16 horizontal SSSE3 mc functions
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_c:             309.1 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_mmxext:        177.3 ( 1.74x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_ssse3:          52.3 ( 5.91x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_c:             279.6 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_mmxext:        148.8 ( 1.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_ssse3:          52.1 ( 5.37x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_c:             332.6 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_mmxext:        177.3 ( 1.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_ssse3:          52.5 ( 6.33x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_c:             288.8 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_mmx:           170.3 ( 1.70x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_ssse3:          51.3 ( 5.63x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_c:             236.2 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_mmx:           144.1 ( 1.64x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_ssse3:          51.3 ( 4.61x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_c:             286.6 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_mmx:           170.1 ( 1.69x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_ssse3:          51.2 ( 5.60x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_mc.asm

Change #273595

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4e05d94424a6c41e3b6e4d9a949faf13c9ce9fc1

Comments

avcodec/x86/vc1dsp_mmx: Remove purely horizontal MMX mc functions
Superseded by SSSE3. By the way, the SSSE3 functions occupy
816B, but the functions removed now occupied 2304B with GCC
and 6512B with Clang (which inlined everything).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_mmx.c

Change #273596

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b0d5914678ab9f7b610b01170fddb8bc83a383d8

Comments

avcodec/x86/vc1dsp_mc: Add size 8 vertical SSSE3 mc functions
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_8_c:              165.6 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_8_mmxext:          44.4 ( 3.73x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_8_ssse3:           18.5 ( 8.97x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_8_c:              152.5 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_8_mmxext:          37.3 ( 4.09x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_8_ssse3:           18.5 ( 8.25x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_8_c:              162.9 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_8_mmxext:          44.1 ( 3.69x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_8_ssse3:           18.3 ( 8.88x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_8_c:              150.5 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_8_mmx:             42.4 ( 3.55x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_8_ssse3:           16.5 ( 9.11x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_8_c:               78.4 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_8_mmx:             36.1 ( 2.17x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_8_ssse3:           16.5 ( 4.76x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_8_c:              144.7 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_8_mmx:             42.6 ( 3.40x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_8_ssse3:           16.3 ( 8.89x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_mc.asm

Change #273597

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a619524605fcf2d8afe3f88ecab433c107fdad47

Comments

avcodec/x86/vc1dsp_mc: Don't xor unnecessarily
m0 is not used at all in this function.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_mc.asm

Change #273598

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bafcf7b73f48c978e2abc897e14078faa059b6ca

Comments

avcodec/x86/vc1dsp_mc: Add size 16 vertical SSSE3 mc functions
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_16_c:             334.4 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_16_mmxext:        177.5 ( 1.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc01_16_ssse3:          52.3 ( 6.40x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_16_c:             306.9 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_16_mmxext:        149.2 ( 2.06x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc02_16_ssse3:          52.2 ( 5.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_16_c:             334.2 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_16_mmxext:        176.5 ( 1.89x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc03_16_ssse3:          51.9 ( 6.44x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_16_c:             311.9 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_16_mmx:           169.8 ( 1.84x)
vc1dsp.put_vc1_mspel_pixels_tab_mc01_16_ssse3:          51.6 ( 6.04x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_16_c:             279.3 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_16_mmx:           144.1 ( 1.94x)
vc1dsp.put_vc1_mspel_pixels_tab_mc02_16_ssse3:          51.7 ( 5.40x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_16_c:             310.6 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_16_mmx:           171.0 ( 1.82x)
vc1dsp.put_vc1_mspel_pixels_tab_mc03_16_ssse3:          51.6 ( 6.02x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_mc.asm

Change #273599

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 01 Jul 2026 20:00:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4c6e8fdbf6408c29cc871dcd08767962ec9ca184

Comments

avcodec/x86/vc1dsp_mmx: Remove purely vertical MMX mc functions
They have been superseded by SSSE3. Notice that the functions removed
occupied 3424B with GCC and 6176B with Clang here, whereas
the SSSE3 functions replacing them occupy only 944B.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_mmx.c

Change #273628

Category	ffmpeg
Changed by	Thomas Guilbert <tguilbertohnoyoudont@chromium.org>
Changed at	Thu 02 Jul 2026 04:49:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f956c41d7b2ecb5802a24dfe158d319b71c8a95a

Comments

Guard against loop underflow
This commits adds an early returns if `len <= pred_order` for RISC-V
architectures, matching the logic on other architectures.

Changed files

libavcodec/riscv/flacdsp_rvv.S

Change #273629

Category	ffmpeg
Changed by	Thomas Guilbert <tguilbertohnoyoudont@chromium.org>
Changed at	Thu 02 Jul 2026 04:49:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b3fb13e8dd6ef2451acaffbe84f178d859f50f68

Comments

Guard lpc methods in flacdsp and add tests
This commit adds additional guards for the RISC-V lpc variants.

It also adds tests to exercise the guarded paths in checkasm/flacdsp.c.

Changed files

libavcodec/riscv/flacdsp_rvv.S
tests/checkasm/flacdsp.c

Change #273630

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 04:57:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	053b2d18fa6714e927dbe8afcf5bc6fd9a9c6a05

Comments

avcodec/aacenc: export discard padding side data when needed
This is in line with other encoders, and is needed by the matroska muxer
to properly signal end trimming samples.

gaplessenc-pcm-to-mov-aac changes as the mov muxer now uses this information
when writing the edit list, so the shorter frame duration is not needed.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/aacenc.c
tests/ref/fate/gaplessenc-pcm-to-mov-aac

Change #273631

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Thu 02 Jul 2026 05:16:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0dac41d2e5e0f4251ed89eedb03a847d5d59ef67

Comments

avfilter/boxblur: reject pixel formats deeper than 16 bits
Fixes: heap buffer overflow
Fixes: CRufFb6TVikL
Fixes: dee7440531 (vf_boxblur: Templatize blur{8,16})
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_boxblur.c

Change #273643

Category	ffmpeg
Changed by	wangbin <wbsecg1ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 09:34:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d9caf5e873722a5bbbd2facacdb67fab6878c4bd

Comments

lavu: monotonic av_gettime_relative for msvc

Changed files

libavutil/time.c

Change #273648

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c5591a91ebb8ab107b136b3790cab771cba896c9

Comments

cbs_apv: relax minimum tile-sizes
APV currently specifies a very large minimum tile size. Just relax it
until the spec is amended.

Changed files

libavcodec/cbs_apv_syntax_template.c

Change #273649

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	876d10f68346f7c1d4efcdb73e06cb4545ea0528

Comments

lavc/apv_encode_vulkan: add a Vulkan APV encoder
This commit adds a compliant Vulkan compute APV encoder.

Changed files

configure
libavcodec/Makefile
libavcodec/allcodecs.c
libavcodec/apv_encode_vulkan.c
libavcodec/vulkan/Makefile
libavcodec/vulkan/apv_encode_dct.comp.glsl
libavcodec/vulkan/apv_encode_tiles.comp.glsl
libavcodec/vulkan/seg_gather.comp.glsl

Change #273650

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	adfa2e3255477213d292590d1489c53c01c443d3

Comments

vulkan_decode: drop ff_vk_decode_prepare_frame_sdr, use exec owned views
For SDR codecs, we have zero dependency on the images we decode. So using
the same scheme as the software code led us to needing unnecessarily to block
on a semaphore wait.

Instead of doing this, just make the imageviews a dependency of the exec context.
Saves on CPU, matters particularly for very fast decoding speeds.

FFv1 gets a slight workaround that's still better than what we did before.

Changed files

libavcodec/vulkan_apv.c
libavcodec/vulkan_decode.c
libavcodec/vulkan_decode.h
libavcodec/vulkan_dpx.c
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_prores.c
libavcodec/vulkan_prores_raw.c

Change #273651

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d040f5d9857af76d3cad3ae679c66e0377ef728f

Comments

ffv1enc_vulkan: gather slices on the GPU, drop the transfer copy
Uses the gather shader from the APV encoder to copy the packet data on
the GPU with a second pass.

Replace all of that with a seg_gather compute dispatch in the
same submission as the encode. It packs the per-slice slots tight and in
order into a contiguous host-visible buffer, so the final packet is just
the sum of the per-slice sizes and the gathered buffer is handed straight
to pkt->buf with no extra submit, no copy, no CPU involvement.

The slots are sized with the version-4 worst-case formula (the realistic
lossless bound, ~17x tighter than the version-3 budget this encoder never
approaches; the emitted bitstream version is unchanged) and rounded down
to a 16-byte multiple so the gather reads them with aligned wide loads.

Removes the transfer queue family, its exec pool, and the buf_regions
scratch array.

Changed files

libavcodec/ffv1enc_vulkan.c
libavcodec/vulkan/Makefile

Change #273652

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5784a775dfbabada9b589f289987a2beef7a9071

Comments

vulkan/prores_raw: reconstruct DC values in 32-bit
DCs can go over int16_t, and when it does, dc_add wraps, so the wrong sign
is reported, which causes errors when decoding more DCs.

Do the prediction in int and narrow to int16 only at store time. The
read_ac_vals() sign change is cosmetic (the value is only ever 0/-1, and
the old code already widened it).

Changed files

libavcodec/vulkan/prores_raw_decode.comp.glsl

Change #273653

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8d433a5b4d325807fb049668dddf15f27c2ccaf1

Comments

avcodec/vulkan_apv: store coefficients in a flat buffer, not the image
The entropy pass wrote each decoded coefficient into the output image as
16-bit scratch, then the iDCT pass read it back, dequantised it and
overwrote the same texel with the final pixel. Routing the coefficients
through the image couples the two passes to the image layout and forces
the entropy shader to address the descriptor-indexed image.

Add a dedicated device-local int16 buffer instead: the entropy shader
writes coefficients into it (one plane per component, MB-aligned coded
size) and the iDCT reads them, so the image is written exactly once, by
the iDCT. The buffer is zero-filled before the entropy dispatch so blocks
with no coded coefficients read as zero, and the decode->iDCT hand-off
becomes a buffer barrier. The output image is still cleared so any padding
the iDCT does not cover stays zero.

Changed files

libavcodec/vulkan/apv_decode.comp.glsl
libavcodec/vulkan/apv_idct.comp.glsl
libavcodec/vulkan_apv.c

Change #273654

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9872cb8c0e908db9dd2d834799e4e0de23f75bc0

Comments

avcodec/vulkan_apv: decode VLC codes branchlessly from a single window
apv_read_vlc() dispatched on the three APV code layouts with two data-
dependent branches and, for the long form, a second show/skip. The longest
legal code plus its optional sign bit fits in a single 32-bit window, so
compute all three candidate (value, length) pairs unconditionally and
select with no divergent branches: one show_bits(32) and one skip_bits per
code. apv_read_vlc_sign() additionally folds the trailing sign bit into the
same window, removing the separate get_bit() refill that DC and AC levels
did. Clamps on the long-form length guard corrupt streams without affecting
legal decodes.

Changed files

libavcodec/vulkan/apv_decode.comp.glsl

Change #273655

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:45:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ae25ca0b3677c3f9ff8c3b5fed78f987b7e10944

Comments

avcodec/vulkan_apv: replace the iDCT tile-position search with a division
The iDCT located its tile by linearly scanning the tile column/row start
tables. The APV tile grid is uniform-sized columns and rows with a single
remainder tile at each edge, so the tile index is just the luma position
divided by the (constant) first step, clamped to the last tile. Single
column/row grids have no step to read and resolve to index 0.

Changed files

libavcodec/vulkan/apv_idct.comp.glsl

Change #273656

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:47:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bb4eb89e1d1069ebddaa49302025f4f216d2b1e7

Comments

Changelog: add new AAC encoder improvements to the changelog

Changed files

Changelog

Change #273657

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 09:48:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f8389eae8782f815158812de49e83d579979a814

Comments

Changelog: add APV Vulkan encoder to the changelog

Changed files

Changelog

Change #273658

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 10:05:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	88c6409a0a8438a8f82df7a459a5179420423333

Comments

avcodec/version: bump minor for recent APV Vulkan encoder and AAC encoder
Helps identify what encoder users are using via a log.

Changed files

libavcodec/version.h

Change #273659

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 10:32:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5ffea4c6f148dfac48aa353b0e63dd284158f6fe

Comments

aacenc: at 192kbps per channel, code the full spectrum
Better to be safer than usual, since the signal contains literally no
audible data, and is so expensive to code, the coder itself is very
reluctant to do so except at very high rates.

Changed files

libavcodec/aacenc.c

Change #273665

Category	ffmpeg
Changed by	Shreesh Adiga <16567adigashreeshohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 11:03:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	915bac7bdc89317b85dc19e9cf6e7aed99be2e68

Comments

avutil/crc: add aarch64 hybrid crc32 NEON PMULL+EOR SIMD implementation
Adding crc32 specialization for aarch64 which uses both PMULL and crc32
instructions to perform 192 bytes fold in one iteration, performing
9x PMULL and 6 crc32 in one loop iteration, obtaining higher performance for
large inputs >8kB. This approach is based on zlib-ng implementation which
is also described at https://github.com/corsix/fast-crc32.

For smaller buffer size, it was observed to be slightly slower, thus only
for input size >8192 this logic is used, for smaller sizes otherwise the
4x PMULL folding method is used along with scalar crc32 instructions for
processing the remainder input size.

On a MediaTek Dimensity 9400 Android device in termux environment,
with normal checkasm seed 0 which picks random buffer size and max buffer size
of 16kB, the data observed on Cortex X925, A720 and X4:
X925 Before:
  crc_32_IEEE_LE_c:                  12762.0
  crc_32_IEEE_LE_crc:                  667.5 (19.11x)
  crc_32_IEEE_LE_pmull_eor3:           346.9 (26.30x)
X925 After:
  crc_32_IEEE_LE_c:                  12707.6
  crc_32_IEEE_LE_crc:                  665.2 (19.10x)
  crc_32_IEEE_LE_pmull_eor3:           292.8 (41.90x)

A720 Before:
  crc_32_IEEE_LE_c:                  23059.1
  crc_32_IEEE_LE_crc:                 1220.7 (18.89x)
  crc_32_IEEE_LE_pmull_eor3:          1198.9 (19.23x)
A720 After:
  crc_32_IEEE_LE_c:                  23293.3
  crc_32_IEEE_LE_crc:                 1209.1 (19.26x)
  crc_32_IEEE_LE_pmull_eor3:          1150.4 (20.24x)

X4 Before:
  crc_32_IEEE_LE_c:                  12405.5
  crc_32_IEEE_LE_crc:                  664.5 (18.67x)
  crc_32_IEEE_LE_pmull_eor3:           498.1 (24.90x)
X4 After:
  crc_32_IEEE_LE_c:                  12457.2
  crc_32_IEEE_LE_crc:                  665.5 (18.72x)
  crc_32_IEEE_LE_pmull_eor3:           468.8 (26.57x)

So it seems to work well on high performance core like X925, and results in about
20% better performance, while having tiny gains on other cores.

Testing for input size of 160 kB after modifying the checkasm crc test to
have buffer size increased to 160kB and always using full capacity instead of
a random size results in below observations:
X925 Before:
  crc_32_IEEE_LE_c:                 210177.1
  crc_32_IEEE_LE_crc:                10313.7 (20.35x)
  crc_32_IEEE_LE_pmull_eor3:          6580.9 (31.83x)
X925 After:
  crc_32_IEEE_LE_c:                 210869.3
  crc_32_IEEE_LE_crc:                10304.8 (20.36x)
  crc_32_IEEE_LE_pmull_eor3:          3098.5 (68.05x)

A720 Before:
  crc_32_IEEE_LE_c:                 387502.5
  crc_32_IEEE_LE_crc:                19196.7 (19.54x)
  crc_32_IEEE_LE_pmull_eor3:         18717.1 (20.63x)
A720 After:
  crc_32_IEEE_LE_c:                 392090.8
  crc_32_IEEE_LE_crc:                19795.1 (18.68x)
  crc_32_IEEE_LE_pmull_eor3:         14971.4 (24.97x)

X4 Before:
  crc_32_IEEE_LE_c:                 196232.0
  crc_32_IEEE_LE_crc:                10378.7 (18.68x)
  crc_32_IEEE_LE_pmull_eor3:          7742.0 (25.29x)
X4 After:
  crc_32_IEEE_LE_c:                 199632.9
  crc_32_IEEE_LE_crc:                10495.8 (18.32x)
  crc_32_IEEE_LE_pmull_eor3:          5448.9 (24.69x)

Seems to result in about 2x gains on X925, 25% on A70 and 40% on X4.
In general the performance gains depends on the CPU Core and input size,
and this optimization benefits large input size especially on high performance
cores like X925 and Apple M series.

Changed files

libavutil/aarch64/crc.S
libavutil/aarch64/crc.h

Change #273667

Category	ffmpeg
Changed by	Shreesh Adiga <16567adigashreeshohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 11:03:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e18e229388b7b5d5bb3bb667533a9e853ae94737

Comments

checkasm/crc: increase size of input buffer from 8k to 16k
For AArch64, AV_CRC_32_IEEE_LE implementation has codepath which is triggered
when input buffer size is >= 8192. Thus to ensure sufficient test coverage,
size of the input buffer in the test file is being increased from 8192 to 16384.

Changed files

tests/checkasm/crc.c

Change #273677

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 02 Jul 2026 11:29:19
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	95a888b9cadb4888d957797ba411b300422e81cc

Comments

aacenc_tns: tighten TNS toggle for short frames
Excessive TNS use for short frames resulted in audible popping at
high rates, this strictens the threshold conservatively.

Changed files

libavcodec/aacenc_tns.c

Change #273713

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 02 Jul 2026 15:29:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6f80e2765492700622596af720534cef33dd31b4

Comments

avformat/spdifenc: bound DTS core_size against the packet size in the HD path
Fixes: out of array read
Fixes: yBSax492UIB9
Fixes: 482d98f69b2 (spdifenc: IEC 61937 encapsulation of DTS-HD for HDMI)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/spdifenc.c

Change #273714

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Thu 02 Jul 2026 15:32:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	11d5f475be95d22d5f0692220cc772b116abc632

Comments

avformat/rtpdec_asf: reject ASF objects smaller than their header
Fixes: infinite loop
Fixes: MzWwJdpZF2Ls
Fixes: c2f3eec445389d67afc8c699ba23915a20cae51c (Implement RTSP-MS/ASF packet parsing.)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/rtpdec_asf.c

Change #273719

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 15:48:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	324617b26f4b4e074caff12af5d7493db16bec46

Comments

tests/fate/libavutil: add FATE test for dovi_meta
Test av_dovi_alloc, av_dovi_metadata_alloc, and av_dovi_find_level.
Verifies that the four inline offset-based accessors (get_header,
get_mapping, get_color, get_ext) return pointers consistent with
the offset fields, that find_level returns the first matching ext
block or NULL for a missing level, and OOM paths via av_max_alloc.

Coverage for libavutil/dovi_meta.c: 63.16% -> 100.00%

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/dovi_meta.c
tests/fate/libavutil.mak
tests/ref/fate/dovi_meta

Change #273720

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 15:48:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7ea44ca6a6fb630a0678e8afbdd6bb49f06baba5

Comments

tests/fate/libavutil: add FATE test for mastering_display_metadata
Test all 5 public functions: av_mastering_display_metadata_alloc,
av_mastering_display_metadata_alloc_size, the create_side_data
variant, av_content_light_metadata_alloc, and its create_side_data
variant. Verifies the {0,1} rational defaults set by get_defaults(),
write/read-back of HDR metadata fields, frame side data attachment
for both mastering display and content light metadata, and OOM
paths via av_max_alloc.

Coverage for libavutil/mastering_display_metadata.c: 86.49% -> 100.00%

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/mastering_display_metadata.c
tests/fate/libavutil.mak
tests/ref/fate/mastering_display_metadata

Change #273721

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 15:48:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	784ab1558941fa2d0e4774abb9c0ac4a0135f74c

Comments

tests/fate/libavutil: add FATE test for video_hint
Test av_video_hint_alloc with 0, 1, and 4 rects, and
av_video_hint_create_side_data. Verifies that av_video_hint_rects
and av_video_hint_get_rect return pointers consistent with
rect_offset and rect_size, write/read-back of rect coordinates,
both hint type values, and OOM paths via av_max_alloc.

Coverage for libavutil/video_hint.c: 0.00% -> 82.05%

The remaining uncovered lines are the nb_rects overflow guard and
the av_buffer_create / av_frame_new_side_data_from_buf failure
cleanup paths, which av_max_alloc() cannot reach since it forces
the first allocation to fail.

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/video_hint.c
tests/fate/libavutil.mak
tests/ref/fate/video_hint

Change #273722

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Thu 02 Jul 2026 16:29:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e85c0e0ab1b5333c9d3c3bc6fbc1574fdd2d9d3d

Comments

avformat/shared: increase CRC size to 32-bits
This increases the amount of corrupted data we can correctly detect
as corrupted from (on average) 2 GiB to 128 TiB, at the default 32 KiB block
size.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #273723

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Thu 02 Jul 2026 16:29:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aa8fc9f946c68565dd0a105437e49899154a29eb

Comments

avformat/shared: add -retry_corrupt option
Similar to -retry_errors, this retries blocks whose only crime is failing
the CRC self-check.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

doc/protocols.texi
libavformat/shared.c

Change #273724

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Thu 02 Jul 2026 16:29:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cce1b4f2289f9b02a3b9848dbc91c87903bf3821

Comments

avformat/shared: hash cache version into filename
Changing version implies a non-backwards compatible disk layout.
Alternatively, we could detect a version mismatch and delete + rebuild the
cache file, but that's nontrivial to get right because of existing processes
that might be concurrently trying to open the same cache file with the old
version.

Overall, cleaner and safer to just separate them by version.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #273725

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Thu 02 Jul 2026 16:38:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dbd495f066a85ba96b17433f4306582aa37c3951

Comments

avformat/vobsub: reuse subtitle streams and bound the stream count
Fixes: heap buffer overflow
Fixes: lqaO5R1BaZGO
Fixes: dbfe61100b (avformat/vobsub: fix several issues.)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/mpeg.c

Change #273726

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Thu 02 Jul 2026 16:38:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1836ef96846937a6cc2443698a693104f5c0b21e

Comments

avcodec/adx: sync decoder channel state on NEW_EXTRADATA
Fixes: out of array access
Fixes: heaNtmHvklpe
Fixes: 92396cee602320c714713ca2d93b53684ad57000 (avformat: add CRI AAX demuxer)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/adxdec.c

Change #273731

Category	ffmpeg
Changed by	Marvin Scholz <epirat07ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 18:23:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6aa9590e7a40ff584da41dfd64d95b098bf0c9c4

Comments

avformat/hevc: use correct format specifier
The numOfArrays variable is an integer, so use %d to print it.

Changed files

libavformat/hevc.c

Change #273732

Category	ffmpeg
Changed by	Marvin Scholz <epirat07ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 18:23:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	909e00f3fe130cca25eaea9bd5541a163d1809bd

Comments

avformat/jvdec: add fall-through annotations

Changed files

libavformat/jvdec.c

Change #273733

Category	ffmpeg
Changed by	Marvin Scholz <epirat07ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 18:23:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7b451cdb2155b79839fe82a37bc7c36cceb49978

Comments

avformat/nuv: add fall-through annotations

Changed files

libavformat/nuv.c

Change #273734

Category	ffmpeg
Changed by	Marvin Scholz <epirat07ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 18:23:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9448d5e0843073492087d656ae7e0896d336f0cf

Comments

avfilter/vf_mcdeint: add fall-through annotations

Changed files

libavfilter/vf_mcdeint.c

Change #273735

Category	ffmpeg
Changed by	Marvin Scholz <epirat07ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 18:24:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	884c05605933ec960681da22232daa6baba1182b

Comments

avcodec/lcevc_parser: replace fall-through comment with annotation

Changed files

libavcodec/lcevc_parser.c

Change #273742

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 02 Jul 2026 20:39:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a8968b14250c85de947fce4df9793b88c8580869

Comments

Makefile, ffbuild/{common,library}: Allow to build DEVTOOLS
These tools run on the target and are supposed to be used
for development; they are basically the same as TESTPROGS,
just without the presumption of living in the tests
subdirectory.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

Makefile
ffbuild/common.mak
ffbuild/library.mak

Change #273743

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 02 Jul 2026 20:39:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd9e6996afe679d713b726e1580a5f855cf7b255

Comments

swscale/uops: Move stuff for generating uops_macros.h out
It is not used for normal builds and is more an auxiliary
dev tool; move the code into a new file, uops_macros_gen.c
to be built as a DEVPROG.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/Makefile
libswscale/tests/sws_ops.c
libswscale/uops.c
libswscale/uops.h
libswscale/uops_list.h
libswscale/uops_macros_gen.c
tests/fate/libswscale.mak

Change #273744

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 02 Jul 2026 20:39:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	da8f9c4cacc2abac68855bf702ed4add2c43f5c3

Comments

swscale/ops: Move stuff for enumerating op lists out
It is not used for normal builds, so move the code into
a new file, uops_macros_gen_template.c to be included
by its only users, namely tests/sws_ops{,_aarch64}.c
and uops_macros_gen.c.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/op_list_gen_template.c
libswscale/ops.c
libswscale/ops.h
libswscale/tests/sws_ops.c
libswscale/tests/sws_ops_aarch64.c
libswscale/uops_macros_gen.c

Change #273746

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 22:20:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	14ed5e8ae31d0c833afbb5ed0e0ca2c7cfc0db87

Comments

forgejo/workflows: rebase PR commits onto target branch before running verification
This simulates the state of the repository after the PR is
rebased-and-merged into the target branch and validates the merged
result rather than the PR branch in isolation.

Without this step, CI may run against a stale branch. A PR that merges
cleanly can still fail verification after merging into target because
the target branch may have advanced and changed tests. Conversely, the
Docker image used for the build may no longer work with the old
repository state when testing an old PR branch. This also allows the PR
to be revalidated by simply rerunning CI, without requiring an explicit
rebase.

This change also fails CI early when rebase conflicts are detected. Since
PRs with conflicts cannot be merged, there is no value in running the
remaining verification steps.

This is bit strict as it likely will fail to rebase merge commits,
should they be present in the PR. However, we generally prefer linear
history. It's rebased mostly to avoid search for merge commit in target
branch. While for PR we can ask API how many commit there are and the
parent of that should be merge-base commit, for target branch it's
unknown how deep this common commit is. It would be trivial if we did
full non-shallow clone, but we do shallow and probably want to keep it
this way. This forces us to just replay commits from PR on the tip of
the current target branch. It's fine, this is what we want in the end of
the day.

Incidentally this also fixes recently merged
dd6ae3e02412021e016983cc52c242c35e2d7ab1, which tries to validate commit
messages, but by default clone depth is 1 so only single commit message
of head commit in PR was merged. After this change all commits will be
visible. However, I still liked the API script that we had before, not
sure why this was completely reworked.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

.forgejo/actions/rebase-pr/action.yml
.forgejo/workflows/lint.yml
.forgejo/workflows/test.yml

Change #273770

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 22:53:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b5bfb7716c4914efdccbc3b507a77d676616be51

Comments

avcodec/ac3enc: drain the buffered samples at the end of encoding
When encoding a stream with an amount of samples multiple of a block, the last
the last 256 samples would be lost as the encoders were not marked as
AV_CODEC_CAP_DELAY.

This can be easily reproduced with:

ffmpeg -f lavfi -i sine -ac 2 -af atrim=start_sample=0:end_sample=4608 -c:a eac3 -f framecrc -

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/ac3enc.c
libavcodec/ac3enc.h
libavcodec/ac3enc_fixed.c
libavcodec/ac3enc_float.c
libavcodec/ac3enc_template.c
libavcodec/eac3enc.c
tests/fate/ac3.mak
tests/ref/fate/ac3-fixed-encode-2
tests/ref/fate/shortest

Change #273771

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 22:53:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b192eb5b71448d87aeee71cd31f1364c40d7fb5c

Comments

avcodec/audio_frame_queue: take into account padding in input frames
If the declared duration is less than the amount of samples in the frame, then
only consider the former as valid.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/audio_frame_queue.c

Change #273772

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 02 Jul 2026 22:53:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d8c1dd4233e51c9946813f332b2c6c4a5667b154

Comments

tests/fate/ac3: add a test to check correct encoder sample output
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/ac3.mak
tests/ref/fate/ac3-fixed-encode-3

Change #273811

Category	ffmpeg
Changed by	Anxious Koisi <ddoaoingohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 03:51:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	63958ce2b5dd1b22f534fc968c86329cbbceef07

Comments

avcodec/apac: Check bitstream_size to avoid integer overflow
Found-by: k00shi_
Fixes: OOM
Fixes: poc.c
Fixes: XMkAe1I9uFro

Changed files

libavcodec/apac.c

Change #273812

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 03 Jul 2026 03:55:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aafb5c655edc76a753275c383ebb139feb032718

Comments

avcodec/mace: reject sample counts that overflow int
Fixes: heap buffer overflow
Fixes: FmXBI2dbgvgD
Fixes: 0eea212943544d40f99b05571aa7159d78667154 (Add avcodec_decode_audio4().)
Found-by: Adrian Junge (vurlo)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/mace.c

Change #273846

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 14:24:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2e9dfa02147112f60d5415128a79b1310e4ad9f9

Comments

.forgejo/actions/rebase-pr: workaround stale `forge.event.pull_request.merge_base` value
Unfortunately, Forgejo does not update the merge-base when PR branch is
updated, the `forge.event.pull_request.merge_base` is stale value, from
the time when PR was created. We cannot relly on it, so try to infer
parent commmit of the PR from the shallow clone that we have.

This can be reproduced by:
1. `git rebase X` where X is some commit on target branch, later known
   as merge-base
2. Create PR. Notice that `forge.event.pull_request.merge_base` == `X`
3. `git rebase X^`
4. Force push the reparented branch. Notice that
   forge.event.pull_request.merge_base == X. This is invalid! X not only
   is no longer a merge-base, but it is completelly unrelated commit,
   which exists only on `target` branch, not on PR branch.
5. ????

Additionally I can say that in fact Forgejo does update the merge-base
value, but it looks like it does that AFTER the actions are run, or
maybe concurrently and there is a race there. I've observed that on 2nd
push of a PR branch, rebased on the same parent/merge-base, the reported
value is actually correct, but we cannot rely on that, because on 1st
run it's incorrect if PR branch was rebased on different parent.

For more complex PR fallback to full unshallow and merge.

While at it make git fetches verbose, so we know which commits are asked
for.

Changed files

.forgejo/actions/rebase-pr/action.yml

Change #273848

Category	ffmpeg
Changed by	Hunter Kvalevog <hunterohnoyoudont@kvog.sh>
Changed at	Fri 03 Jul 2026 14:49:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6d72600a301c441e4a6d46663fbbfdad7e021068

Comments

fftools/ffplay: remove SDL_AUDIO_ALSA_SET_BUFFER_SIZE
SDL_AUDIO_ALSA_SET_BUFFER_SIZE was removed in SDL 2.0.9:
https://github.com/libsdl-org/SDL/commit/c591429542f429aac4979fb6ae9bbe024491a1f

Signed-off-by: Hunter Kvalevog <hunter@kvog.sh>

Changed files

fftools/ffplay.c

Change #273851

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 14:53:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9011e7083e63c08555fd273fefb57532d1643b8c

Comments

tests/fate/ffmpeg: remove audio stream from fate-autorotate
It's not relevant to the test, and it's often affected by unrelated changes to the
ac3 encoder.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/ffmpeg.mak
tests/ref/fate/autorotate

Change #273858

Category	ffmpeg
Changed by	tknaveen <unbornboron69ohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 15:29:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	766074133f9711e440144a940ac0ad32a8b8e5d7

Comments

avfilter/dnn_backend_torch: fix missing ret in async error path
ff_dnn_start_inference_async() return value was not stored in ret,
causing execute_model_th() to return success on async startup failure.
This left vf_dnn_processing stuck in its flush loop waiting on a
task that was never started.

Changed files

libavfilter/dnn/dnn_backend_torch.cpp

Change #273871

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 03 Jul 2026 16:09:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	061b28fad6e804b75c87ad2b3136827270fe31e6

Comments

avcodec/x86/me_cmp: Improve median_sad functions
The median_abs functions involve computing a predictor
for a pixel difference; deviations from this predicted
pixel difference are summed to the result. For the leftmost
element of a row, the predictor is simply the leftmost
element of the last row (if any; otherwise zero).
For the other elements, it is the median of top, left
and (top + left - topleft).

The current approach to deal with this is to treat
the first element specially (by using a special accumulator
just for it and mask at the end). But there is a better way:
If the left and topleft predictors are zero, the median prediction
is top and therefore yields the correct result for the leftmost
element, obviating the need for the special case.

Creating the registers with zeroed left predictors is easy:
Just shift in the other direction to how it is done now.

Old benchmarks:
  median_sad_0_c:            429.9
  median_sad_0_ssse3:         43.7 ( 9.83x)
  median_sad_1_c:            189.9
  median_sad_1_ssse3:         24.5 ( 7.75x)

New benchmarks:
  median_sad_0_c:            431.1
  median_sad_0_ssse3:         39.7 (10.84x)
  median_sad_1_c:            190.6
  median_sad_1_ssse3:         20.4 ( 9.36x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/me_cmp.asm

Change #273872

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 03 Jul 2026 16:09:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	67dfae9c5fad698ea4a7fa4f81b1fe4be1f1f4fd

Comments

avcodec/x86/me_cmp: Reduce amount of registers used
One can avoid one scratch register in MEDIAN_ABS_ACC
(by clobbering topleft which isn't needed lateron).
Furthermore, the earlier code requested more registers
in cglobal than it actually used: median_sad8 only needs
seven registers, median_sad16 only 11. One of these
registers is actually unclobbered (read-only), so
median_sad8 does not need to save and restore and
non-volatile registers on Win64.

Furthermore, the restriction of median_sad8 to x64 can now
be lifted.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/me_cmp.asm
libavcodec/x86/me_cmp_init.c

Change #273873

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 03 Jul 2026 16:09:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f3343f51b159c420724c04315e457704f112449e

Comments

tests/checkasm/motion: Improve reporting
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/motion.c

Change #273874

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 03 Jul 2026 16:09:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fad2e0bc504d0619453151cdb2a88a744dee34bc

Comments

tests/checkasm/motion: Improve randomizing buffers
Randomize four bytes at a time and only do it if needed.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/motion.c

Change #273922

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 20:43:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a37171bed554305b7b83c521ccd25ef40806347f

Comments

tests/fate/hevc: add scale filter dependency to fate-hevc-bsf-mp4toannexb-new-extradata
Should fix failures when said filter is not compiled in.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/hevc.mak

Change #273936

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Fri 03 Jul 2026 21:47:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2f779272e061fbdf2725009f9e327c2f7a6b0a61

Comments

libavformat/img2enc: add update_filemtime option
Add a new boolean option -update_filemtime to the image2 muxer that
sets each output file's modification time based on the creation_time
metadata plus the frame's PTS offset.

This is useful when extracting frames from dashcam or action camera
footage where wall-clock timestamps should be preserved on the output
files, allowing photo management tools to sort frames by capture time
without post-processing.

The option requires creation_time metadata to be set (via -metadata
creation_time=...). If not present, a warning is logged and the
option is silently disabled. When PTS is unavailable, the creation
time is used as-is without frame offset.

Uses utimes() on POSIX and _utime() on Windows to set file timestamps
with microsecond and second precision respectively.

Includes a FATE roundtrip test that writes frames with a known
creation_time, reads them back using the demuxer's -ts_from_file
option, and verifies the PTS values match the expected timestamps.

Closes: https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/22537
Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

doc/muxers.texi
libavformat/img2enc.c
tests/fate-run.sh
tests/fate/image.mak
tests/ref/fate/img2-update-filemtime

Change #273947

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Sat 04 Jul 2026 01:53:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ddf8f40301af20ad985cf369d1eb6d114be0c8f0

Comments

avdevice/avfoundation: wait for frame consumption to avoid dropping A/V frames
The capture callback dropped the previous frame instead of waiting for it
to be consumed, so frames were lost when avf_read_packet() fell behind;
39fbd06314 (return EAGAIN instead of waiting) made it easy to hit.

Add back a condition variable: the capture callback blocks until
avf_read_packet() takes the current frame, and the reader waits on it when
no frame is ready. An is_stopping flag set in destroy_context() wakes both
sides so teardown cannot deadlock. observed_quit is now set under the lock
and broadcast as well, so a blocked reader wakes and returns EOF when a
transport-control device stops delivering frames.

Based on a patch by Zhongxin Zhuang <zx.zhuang@hotmail.com>; here
unlock_frames() only broadcasts and unlocks, so a read no longer releases
the other stream's still-unconsumed frame.

Verified by capturing camera+mic for 8s and comparing delivered packet
counts before/after this change:

  ffmpeg -f avfoundation -pixel_format nv12 -framerate 30 -i "0:0" -t 8 \
         -vf "scale=2560:1440,hqdn3d" -c:v libx264 -preset medium \
         -c:a aac out.mp4
  ffprobe -count_packets -show_entries stream=nb_read_packets out.mp4

  slow consumer (ideal: video ~240, audio ~375)
                     video   audio
    before (EAGAIN)   174     163   (audio ~57% dropped)
    after  (condvar)  234     376   (no drops)

  fast consumer (-preset ultrafast, no filter)
    before            240     328
    after             238     376

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavdevice/avfoundation.m

Change #273950

Category	ffmpeg
Changed by	Jack Lau <jacklau1222gmohnoyoudont@gmail.com>
Changed at	Sat 04 Jul 2026 04:39:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6f2f3755a06bdab40028dd8bfcfe91e057aac845

Comments

avformat/whip: disable DTLS certificate verification
Fix DTLS failure since 9549c9ad79ee399ff469a420223d4b6118498f7e

WebRTC does not verify CA because it allows self-signed cerificate.

Signed-off-by: Jack Lau <jacklau1222gm@gmail.com>

Changed files

libavformat/whip.c

Change #273984

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Sat 04 Jul 2026 18:52:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	22b08e8cc1cc0b9504f75cf46f6b6698247c3616

Comments

swscale/x86: fix SIGILL in u8 SCALE on SSE4-only CPUs
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libswscale/x86/ops_int.asm

Change #274029

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Sun 05 Jul 2026 01:53:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0fe4bd4b435b1de950890d37b1e4df7bae3655b1

Comments

avformat/imfdec: check for a missing or empty Path in the asset map
Fixes: NULL pointer dereference
Fixes: vIIX7oqa3coO
Fixes: 73f6cce9361 (avformat/imf: Demuxer)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/imfdec.c

Change #274030

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Sun 05 Jul 2026 01:53:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a76190152d0d1e90f38a761e0a1443c1dcb4e7a8

Comments

avformat/imfdec: reject virtual tracks with no resources
Fixes: NULL pointer dereference
Fixes: kia6rom3mZdr
Fixes: 73f6cce9361 (avformat/imf: Demuxer)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/imfdec.c

Change #274031

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 01:59:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4c6217477fc64305055b37d9d1d0d76d30e37f97

Comments

avcodec/nvdec: don't double free the fdd-owned context on the sep_ref error path
Fixes: double free
Fixes: rpSz7v3yq2u8
Fixes: 72982f8cb5dad6252a14226d28128313eed4a5ff (avcodec/nvdec: add support for separate reference frame)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/nvdec.c

Change #274032

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 02:00:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3b28a08a32c3599410c7f7cd3c1032c58f34dbe0

Comments

avformat/wavdec: fix integer overflow in W64 chunk size check
Fixes: 523658585/clusterfuzz-testcase-minimized-ffmpeg_dem_W64_fuzzer-5454488994643968
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/wavdec.c

Change #274033

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 02:05:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f50e22c68d1e579a3e2818a7d93d6eed092ad387

Comments

swscale/output: avoid signed overflow in yuv2rgb_write_full luma scaling
Fixes: 523522305/clusterfuzz-testcase-minimized-ffmpeg_SWS_fuzzer-4574409678716928
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libswscale/output.c

Change #274034

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 02:07:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4da9812e25894fb51d62a8875cfa8eb39b5e20f5

Comments

avfilter/vf_quirc: resize the quirc buffers when the input size changes
Fixes: out of array access
Fixes: JbvzNObhorBp
Fixes: 030e140145 (lavfi: add quirc filter)
Found-by: Adrian Junge (vurlo)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_quirc.c

Change #274035

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 02:08:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e7cbfd1c507b57a806a5825b87d609963e862c8c

Comments

avcodec/lcldec: zero the not-decoded tail to avoid heap disclosure
Fixes: use of uninitialized memory
Fixes: CsNDKB1K1U0C
Fixes: e2c3aa8e2b (avcodec/lcldec: More space for rgb24)
Found-by: Adrian Junge (vurlo)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/lcldec.c

Change #274036

Category	ffmpeg
Changed by	Vladimir Panteleev <gitohnoyoudont@thecybershadow.net>
Changed at	Sun 05 Jul 2026 02:41:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	734cbb6617036e171ed61fb990a768891bfd6030

Comments

vf_photosensitivity: scale badness threshold with history size
This commit attempts to address feedback from this filter's users, by
improving the filter's behavior at the start of playback or
immediately after seeking.

In these situations, the history buffer is empty, but because we did
not previously track its size, we were calculating the weighted
average as if the corresponding frames had zero badness. This caused
the filter to behave differently and possibly produce false negatives
when the history was not fully populated.

Address this by instead taking into account the history size when
calculating cumulative badness. To accomplish this:

- Add a history_size field to to PhotosensitivityContext, tracking how
  much of the history buffer is populated.

- Change the semantics of PhotosensitivityContext::badness_threshold.
  Previously, it was premultiplied by the maximum history size. This
  is no longer done, so that we can multiply it to the live
  history_size on the fly instead.

- Calculate the badness threshold on a per-frame basis. The result is
  now stored in a badness_threshold local variable.

This commit only changes the filter's behavior for the first
PhotosensitivityContext::nb_frames (configurable as the "frames"
filter option) frames. For successively filtered frames, the behavior
is unchanged.

Changed files

libavfilter/vf_photosensitivity.c

Change #274037

Category	ffmpeg
Changed by	Vladimir Panteleev <gitohnoyoudont@cy.md>
Changed at	Sun 05 Jul 2026 02:41:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	722493d3e13b81a45c6ca021a8c684d2d73d042a

Comments

vf_photosensitivity: add blend option
This commit attempts to address feedback from this filter's users, by
introducing a new option which controls the amelioration mechanism.

The "blend" option is a factor which is multiplied by the difference
in badness (between the threshold and the currently accumulated
badness). This difference normally controls how much of the next frame
we can let through without making it exceed the badness threshold.

Setting the option to zero effectively puts the filter into a mode
where it always duplicates the last frame (which did not put
accumulated badness over the threshold) instead of attempting to blend
in new frames.  I have received reports that this mode is preferable
to users for some types of media.

Changed files

doc/filters.texi
libavfilter/vf_photosensitivity.c

Change #274038

Category	ffmpeg
Changed by	Vladimir Panteleev <gitohnoyoudont@cy.md>
Changed at	Sun 05 Jul 2026 02:41:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dd057bd8569a1fb74d5edc9ff56eee83bddb198f

Comments

fate/filter-video: add photosensitivity blend test
Add a small self-contained lavfi test for the photosensitivity filter's
blend path. The graph produces one black RGB frame followed by two white
RGB frames, then runs:

    photosensitivity=frames=2:threshold=95:blend=0.5

The threshold is chosen so that the first black-to-white transition
barely exceeds the detector limit. A full 8x8 RGB grid transition has
badness 64 * 3 * 255 = 48960, while threshold=95 gives 48640 for the
first checked history window. That forces the filter into the blend
branch without needing any external sample.

With blend=0.5, the runtime factor is 48640 / 48960 * 0.5, or about
0.4967. The fixed-point blender truncates this to an input-frame weight
of 127/256, so blending black toward white produces an RGB component
value of 126. The reference therefore records black, the blended gray
frame, and then the following accepted white frame.

This makes the otherwise cryptic CRCs tie directly to the blend
calculation and history update behavior.

Co-authored-by: Codex CLI <noreply@openai.com>

Changed files

tests/fate/filter-video.mak
tests/ref/fate/filter-photosensitivity-blend

Change #274039

Category	ffmpeg
Changed by	Vladimir Panteleev <gitohnoyoudont@cy.md>
Changed at	Sun 05 Jul 2026 02:41:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e058e8c0c533add4878bf73ff1f148c5ea879360

Comments

vf_photosensitivity: fix metadata when duplicating frames
When the blending factor was calculated to be 0, the hot path skipped
updating variables later emitted in metadata.

Fix the flow to ensure they are populated consistently.

Extend the FATE test suite to check metadata output.

Changed files

libavfilter/vf_photosensitivity.c
tests/fate/filter-video.mak
tests/ref/fate/filter-metadata-photosensitivity-blend0

Change #274040

Category	ffmpeg
Changed by	Omkhar Arasaratnam <omkharohnoyoudont@linkedin.com>
Changed at	Sun 05 Jul 2026 04:20:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dcc18b60779b0d228b37621b6485cb74b2d6f352

Comments

avformat/mov: bound the iref thmb/cdsc entry count by the sub-box size
Fixes: excessive parsing past the iref sub-box
Fixes: TMwMAJfgy3l9
Fixes: 74e430202d933284fc38b591d4b3a12464e8aec6 (avformat/mov: make items referencing items generic)
Found-by: Omkhar Arasaratnam
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/mov.c

Change #274041

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Sun 05 Jul 2026 04:34:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c364ab176f722bdabc886845e770c9eacbc1e3e5

Comments

avformat/hls: reject Master Playlist tags in a Media Playlist
Fixes infinite loop on crafted HLS playlists.

Fixes: recursion.m3u8
Fixes: cEy2cxyyPaLH
Fixes: cd223e0b4e (Add Apple HTTP Live Streaming demuxer)
Found-by: BapToutatis
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c

Change #274042

Category	ffmpeg
Changed by	Zhao Zhili <quinkblackohnoyoudont@foxmail.com>
Changed at	Sun 05 Jul 2026 06:15:26
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	056192a835330ec476ca5254302b682c7c8d2623

Comments

avformat/mods: fix keyframe flagging in read_packet
avformat_index_get_entry_from_timestamp() was called with the file
position passed as wanted_timestamp, so the lookup was meaningless.
Use ffstream(st)->cur_dts as the wanted timestamp.

Changed files

libavformat/mods.c

Change #274043

Category	ffmpeg
Changed by	Zhao Zhili <quinkblackohnoyoudont@foxmail.com>
Changed at	Sun 05 Jul 2026 06:15:26
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8b1118818bc9fea829ebd07aa31dd0461b494084

Comments

avformat/mods: bail out on av_get_packet() failure

Changed files

libavformat/mods.c

Change #274044

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 05 Jul 2026 06:22:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	14604ec4b797248222cdd72a74d39abbdad9b343

Comments

avformat/iff: reject invalid data_size in ANIM
Fixes: 520277418/clusterfuzz-testcase-minimized-ffmpeg_dem_IFF_fuzzer-6733106211258368
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/iff.c

Change #274054

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 05 Jul 2026 09:02:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	97cbffe9172f1614af56363022f560700fa0c36b

Comments

all: Don't discard const through {mem,str,strr}chr
C23 made these generic functions that no longer cast const
away for you, leading to warnings when compiling with C23
and a recent enough toolchain (glibc supports this since 2.43).
This commit fixes all the warnings that can simply be fixed
by adding const, without adding casts.
(the latter excludes parse_forced_key_frames() in ffmpeg_mux_init.c).

Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

fftools/textformat/tf_json.c
libavcodec/htmlsubtitles.c
libavcodec/microdvddec.c
libavcodec/pnm_parser.c
libavfilter/graphparser.c
libavformat/avformat.c
libavformat/format.c
libavformat/http.c
libavutil/avstring.c
libavutil/channel_layout.c
libavutil/mem.c

Change #274081

Category	ffmpeg
Changed by	Pavel Kohout <disclosureohnoyoudont@aisle.com>
Changed at	Sun 05 Jul 2026 19:07:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	92737390dc133daadce47dd7d2ec8ef3d9ebcbed

Comments

avcodec/vulkan_hevc: reject too many VPS HRD parameter sets
Fixes: stack buffer overflow
Fixes: tD7Mj0ST7ND3
Fixes: 82864c21112157951ce91b4430a9018edd02f5ab (vulkan_hevc: use VK_KHR_video_maintenance2 if available)
Found-by: Pavel Kohout (Aisle Research)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/vulkan_hevc.c

Change #274089

Category	ffmpeg
Changed by	Tristan Matthews <tmatthohnoyoudont@videolan.org>
Changed at	Sun 05 Jul 2026 19:47:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	79adb692a2720647e9dbe57de33fd50b41db295c

Comments

rvv: add optimized h264 intra prediction functions
Reviewed-by: Nathan E. Egge <unlord@xiph.org>
Reviewed-by: Rémi Denis-Courmont <remi@remlab.net>

Benched on SpaceMIT K1 for VLEN 256 and CanMV-K230 for VLEN 128.

                         C       RVV256           C       RVV128
pred8x8_horizontal_8     31.1    19.8  (1.57x)    39.9    19.3  (2.07x)
pred8x8_plane_8         175.6    78.0  (2.25x)   175.7    89.6  (1.96x)
pred16x16_dc_8           56.1    44.2  (1.27x)    73.6    53.2  (1.38x)
pred16x16_dc_128_8       21.1    18.5  (1.14x)    37.2    28.3  (1.31x)
pred16x16_horizontal_8   84.9    37.3  (2.28x)   106.8    47.4  (2.25x)
pred16x16_left_dc_8      43.6    37.9  (1.15x)    60.8    47.4  (1.28x)
pred16x16_plane_8       644.8   130.0  (4.96x)   610.8   182.3  (3.35x)
pred16x16_top_dc_8       36.7    29.2  (1.26x)    53.8    38.5  (1.40x)
pred16x16_vertical_8     27.9    19.2  (1.45x)    42.3    28.9  (1.46x)

Changed files

libavcodec/h264pred.c
libavcodec/h264pred.h
libavcodec/riscv/Makefile
libavcodec/riscv/h264_intrapred_init.c
libavcodec/riscv/h264_intrapred_rvv.S

Change #274122

Category	ffmpeg
Changed by	Jamaika1 <lukaszcz18ohnoyoudont@wp.pl>
Changed at	Mon 06 Jul 2026 00:31:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c74f27955949dea8b2b33306cf0ea864ed7bf365

Comments

avfilter/vf_quirc: Changed character specifier
```
vf_quirc.c: In function 'filter_frame':
vf_quirc.c:104:46: warning: format '%ld' expects argument of type 'long int', but argument 5 has type 'int64_t' {aka 'long long int'} [-Wformat=]
  104 |            "Found count %d codes in image #%ld\n", codes_count, inl->frame_count_out);
      |                                            ~~^                  ~~~~~~~~~~~~~~~~~~~~
      |                                              |                     |
      |                                              long int              int64_t {aka long long int}
      |                                            %lld
```

Changed files

libavfilter/vf_quirc.c

Change #274123

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 06 Jul 2026 02:16:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	976490dcc3dbe0684851c30eff88a85d4d386d1a

Comments

avformat/hls_sample_encryption: Validate ADTS frame length against packet
Fixes: out of array access
Fixes: playlist.m3u8 / make_poc.py
Fixes: rJ50u41V7ctk
Fixes: ff958b3846 (libavformat/hls: add support for decryption of HLS media segments encrypted using SAMPLE-AES encryption method)
Found-by: Clouditera Security Research Team <haoyuliu@clouditera.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/hls_sample_encryption.c

Change #274126

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Mon 06 Jul 2026 06:46:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f2cd9282a952853b91b6c4a5615e05c856efb256

Comments

Changelog: move iTerm2 muxer entry from 9.0 to <next>
The iTerm2 inline image protocol muxer did not make the 9.0 release
cut.

Changed files

Changelog

Change #274127

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 06 Jul 2026 08:39:19
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c6498178bbfcdb6cc67ce01242f3faeb9d039119

Comments

avfilter/vsrc_gfxcapture_winrt: add missing system_error include
Fixes: error: no type named 'system_error' in namespace 'std'
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavfilter/vsrc_gfxcapture_winrt.cpp

Change #274211

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Mon 06 Jul 2026 18:26:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	32a256f619a344a7a351b89709f4a7de732268eb

Comments

avcodec: add a graph based bitstream filtering API
Tagged experimental for now, as it may see modifications depending on
requirements.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

configure
doc/APIchanges
libavcodec/Makefile
libavcodec/bitstream_filters.c
libavcodec/bitstreamfilter.c
libavcodec/bsf.c
libavcodec/bsf.h
libavcodec/bsf/Makefile
libavcodec/bsf/filters.h
libavcodec/bsf/sink.c
libavcodec/bsf/source.c
libavcodec/bsf_internal.h
libavcodec/bsfgraph.c
libavcodec/version.h

Change #274212

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Mon 06 Jul 2026 18:27:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f2b3888754513ff728ecb8a4bf15fc2cb18d6b6a

Comments

avcodec/bsf: add an LCEVC stream merging bsf
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

Changelog
libavcodec/bitstream_filters.c
libavcodec/bsf/Makefile
libavcodec/bsf/lcevc_merge.c
libavcodec/bsf/packetsync.c
libavcodec/bsf/packetsync.h
libavcodec/version.h

Change #274213

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Mon 06 Jul 2026 18:27:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4eb56054c203a4e5139ef485ec51ffa214cb469f

Comments

fftools/ffmpeg_demux: merge LCEVC enhancement payloads into the base stream
This is a temporary implementation until support for bitstream filter graphs is
generically added to the scheduler.

The dual track tests are as such changed so their output is not affected by builds
with liblcevc-dec enabled.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_demux.c
tests/fate/mpegts.mak
tests/ref/fate/mpegts-lcevc-h264-dual-track
tests/ref/fate/mpegts-lcevc-hevc-dual-track
tests/ref/fate/mpegts-lcevc-vvc-dual-track

Change #274233

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Tue 07 Jul 2026 02:09:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2dfe02aff861836445735735aca1d5e8ec869721

Comments

fftools/ffmpeg_demux: free stream group reference array in DemuxStream
Missed in 4eb56054c203a4e5139ef485ec51ffa214cb469f.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_demux.c

Change #274234

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Tue 07 Jul 2026 04:39:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	160737cf0da1915d8499881a8021d170f927411d

Comments

fftools/opt_common: fix format string bug in print_program_info
FFMPEG_CONFIGURATION and FFMPEG_VERSION were concatenated into av_log
format strings, so any % in them was interpreted as a conversion
specifier and consumed unrelated varargs, causing undefined behavior
ranging from garbled output to a segfault on `ffmpeg -version`. Pass
both as %s arguments instead.

Fixes issue #23662.

The FFMPEG_CONFIGURATION fix is based on a patch by FinnRG; the
FFMPEG_VERSION case is fixed here as well.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

fftools/opt_common.c

Change #274264

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 15:06:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	df2649ffb128c6846c523a5cb755e8b789009e4e

Comments

avcodec/vvc/dec: merge the SAO and ALF pixel buffer tab loops
Both loops iterate over the same components and compute the same
per-component width and height. The order of tabs within a TabList
does not matter, so build all pixel buffer tabs in a single loop.

Changed files

libavcodec/vvc/dec.c

Change #274265

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 15:06:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a41f5431137029215bc444af9cf5f16ce33e0400

Comments

avcodec/vvc/dec: fix pixel buffer tab leak on chroma format change
Fixes: memleak
Fixes: 518575619/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VVC_fuzzer-5161396783611904
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libavcodec/vvc/dec.c

Change #274277

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 19:30:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c5150ad48b2c6e14b01a4f73b5922f00ca200869

Comments

avcodec/hevc/ps: avoid signed overflow reading act qp offsets
Fixes: 523734452/clusterfuzz-testcase-minimized-ffmpeg_dem_HEVC_fuzzer-5246801451220992
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/hevc/ps.c

Change #274278

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 19:31:26
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	90a285ca782ebe633ba40ab123b6ff0cde9085a8

Comments

avcodec/jpeg2000dec: avoid signed overflow in ROI shift
Fixes: 522402419/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_JPEG2000_DEC_fuzzer-5987463315914752
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/jpeg2000dec.c

Change #274279

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 19:35:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fe47696aa1eb3b4f2359b9c4416c529912deee38

Comments

avcodec/lzf: pad the decompressed buffer
Fixes: out of array read
Fixes: oddlist.mov / poc.mov
Found-by: Clouditera Security; Z.ai Security; NSFOCUS
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/lzf.c
libavcodec/lzf.h

Change #274280

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 19:43:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b3712addc9636f428a751586c903b4b1c7be7fb3

Comments

avfilter/vf_v360: reject dimensions too small for the projection
Fixes: out of array read
Fixes: assertion failure
Fixes: mQzloVqnivHQ
Found-by: Anthony Hurtado
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_v360.c

Change #274281

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 07 Jul 2026 19:43:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a73d648f6e3e5224dfbee73005a59c2dc161445f

Comments

avfilter/vf_v360: keep remap source coordinates in bounds
Two runtime paths could compute out-of-range source coordinates for
degenerate projection geometry, causing heap-buffer-overflow reads

Fixes: out of array read
Fixes: assertion failure
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_v360.c

Change #274284

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Tue 07 Jul 2026 20:31:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	813c28d1ce216502ee21d4912980689ad60333d3

Comments

avformat/dashdec: check NULL pointer of av_strtok value before use it
fix issue: issues/23057 POC1
release seg memory and return NULL if av_strtok return NULL.

Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/dashdec.c

Change #274285

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Tue 07 Jul 2026 20:31:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5e0ca682a43f3c3d2b443c0d86d6a3d4d6ad4dd7

Comments

avformat/dashdec: check NULL pointer before use str_end_offset
fix issue: issues/23057 POC2

Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/dashdec.c

Change #274287

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Tue 07 Jul 2026 20:31:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0c0dd61b8f0cee6ed53f735d06b73998e13b6fea

Comments

avformat/dashdec: fix integer truncation in calc_max_seg_no()
When computing the segment count for a SegmentTimeline entry with
repeat="-1" (indefinite repeat, typical for live streams), the
expression:

  length_of_each_segment = pls->timelines[i]->duration /
                           pls->fragment_timescale;
  num = c->period_duration / length_of_each_segment;

performs two successive integer divisions.  The first division
truncates toward zero; if the segment duration in timescale ticks
is smaller than the timescale (duration < fragment_timescale),
length_of_each_segment becomes 0, and the second division triggers
a division by zero.

Fix this by using av_rescale(), which computes

  period_duration * fragment_timescale / duration

with 64-bit intermediate precision, avoiding both the truncation
and the zero-division risk.  If duration is 0 (invalid manifest),
explicitly fall back to first_seq_no instead of crashing.

Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/dashdec.c

Change #274289

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Tue 07 Jul 2026 20:31:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6a351a3113e407885a6ca6787b55fbf1b78cbee5

Comments

avformat/dashdec: check fragment_timescale before division in refresh_manifest
fix issue: issues/23057 POC4

In refresh_manifest(), when timelines are present, fragment_timescale is
used as the denominator to compute currentTime. If fragment_timescale is
zero (e.g., timescale attribute missing from a crafted manifest with
SegmentTimeline), this results in a division by zero.

Add fragment_timescale > 0 guard to both the video and audio timeline
processing paths to prevent the crash.

Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/dashdec.c

Change #274290

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Tue 07 Jul 2026 21:04:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	089e86692c58822e9ac8e56d890066e99a0056f2

Comments

avcodec/bsf: fix codec-id match loop in av_bsf_link
The break statement only exits the inner j loop, so the outer i loop
always runs to the AV_CODEC_ID_NONE terminator and the match result is
never recorded.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/bitstreamfilter.c

Change #274291

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Tue 07 Jul 2026 21:04:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3b08818dd80bf634a44b3db159de05af45c5858d

Comments

avcodec/bsf: fix wrong pad index in av_bsf_link diagnostics
The error-path log loops read pad 0's codec id list instead of the
actual srcpad/dstpad, printing the wrong supported-codec list when a
link is rejected.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/bitstreamfilter.c

Change #274292

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Tue 07 Jul 2026 21:04:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2b5a258ad59496a3c1e69ca71edbaf4e16b7632a

Comments

avcodec/bsf: tighten loop variable scope in av_bsf_link
And replace tab-indented log lines with spaces. No functional change.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/bitstreamfilter.c

Change #274293

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Tue 07 Jul 2026 21:04:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1d5550613bd70343034d949a395ff888b2a8c561

Comments

avcodec/bsf/source: fix packet leak on av_packet_ref failure
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/bsf/source.c

Change #274297

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 08 Jul 2026 05:37:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	741e5630771c2e2068e3f6b4609ada28ca3d1de1

Comments

aarch64/h26x: parameterize calc_all step macro name
So callers can pass a named step macro instead of relying on a locally
redefined `calc` followed by `.purgem calc`.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/aarch64/h26x/epel_neon.S
libavcodec/aarch64/h26x/qpel_neon.S

Change #274298

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 08 Jul 2026 05:37:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ee9aba4c5560cc7236591a09a0928176f66583c4

Comments

aarch64/h26x: give step macros distinct names per function
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/aarch64/h26x/epel_neon.S
libavcodec/aarch64/h26x/qpel_neon.S

Change #274299

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 08 Jul 2026 05:37:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	45bc6c0c08bbbee3493084ff8816aa61350b62b5

Comments

aarch64/h26x: move step macros before their functions
No functional changes.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/aarch64/h26x/epel_neon.S
libavcodec/aarch64/h26x/qpel_neon.S

Change #274300

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Wed 08 Jul 2026 05:54:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	927f205eb8de44fc106a36f00ea9d713c813a4f3

Comments

avutil/hwcontext_vulkan: fix CUDA interop for packed formats
The current export_to_cuda logic only works for planar and semi-planar formats.
When presented with a single plane packed format, it will incorrectly
calculate the number of channels, resulting in failures later on when cuda
code tries to access the frame. Let's fix it.

Changed files

libavutil/hwcontext_vulkan.c

Change #274301

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Wed 08 Jul 2026 05:54:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	30ba440386deb62eb9e489c8939812138a71afe6

Comments

avutil/hwcontext_vulkan: CUDA interop semaphores are per image
The current semaphore logic dates back to a time where we did not have any
multiplane images, and it has not kept up. As a result, we currently try and
manipulate too many semaphores when dealing with multiplane images, leading to
errors and crashes. Let's fix it.

Changed files

libavutil/hwcontext_vulkan.c

Change #274302

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Wed 08 Jul 2026 05:54:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c29d710cd5d0f80bbb56f7ec1f35d5fb7ed44d05

Comments

avutil/hwcontext_vulkan: reject unsupported semi-planar CUDA imports
Cuda currently only supports packed and single-component planar formats, but
fails to import semi-planer (eg: NV12, P010). Even though there are now
semi-planar cuda array formats, which the latest nvdec can use, these formats
are not used when mapping Vulkan imports. Maybe they'll fix that some day.

But until then, let's explicitly detect the case and return a clear error
message for the user.

Exercising this error path revealed that vulkan_free_internal() frees
f->internal on a transfer error and then runs again when the frame is
destroyed, dereferencing the freed pointer; let's make it idempotent.

Changed files

libavutil/hwcontext_vulkan.c

Change #274346

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 08 Jul 2026 14:37:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c57660fb18f058e8ead224e840b242d9c68fd3c4

Comments

avfilter/vf_scale_cuda: fix non-scaling format conversion
A format-only conversion leaves the auto use_filters value unresolved.
Regression since 5d0748243f8.

Fix issue #23737

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavfilter/vf_scale_cuda.c

Change #274352

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Thu 09 Jul 2026 02:35:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ec6c89c1bb0a6b724b5c3930b7f53f1900b2af70

Comments

avcodec/x86/h264_intrapred: SSE2 impl. of pred8x8l_dc_8
Deprecate MMX. Remove the SSSE3 impl. since we no longer use palignr.
The SSE2 impl. uses several more instructions but is slightly faster.

pred8x8l_dc_8_mmxext:                                   20.5 ( 1.66x)
pred8x8l_dc_8_ssse3:                                    17.9 ( 1.90x)
pred8x8l_dc_8_sse2:                                     15.9 ( 2.15x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #274356

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 09 Jul 2026 04:53:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	15bbf3a21d11847de21aa429ec22acf9424953ec

Comments

avformat/spdif: fix reading past the buffer when probing AAC in IEC 61937
Fixes: read of uninitialized memory
Fixes: 528525234/clusterfuzz-testcase-minimized-ffmpeg_dem_W64_fuzzer-6429204661731328
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libavformat/spdifdec.c

Change #274357

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 09 Jul 2026 04:54:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8de8405796df0c11c52772dd14bf6ed552d25c07

Comments

avcodec/tdsc: reject out-of-frame cursor position
Fixes: signed integer overflow
Fixes: 519649309/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TDSC_fuzzer-6322382028734464
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libavcodec/tdsc.c

Change #274394

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Thu 09 Jul 2026 15:25:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8ad6288553ffb99f1260d4770683f9d6c367e16b

Comments

avfilter/vf_libplacebo: add flag to inherit the input's Vulkan device
vf_libplacebo is unique in that it can function as both a Vulkan filter, and a
software filter, depending on how it the filter graph is configured. While this
is very flexible, it create a problem in situations where the filter does not
receive a Vulkan hw frames context up front. When that doesn't happen, it will
initialise its own standalone pl context, and then fail to interoperate with
a context provided via the input link. This then leads to graph failures.

There are two primary scenarios where this existing logic breaks what should
be valid configurations:
* When the global filter hw device is a cuda device and we use something like
  `hwupload=derive_device=vulkan` to pass frames from cuda to vulkan
* In mpv (and probably other media players) which don't set the global filter
  hw device at all. In this case, it's impossible to configure vf_libplacebo
  to use a hw frames ctx, even if it's using Vulkan for everythng else. This
  prevents the use of vf_libplacebo in any fully hardware accelerated pipeline
  in mpv

There are various ways we could imagine addressing it - such as allowing the
filter to discard the initial pl context and recreating it based on the passed
in device, but it's easier to reason about if we add a flag that explicitly
tells the filter that it should inherit the device context from the input link.
This puts the filter into a mode that works like all the other Vulkan filters.

This requires explicit configuration from the user, but the intent is clearer,
and the user can always know when it's necessary as they define the filter
graph.

Changed files

doc/filters.texi
libavfilter/vf_libplacebo.c

Change #274415

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Thu 09 Jul 2026 20:39:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b4b30ff8fd1580c4fc1330afdf6c8fde242e1754

Comments

libavutil/vulkan: fix elem_size computation in vulkan_export_to_cuda

Changed files

libavutil/hwcontext_vulkan.c

Change #274430

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 09 Jul 2026 23:39:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	40ce1513c6a883bc2b96d33d968013b89dfe76af

Comments

avcodec/mjpegdec: reject flipping of bayer images
Fixes: assertion failure
Fixes: 523522308/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_DEC_fuzzer-4771843818192896
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/mjpegdec.c

Change #274446

Category	ffmpeg
Changed by	Vignesh Venkat <vigneshvohnoyoudont@google.com>
Changed at	Fri 10 Jul 2026 03:56:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c4e41b54cd9fba6f73ee4500a459430ed917722

Comments

avcodec: Add itut-t35 metadata as a codec type
Signed-off-by: Vignesh Venkat <vigneshv@google.com>

Changed files

doc/APIchanges
libavcodec/codec_desc.c
libavcodec/codec_id.h
libavcodec/version.h

Change #274447

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Fri 10 Jul 2026 05:20:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0baa71b53cd9dc6ecf0cf9a5c271bcbde04135ba

Comments

avutil/hwcontext_vulkan: query the actual image format in try_export_flags
GetPhysicalDeviceImageFormatProperties2() must be queried with the format
that will actually be used to create the image, otherwise the external
memory export capability check answers for a different format than the one
being allocated.

When AV_VK_FRAME_FLAG_DISABLE_MULTIPLANE is set (e.g. by ffplay for CUDA
interop), vulkan_frames_init selects the per-plane fallback format via
vkfmt_from_pixfmt2(disable_multiplane=1) and stores it in hwctx->format[].
The image is then created with that fallback format. Querying the
multiplane vkf instead returns capability for a format that is never
instantiated, and on NVIDIA GPUs with OPTIMAL tiling the answer differs
between the two formats, producing broken CUDA hwaccel output.

Read hwctx->format[0], which is populated by vulkan_frames_init before
vulkan_pool_alloc runs, so the query matches the image regardless of
whether the multiplane or fallback path was taken.

Fix artifacts with `ffplay -hwaccel cuda foo.mp4`.

Changed files

libavutil/hwcontext_vulkan.c

Change #274449

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 10 Jul 2026 06:41:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d09d5afc3aebede25d2d245ee23b75a47ea17c3a

Comments

avutil/hwcontext_vulkan: declare cuda as a transfer-compatible format
The cuda hwcontext has listed AV_PIX_FMT_VULKAN among its supported
formats since Vulkan interop was first added, so a device's frame
constraints advertise Vulkan as a transfer-compatible ("sw") format.
This is what allows an hwupload targeting a cuda device to accept a
Vulkan frame as input during format negotiation.

The Vulkan hwcontext never made the reciprocal declaration, so an
hwupload targeting a Vulkan device would not accept a cuda frame, and a
cuda -> Vulkan upload could not be configured even though the transfer
itself is supported. Advertise AV_PIX_FMT_CUDA in the Vulkan frame
constraints to ensure we have consistent behaviour in both directions.

Changed files

libavutil/hwcontext_vulkan.c

Change #274474

Category	ffmpeg
Changed by	Tobias Rapp <t.rappohnoyoudont@noa-archive.com>
Changed at	Fri 10 Jul 2026 12:12:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	35f8f4bdc075db1ae63d999edf7ae682fcaf3cf7

Comments

avformat/wavdec: limit memory allocated for BEXT coding history data
Avoid excessive memory allocation when parsing coding history data from
BWF/WAV files by using a size limit of 1MB. When the chunk size limit is
exceeded read fails with an error log message.

Patch-by: Leander-Amz in https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20642#issuecomment-48126

Changed files

libavformat/wavdec.c

Change #274504

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Fri 10 Jul 2026 19:45:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	19a2758e55a6edcc185061958740fcac4b83378f

Comments

avformat/mvrdec: add MVR CCTV demuxer
The format stores video-only H.264 in a fixed header + index table +
chunk layout. Each chunk is a self-contained GOP, enabling simple
random access via the index.

Fix #23540

Changed files

Changelog
libavformat/Makefile
libavformat/allformats.c
libavformat/mvrdec.c
libavformat/version.h

Change #274509

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 11 Jul 2026 04:09:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7c116b536d0d64c96dd15b096aaab1a76cce3b0

Comments

avcodec/mjpegdec: reject redundant scans of a sequential image
Fixes: Timeout
Fixes: 527085539/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AMV_DEC_fuzzer-6502692927897600
Fixes: 523663551/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MJPEG_DEC_fuzzer-5493410811936768
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/mjpegdec.c
libavcodec/mjpegdec.h

Change #274510

Category	ffmpeg
Changed by	Nil Fons Miret <nilfohnoyoudont@netflix.com>
Changed at	Sat 11 Jul 2026 04:46:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6bae3bd38558363542fca509cb7677d727f2beb9

Comments

avformat/wavdec: move ID3v2 auto-parsing flag from w64 to wav
9d037c54f2 (avformat/demux: avoid unconditional ID3v2 tag consumption)
introduced FF_INFMT_FLAG_ID3V2_AUTO to opt formats into generic ID3v2
parsing, but mistakenly set it on ff_w64_demuxer instead of
ff_wav_demuxer (both live in wavdec.c, with w64 at the end of the file).
As a result, WAV files carrying a prepended ID3v2 tag were no longer
parsed generically, while w64 attempted to parse an ID3v2 tag it didn't
need. Move the flag to the wav demuxer, restoring the originally
intended behavior.

Reported-by: Daniel Verkamp <daniel@drv.nu>
Signed-off-by: Nil Fons Miret <nil.fons@gmail.com>

Changed files

libavformat/wavdec.c

Change #274514

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 11 Jul 2026 10:10:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	300cac3078185718de65bfd66b75b5dcfc57a5c0

Comments

apv_decode: call ff_thread_finish_setup()
"[...] decoders with update_thread_context callback have to call
ff_thread_finish_setup() themselves; it is not done implicitly in
ff_thread_get_buffer() any more.
Not doing so makes the decoder effectively single-threaded."

Changed files

libavcodec/apv_decode.c

Change #274520

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 11 Jul 2026 15:45:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1f2484023936c4da95035459ebf11fc7730f18af

Comments

avdevice/alsa: auto-detect channel layout when capturing audio

Changed files

doc/indevs.texi
libavdevice/alsa.c
libavdevice/alsa.h
libavdevice/alsa_dec.c

Change #274521

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sat 11 Jul 2026 16:10:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	87bcbbff5722867d1d9e82c12299bd23382782c8

Comments

avcodec/libfdk-aacenc: don't propagate a skip samples value higher than the frame size
This puts the burden of keeping track of remaining samples on the user.
It's best if we just report the exact amount of discardable samples
within the packet's duration.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libfdk-aacenc.c

Change #274522

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sat 11 Jul 2026 16:10:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c31a781412a3aa801cc0997ba001177f258f4d80

Comments

avcodec/audio_frame_queue: add skip samples side data to output packets
Instead of reimplementing the same code on every afq user, just do it here.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/aacenc.c
libavcodec/ac3enc.c
libavcodec/aptxenc.c
libavcodec/audio_frame_queue.c
libavcodec/audio_frame_queue.h
libavcodec/audiotoolboxenc.c
libavcodec/libfdk-aacenc.c
libavcodec/libmp3lame.c
libavcodec/libopencore-amr.c
libavcodec/libopusenc.c
libavcodec/libshine.c
libavcodec/libspeexenc.c
libavcodec/libvorbisenc.c
libavcodec/mlpenc.c
libavcodec/nellymoserenc.c
libavcodec/opus/enc.c
libavcodec/ra144enc.c
libavcodec/vorbisenc.c
tests/ref/fate/ac3-fixed-encode-2
tests/ref/fate/ac3-fixed-encode-3
tests/ref/fate/copy-shortest1
tests/ref/fate/copy-shortest2
tests/ref/fate/ffmpeg-filter_complex_audio
tests/ref/fate/opus-enc-silence
tests/ref/fate/shortest

Change #274523

Category	ffmpeg
Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
Changed at	Sat 11 Jul 2026 16:23:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	24f3ab761e4648105d5734f72f94c3e2ddd9c9fa

Comments

lavc/hevcdec: precompute the significance context tables per scan
The significance flag loop looked up scan_x_off/scan_y_off and then
ctx_idx_map for every decoded bin. The composition of the two is
constant per scan type, so precompute the three permuted tables and
index them by scan position directly. The SCAN_HORIZ table is the
original one, horizontal scan order being raster order.

About 2.5% faster single threaded decoding on high bitrate streams.

Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

Changed files

libavcodec/hevc/cabac.c

Change #274524

Category	ffmpeg
Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
Changed at	Sat 11 Jul 2026 16:23:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ea8ca20168f103733e10d90e2954d5f09f9a8a13

Comments

lavc/hevcdec: skip the coefficient clear for DC-only blocks
idct_dc reads coeffs[0] only and writes the whole block, so the
clear before residual decoding is not needed on that path. Move it
after the last significant coefficient position is known and compute
col_limit once instead of at the transform call.

Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

Changed files

libavcodec/hevc/cabac.c

Change #274525

Category	ffmpeg
Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
Changed at	Sat 11 Jul 2026 16:23:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5d6f409cdd65c3943d1b7447524008bba4ac36c9

Comments

lavc/hevcdec: decode runs of bypass bins with one integer division
Reading a bypass bin doubles the offset and conditionally subtracts
range << 17, which is one step of a binary long division: a run of n
bins is the quotient of the offset, shifted and refilled upfront, by
range << 17, and the new offset is its remainder. Decisions only
depend on multiples of 2^17, so the reordered refill cannot change
any of them; a quotient overflowing n bits, which needs an offset
closer to the range than the pending low bits can distinguish, falls
back to the serial loop.

Use it for the coefficient sign runs, the Golomb-Rice suffixes and
the unary prefixes, whose length is the number of leading ones of a
quotient peeked without committing the bins past the terminating
zero. Sign bins are equiprobable by construction, so on top of the
saved renormalisation work this removes data dependent branch
mispredictions. About 3% faster single threaded decoding on high
bitrate streams.

Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

Changed files

libavcodec/hevc/cabac.c

Change #274526

Category	ffmpeg
Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
Changed at	Sat 11 Jul 2026 16:23:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4ff36f83eb5c644de7a694ffe8060b19d63b0c4f

Comments

lavc/hevcdec: remove data dependent branches from residual decoding
Freshly decoded bins are close to equiprobable, so branching on them
mispredicts half of the time. Apply the coefficient signs and the
clipping with the xor/subtract idiom and FFMIN/FFMAX, store the
significant coefficient positions unconditionally while advancing the
count by the flag value, and derive the greater1 contexts and the
first greater1 index arithmetically from a flag mask. About 7% faster
single threaded decoding on high bitrate streams.

Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

Changed files

libavcodec/hevc/cabac.c

Change #274527

Category	ffmpeg
Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
Changed at	Sat 11 Jul 2026 16:23:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6179982c9c0ae6ff5092b4df3a173a79e2227785

Comments

avcodec/x86/cabac: compute renormalisation shifts with lzcnt
The norm_shift loads sit on the serial range dependency chain and
cost several cycles where lzcnt costs one. Builds targeting CPUs
without LZCNT keep the exact original code. About 1.5% faster CABAC
heavy H.264 and HEVC decoding.

Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

Changed files

libavcodec/x86/cabac.h

Change #274531

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 11 Jul 2026 19:14:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c1be9b4a81774105f355a967a59bc0e4bdbd4e97

Comments

Add FATE tests for stale metadata during re-encoding.

Changed files

tests/fate/id3v2.mak
tests/fate/matroska.mak
tests/fate/mov.mak
tests/ref/fate/id3v2-reenc-delete-metadata
tests/ref/fate/id3v2-reenc-delete-metadata-map-metadata
tests/ref/fate/id3v2-reenc-remux-keep
tests/ref/fate/matroska-reenc-chapter-nofilter
tests/ref/fate/matroska-reenc-delete-metadata
tests/ref/fate/matroska-reenc-delete-metadata-filter-output
tests/ref/fate/matroska-reenc-delete-metadata-spec-scope
tests/ref/fate/mov-reenc-delete-format-metadata
tests/ref/fate/mov-reenc-delete-stream-metadata
tests/ref/fate/mov-reenc-delete-stream-metadata-global-tag

Change #274534

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 11 Jul 2026 19:14:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4641be96e5c97c3a91bba0980cb7377586156195

Comments

fftools/ffmpeg_mux_init.c: delete encoder-specific metadata when re-encoding.

Changed files

doc/ffmpeg.texi
fftools/ffmpeg.h
fftools/ffmpeg_mux_init.c
fftools/ffmpeg_opt.c
tests/fate/id3v2.mak
tests/fate/matroska.mak
tests/fate/mov.mak
tests/ref/fate/caf-alac-remux
tests/ref/fate/cover-art-flac-remux
tests/ref/fate/cover-art-mp3-id3v2-remux
tests/ref/fate/enhanced-flv-hevc
tests/ref/fate/enhanced-flv-multitrack
tests/ref/fate/generic-tags-remux-asf
tests/ref/fate/generic-tags-remux-riff
tests/ref/fate/generic-tags-remux-vorbiscomment
tests/ref/fate/h264-bsf-redundant-pps-side-data
tests/ref/fate/h264-bsf-redundant-pps-side-data2
tests/ref/fate/id3v2-reenc-delete-metadata
tests/ref/fate/id3v2-reenc-delete-metadata-keep
tests/ref/fate/id3v2-reenc-delete-metadata-keep-format
tests/ref/fate/id3v2-reenc-delete-metadata-keep-stream
tests/ref/fate/id3v2-reenc-delete-metadata-map-metadata
tests/ref/fate/matroska-alac-remux
tests/ref/fate/matroska-crop
tests/ref/fate/matroska-dovi-hvce-mp4-to-mkv
tests/ref/fate/matroska-dovi-write-config7
tests/ref/fate/matroska-dovi-write-config8
tests/ref/fate/matroska-non-rotation-displaymatrix
tests/ref/fate/matroska-reenc-chapter-nofilter
tests/ref/fate/matroska-reenc-delete-audio-metadata
tests/ref/fate/matroska-reenc-delete-metadata
tests/ref/fate/matroska-reenc-delete-metadata-filter-output
tests/ref/fate/matroska-reenc-delete-metadata-keep
tests/ref/fate/matroska-reenc-delete-metadata-keep-prefix-nomatch
tests/ref/fate/matroska-reenc-delete-metadata-keep-stream-video
tests/ref/fate/matroska-reenc-delete-metadata-keep-stream-video-suffix
tests/ref/fate/matroska-reenc-delete-metadata-keep-stream-video1
tests/ref/fate/matroska-reenc-delete-metadata-spec-scope
tests/ref/fate/mov-cover-reenc-keeps-audio-format-tags
tests/ref/fate/mov-reenc-delete-format-metadata
tests/ref/fate/mov-reenc-delete-stream-metadata
tests/ref/fate/mov-reenc-delete-stream-metadata-global-tag
tests/ref/fate/rgb24-mkv

Change #274537

Category	ffmpeg
Changed by	KawaiiEngine (Sashimiso) <209421934+KawaiiEngineohnoyoudont@users.noreply.github.com>
Changed at	Sat 11 Jul 2026 19:17:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1814fc7e4a79229e1fcd34a2ed5b8b6f53f9f23d

Comments

avutil/hwcontext_cuda: fail derive if no CUDA device matches
When CUDA enumeration succeeds but none of the device UUIDs matches the
source Vulkan device, ret still holds zero from the last successful CUDA
call.  The error path therefore uninitializes the destination and returns
success.

av_hwdevice_ctx_create_derived() then returns a CUDA device context without
a CUDA context.  Later frame allocation fails with an unrelated error.

Set ret to AVERROR(ENODEV) for the no-match case, consistent with equivalent
Vulkan and OpenCL device selection failures.

Signed-off-by: KawaiiEngine (Sashimiso) <209421934+KawaiiEngine@users.noreply.github.com>

Changed files

libavutil/hwcontext_cuda.c

Change #274539

Category	ffmpeg
Changed by	Frank Plowman <postohnoyoudont@frankplowman.com>
Changed at	Sat 11 Jul 2026 22:36:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	265d39e551956d911a0c1c52bff5186a6bae660e

Comments

lavc/hevc: Always generate missing refs for foll pictures
Since bc1a3bfd2cbc01ffa386312662af8a014890d861, missing reference
pictures are not replaced with generated pictures unless `-flags
show_corrupt` is used.  For `ST_FOLL`/`LT_FOLL` reference pictures,
however, generation of missing references is required by the spec per
8.3.3.  We should not require the `show_corrupt` flag to be used in
order to be spec-conformant, so this patch removes the `show_corrupt`
behaviour added in bc1a3bfd2cbc01ffa386312662af8a014890d861 for
`ST_FOLL`/`LT_FOLL` references and instead unconditionally generates
reference pictures in these sets when unavailable

Fixes: NUT_A_ericsson_5
Fixes: RPS_D_ericsson_6

Changed files

libavcodec/hevc/refs.c
tests/fate/hevc.mak

Change #274542

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 11 Jul 2026 23:53:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	501d8eb62dd4f91dd19fb874f555db7ca3fde317

Comments

avcodec/bsf/dts2pts: chain same-gop nodes instead of collecting into a fixed array
hevc_queue_frame() collected all tree nodes of the current gop into a
fixed nodes[HEVC_MAX_DPB_SIZE * 2] array and asserted the count stayed
within it. For a crafted HEVC stream the tree can hold slightly more
same-gop nodes than that estimate (observed 34 vs 32), tripping the
av_assert0 and aborting. Link the matching nodes into a list through a
next pointer in the node instead; this has no size limit and needs
neither extra passes nor allocation.

Fixes: assertion failure
Fixes: 519466146/clusterfuzz-testcase-minimized-ffmpeg_BSF_DTS2PTS_fuzzer-5238235193475072
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Co-Authored-By: Fable-5

Changed files

libavcodec/bsf/dts2pts.c

Change #274543

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 11 Jul 2026 23:53:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4133db39b2bd1491e41457e819ed3ca96ef555ff

Comments

avcodec/bsf/dts2pts: fix removal of 2nd field nodes
alloc_and_insert_node() inserts the nodes for 2nd fields with
duration / poc_diff added to the timestamp, but the removal loop
compared all nodes of a frame against the unadjusted first timestamp,
so 2nd field nodes never matched and stayed in the tree until close.
Advance the compared timestamp the same way the insertion does.

Fixes: tree nodes leaking on every field coded frame

Co-Authored-By: Fable-5

Changed files

libavcodec/bsf/dts2pts.c

Change #274544

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 11 Jul 2026 23:53:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dd941af8ac9c97017b335832714457fe9dd7f6be

Comments

avcodec/bsf/dts2pts: evict unconsumable nodes from the poc tree
Each pending packet consumes up to poc_diff tree nodes when it is
output, so the packets in the FIFO can consume at most nb_pending nodes
in total. Frames whose tree lookup misses on output leave their nodes
behind, and damaged or crafted streams can make that happen
indefinitely, growing the tree without limit.

Track the node count and insertion order, keep the leftovers of up to
MAX_DAMAGED_FRAMES frames and beyond that evict the nodes unconsumed
the longest. Timestamps of valid frames are unaffected unless more
frames than that are damaged. No eviction is done at EOF, where nodes
are deliberately kept to regenerate timestamps from.

Of all h264/hevc conformance samples only MR3_TANDBERG_B.264 triggers
evictions, and no sample changes output.

Fixes: unbounded memory growth with damaged streams

Co-Authored-By: Fable-5

Changed files

libavcodec/bsf/dts2pts.c

Change #274545

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	56e39bd20b418b9298488bd5a261a0ee5ae21525

Comments

avutil/downmix_info: add a new API to propagate pre-made downmix matrixes
Useful for formats where they provide the downmix matrix in the bistream rather
than scale factors for channel groups.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/APIchanges
libavutil/downmix_info.c
libavutil/downmix_info.h
libavutil/version.h

Change #274546

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7e68fdd4dabc02cb76c7924398928a49031847a8

Comments

avutil/frame: add a new side data type for downmix matrix
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/APIchanges
libavutil/frame.h
libavutil/side_data.c
libavutil/version.h

Change #274547

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6336cffdceeece9b7d222d5dc88705db08f88aac

Comments

avfilter/af_ashowinfo: support printing downmix matrix side data
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavfilter/af_ashowinfo.c

Change #274548

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9646228fd676ff59168dd546924705d2fb3f250c

Comments

swresample/rematrix: print custom downmix matrix coefficients
Similar to how it's done to the matrix built in auto_matrix().

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libswresample/rematrix.c

Change #274549

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b662613b63e902e3978bc3977e4ad3ccb738110f

Comments

avfilter/af_aresample: also look for downmix matrix side data
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavfilter/af_aresample.c

Change #274550

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e1e6ebb065ebfb2a08a4adc4b4aae0a85640018d

Comments

fftools/ffmpeg_filter: propagate downmix matrix side data to the filterchain
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_filter.c

Change #274551

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	57708daf619c1de0ea25366ef11e527f292acd55

Comments

avcodec/dca_core: export downmix matrix coefficients
$ ./ffmpeg -downmix stereo -i $fate-suite/dts/dcadec-suite/core_51_24_48_768_1.dtshd -y native.wav

Before:
$ ./ffmpeg -i $fate-suite/dts/dcadec-suite/core_51_24_48_768_1.dtshd -af aresample,aformat=channel_layouts=stereo -y swr.wav
[...]
[Parsed_aresample_0] [SWR] Using fltp internally between filters
[Parsed_aresample_0] [SWR] Matrix coefficients:
[Parsed_aresample_0] [SWR] FL: FL:0.414214 FR:0.000000 FC:0.292893 LFE:0.000000 SL:0.292893 SR:0.000000
[Parsed_aresample_0] [SWR] FR: FL:0.000000 FR:0.414214 FC:0.292893 LFE:0.000000 SL:0.000000 SR:0.292893
[Parsed_aresample_0] ch:6 chl:5.1(side) fmt:fltp r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[...]

$ tests/tiny_psnr.exe native.wav swr.wav s16
stddev:18907.06 PSNR: 10.80 MAXDIFF:51590 bytes:     8192/     8192

After:
$ ./ffmpeg -i $fate-suite/dts/dcadec-suite/core_51_24_48_768_1.dtshd -af aresample,aformat=channel_layouts=stereo -y swr.wav
[...]
[Parsed_aresample_0] [SWR] Using fltp internally between filters
[Parsed_aresample_0] [SWR] Custom matrix coefficients:
[Parsed_aresample_0] [SWR] FL: FL:-0.100006 FR:0.000000 FC:0.000000 LFE:0.000000 SL:-0.707092 SR:0.000000
[Parsed_aresample_0] [SWR] FR: FL:0.000000 FR:-0.100006 FC:0.000000 LFE:0.000000 SL:0.000000 SR:-0.707092
[Parsed_aresample_0] ch:6 chl:5.1(side) fmt:fltp r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[...]

$ tests/tiny_psnr.exe native.wav swr.wav
stddev:    0.02 PSNR:129.44 MAXDIFF:    1 bytes:     8192/     8192

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/dca_core.c
libavcodec/dcadec.c
libavcodec/dcadec.h
tests/fate/dca.mak

Change #274552

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 00:32:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c16840b826584b7682078cec4d5e8405b06f03c8

Comments

avcodec/dca_xll: export downmix matrix coefficients
Only for primary channel sets if they embeed coefficients for now.

$ ./ffmpeg -downmix stereo -i $fate-suite/dts/dcadec-suite/xll_51_24_48_768.dtshd -y native.wav

Before:
$ ./ffmpeg -i $fate-suite/dts/dcadec-suite/xll_51_24_48_768.dtshd -af aresample,aformat=channel_layouts=stereo -y swr.wav
[...]
[Parsed_aresample_0] [SWR] Using fltp internally between filters
[Parsed_aresample_0] [SWR] Matrix coefficients:
[Parsed_aresample_0] [SWR] FL: FL:0.414214 FR:0.000000 FC:0.292893 LFE:0.000000 SL:0.292893 SR:0.000000
[Parsed_aresample_0] [SWR] FR: FL:0.000000 FR:0.414214 FC:0.292893 LFE:0.000000 SL:0.000000 SR:0.292893
[Parsed_aresample_0] ch:6 chl:5.1(side) fmt:s32p r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[...]

$ tests/tiny_psnr.exe native.wav swr.wav
stddev:21178.05 PSNR:  9.81 MAXDIFF:49613 bytes:     8192/     8192

After:
$ ./ffmpeg -i $fate-suite/dts/dcadec-suite/xll_51_24_48_768.dtshd -af aresample,aformat=channel_layouts=stereo -y swr.wav
[...]
[Parsed_aresample_0] [SWR] Using fltp internally between filters
[Parsed_aresample_0] [SWR] Custom matrix coefficients:
[Parsed_aresample_0] [SWR] FL: FL:-0.100006 FR:0.000000 FC:0.000000 LFE:0.000000 SL:-0.707092 SR:0.000000
[Parsed_aresample_0] [SWR] FR: FL:0.000000 FR:-0.100006 FC:0.000000 LFE:0.000000 SL:0.000000 SR:-0.707092
[Parsed_aresample_0] ch:6 chl:5.1(side) fmt:s32p r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[...]

$ tests/tiny_psnr.exe native.wav swr.wav
stddev:    0.71 PSNR: 99.22 MAXDIFF:    1 bytes:     8192/     8192

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/dca_xll.c
tests/fate/dca.mak
tests/ref/fate/dca-xll_51_24_48_768-dmix_2_matrix

Change #274553

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Sun 12 Jul 2026 02:02:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	683fae35e1bf9cc69d903eacc53cf5dde0ca7c7a

Comments

avformat/internal: fix typo allow_unkown -> allow_unknown
The declaration of ff_parse_opts_from_query_string() misspelled
"unknown" as "unkown". The implementation in utils.c already uses
the correct spelling.

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavformat/internal.h

Change #274554

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Sun 12 Jul 2026 02:02:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a6ed750664d6d8760bcee88f8aacec2e7d1b3ab7

Comments

avformat/utils: fix ff_mkdir_p() swallowing intermediate mkdir errors
When creating nested directories (e.g. /a/b/c), a genuine mkdir()
failure for an intermediate component was overwritten by later
attempts, making the original failure harder to diagnose.

Stop immediately on intermediate errors other than EEXIST, preserving
errno for the caller. Existing path components remain non-fatal, as
required by mkdir -p semantics. Add a regression test for creating a
child below an existing parent directory.

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavformat/Makefile
libavformat/tests/.gitignore
libavformat/tests/mkdir.c
libavformat/utils.c
tests/fate/libavformat.mak

Change #274555

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Sun 12 Jul 2026 02:02:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8656a45be55a424af93e052e386ccac2ac46608b

Comments

avformat/tls: remove orphan ff_openssl_init/ff_openssl_deinit declarations
These declarations were left behind by f0913b3f16 which
removed support for OpenSSL < 1.1.0. Since OpenSSL >= 1.1.0
auto-initializes, neither function has an implementation or
caller.

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavformat/tls.h

Change #274557

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Sun 12 Jul 2026 02:02:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fe38a1e257f85ad2e16e4b3bd0bde9f6873592e7

Comments

avformat/network: make ff_network_init() follow standard AVERROR convention
ff_network_init() used to return 0 on failure and 1 on success,
requiring callers to check !ff_network_init(). Change it to return
a negative AVERROR(EIO) on failure and 0 on success, matching the
standard FFmpeg convention where negative values indicate errors.

Update all callers (avio.c, rtsp.c, rtspdec.c, sapdec.c, sapenc.c,
utils.c) to use the standard (ret = ff_network_init()) < 0 pattern.

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavformat/avio.c
libavformat/network.c
libavformat/rtsp.c
libavformat/rtspdec.c
libavformat/sapdec.c
libavformat/sapenc.c

Change #274561

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 12 Jul 2026 06:34:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	30b6ccb59af96dd1fae7668fdbf503527a37c2fc

Comments

vulkan: fix u32vec2buf/u32vec4buf alignment values
We were optimistic, looking through the GLSL specs, it was illegal to use
an alignment value which satisfied scalar alignment when doing vector loads.
At best, the GPU compiler split it up into multiple parts.

Changed files

libavcodec/vulkan/common.glsl

Change #274562

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 12 Jul 2026 06:34:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	43d0d365f3bb4ca9f6e5e6615a4f7aecc6e9df9d

Comments

ffv1enc_vulkan: fix VRAM size detection
The reduction accumulated into the zero-initialized context field
instead of the local variable, so the detected VRAM was whatever
device-local heap the implementation listed last, usually the
256MiB BAR heap rather than the multi-GiB VRAM heap.

This made the encoder take the host-memory fallback (or fail the
allocation outright at 12K and above) on hardware with plenty of VRAM.

Changed files

libavcodec/ffv1enc_vulkan.c

Change #274563

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 12 Jul 2026 06:34:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	720bc0b916c0396ada52490cc75c652d42ccf9c7

Comments

vulkan/seg_gather: avoid scratch spills and serial prefix summing
Compilers lowered the funnel indexer to scratch (spilling the data),
during the tight copy loop into something that spills every load.
Just use a case to select from the 2 source vectors.

Summing the offsets at the end took far longer than expected and
blocked all invocs. Just use a trivial subgroup add + atomic add
to finish the sum quickly and with less contention.

Changed files

libavcodec/vulkan/seg_gather.comp.glsl

Change #274564

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 12 Jul 2026 06:34:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	48236ce1a8c0d37c31ce5c6394e4bd284b8b3bc1

Comments

ffv1enc_vulkan: gather into VRAM and then copy, drop the host-only path
The gather shader wrote the finished bitstream directly into a
host visible buffer with BDA stores.

Keep the gather output in VRAM and append a single copy-engine
transfer to the same submission to move it into the host-visible
readback buffer. DMA readback into cached system memory is the one
path every driver optimizes. Still no CPU involvement.

The host path was awfully slow and only really used because of FFv1's
insane overallocation.

Changed files

libavcodec/ffv1enc_vulkan.c

Change #274565

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 12 Jul 2026 06:34:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a09be9b91e8e1219f297586873b0d7322b47df96

Comments

apv_encode_vulkan: read the gathered bitstream back with the copy engine
Same as FFv1.

Changed files

libavcodec/apv_encode_vulkan.c

Change #274616

Category	ffmpeg
Changed by	Shreesh Adiga <16567adigashreeshohnoyoudont@gmail.com>
Changed at	Sun 12 Jul 2026 21:44:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7b9eeadda06edd6ef7f604d5c3863f30722c7c15

Comments

tests/checkasm/aes: fix memcmp size argument
sizeof(16 * count) evaluates to 4 and thus only the first 4 bytes of aes
output was being compared. Removing the sizeof operator will ensure that
the whole output buffer will be compared.

Changed files

tests/checkasm/aes.c

Change #274639

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2e3bc3ebb268666146e4a902a3afe8e8dbd802ad

Comments

avcodec/x86/vp9lpf: Combine comparisons
The VP9 loop filter involves comparing several unsigned
values against the same value and combines the results:
a <= I && b <= I && ...
This is equivalent to max(a,b,...) <= I and using this
form is beneficial: Each comparison needs one pxor*
and one pcmpgtb and every comparison except the first
uses one por to combine the result. Switching to max
instead results in one comparison and (operands - 1)
pmaxub to create the maximum; the latter can be relaxed
to por if the final comparator is one**, as it is for
flat8{in,out}.

Doing so saves 4113B of .text here and is beneficial
for performance:

Old benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                78.0
  vp9_loop_filter_h_4_8_8bpp_mmxext:           74.2 ( 1.05x)
  vp9_loop_filter_h_8_8_8bpp_c:                85.0
  vp9_loop_filter_h_8_8_8bpp_mmxext:          101.9 ( 0.83x)
  vp9_loop_filter_h_16_8_8bpp_c:              140.3
  vp9_loop_filter_h_16_16_8bpp_c:             250.7
  vp9_loop_filter_h_16_16_8bpp_sse2:          176.1 ( 1.39x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         172.9 ( 1.38x)
  vp9_loop_filter_h_16_16_8bpp_avx:           165.5 ( 1.46x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        167.5
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      70.5 ( 2.32x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     69.7 ( 2.40x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       70.3 ( 2.38x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        175.5
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      93.3 ( 1.87x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     93.3 ( 1.87x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       90.1 ( 1.95x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        177.1
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      94.3 ( 1.82x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     90.6 ( 1.92x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       88.8 ( 1.99x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        186.4
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      94.7 ( 1.95x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     93.7 ( 1.98x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       89.7 ( 2.07x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        176.0
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      29.7 ( 5.92x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     29.3 ( 6.02x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       30.5 ( 5.76x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        187.7
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      65.1 ( 2.82x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     65.2 ( 2.84x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       56.0 ( 3.35x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        188.3
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      60.7 ( 3.03x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     64.8 ( 2.80x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       56.8 ( 3.21x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        199.4
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      65.8 ( 3.01x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     57.6 ( 3.46x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       56.3 ( 3.53x)
  vp9_loop_filter_v_4_8_8bpp_c:                82.8
  vp9_loop_filter_v_4_8_8bpp_mmxext:           32.4 ( 2.50x)
  vp9_loop_filter_v_8_8_8bpp_c:                91.9
  vp9_loop_filter_v_8_8_8bpp_mmxext:           65.6 ( 1.40x)
  vp9_loop_filter_v_16_8_8bpp_c:              156.2
  vp9_loop_filter_v_16_16_8bpp_c:             291.3
  vp9_loop_filter_v_16_16_8bpp_sse2:          125.7 ( 2.25x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         123.1 ( 2.31x)
  vp9_loop_filter_v_16_16_8bpp_avx:           115.3 ( 2.51x)

New benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                78.8
  vp9_loop_filter_h_4_8_8bpp_mmxext:           62.4 ( 1.26x)
  vp9_loop_filter_h_8_8_8bpp_c:                85.8
  vp9_loop_filter_h_8_8_8bpp_mmxext:           86.5 ( 0.97x)
  vp9_loop_filter_h_16_8_8bpp_c:              139.3
  vp9_loop_filter_h_16_16_8bpp_c:             251.3
  vp9_loop_filter_h_16_16_8bpp_sse2:          164.7 ( 1.45x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         164.2 ( 1.49x)
  vp9_loop_filter_h_16_16_8bpp_avx:           155.2 ( 1.56x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        168.3
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      60.1 ( 2.80x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     60.1 ( 2.80x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       60.4 ( 2.79x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        178.1
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      83.5 ( 2.07x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     84.0 ( 2.02x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       78.0 ( 2.28x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        181.7
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      78.3 ( 2.20x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     85.1 ( 2.12x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       79.8 ( 2.27x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        184.8
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      85.6 ( 2.14x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     85.1 ( 2.09x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       77.6 ( 2.38x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        176.0
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      28.1 ( 6.27x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     32.3 ( 5.42x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       27.4 ( 6.43x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        187.4
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      55.3 ( 3.39x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     54.7 ( 3.42x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       53.7 ( 3.48x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        189.6
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      62.5 ( 3.00x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     61.3 ( 3.07x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       54.3 ( 3.47x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        199.5
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      55.1 ( 3.52x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     54.4 ( 3.66x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       53.0 ( 3.69x)
  vp9_loop_filter_v_4_8_8bpp_c:                84.3
  vp9_loop_filter_v_4_8_8bpp_mmxext:           32.0 ( 2.62x)
  vp9_loop_filter_v_8_8_8bpp_c:                94.3
  vp9_loop_filter_v_8_8_8bpp_mmxext:           63.7 ( 1.48x)
  vp9_loop_filter_v_16_8_8bpp_c:              155.8
  vp9_loop_filter_v_16_16_8bpp_c:             292.0
  vp9_loop_filter_v_16_16_8bpp_sse2:          119.5 ( 2.39x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         119.1 ( 2.39x)
  vp9_loop_filter_v_16_16_8bpp_avx:           111.0 ( 2.61x)

*: to shift unsigned values to signed ones for pcmpgtb
**: or any comparison like <= 2^n-1

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274641

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	10232fdd710cd15670119c71375d7d64355cde40

Comments

avcodec/x86/vp9lpf: Don't do work twice
max(abs(p1-p0),abs(q1-q0)) is currently computed twice.
Stop doing so. This saves 1681B of .text and gives
a small performance boost.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274642

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fee9a4fedfc0d89b6213d7b8f1c12efaf7fbaa43

Comments

avcodec/x86/vp9lpf: Improve deriving masks
Deriving the fm, flat8in and flat8out masks involves
calculating max{abs(a_i-a), 1<=i<=N} (with N being two or four).
The current code does this by calculating the abs individually
(two psubusb and a por), followed by combining the results
(another por or pmaxub). But there is a better way:
Calculate the min m and max M of the a_i. Then the result equals
max{(M-a)^+,(a-m)^+}, which can be calculated with two
psubusb, followed by combining the result. E.g. for N=2
this saves two por and exchanges two psubusb with pmaxub/pminub.

Doing so saves 2062B of .text here; performance is basically
flat (the length of the critical section is unchanged).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274643

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b5970d7cb9b5ed1754ab539d5290aae86dc8ef3c

Comments

avcodec/x86/vp9lpf: Avoid reloading constant
pb_80 is still in a register.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274644

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c727bc3480640c266176134eb813b2b046165a7d

Comments

avcodec/x86/vp9lpf: Avoid inverting masks
Change the comparison instead to already yield the inverted mask.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274645

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0c3a212c74399586c671f939be30e8e596335bcb

Comments

avcodec/x86/vp9lpf: Avoid unnecessary stores
P1, P0, Q0 and Q1 are kept in registers throughout the whole
function (when using xmm registers on x64), so they need not
be stored on the stack.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274646

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a619985145c69906707a8040327d0e80dc6aa8ae

Comments

avcodec/x86/vp9lpf: Combine filter2 and filter4 cases
There are two differences between the filter2 and filter4 case.
The first one is

if (hev) {
    int f = av_clip_intp2(p1 - q1, BIT_DEPTH - 1);
    f = av_clip_intp2(3 * (q0 - p0) + f, BIT_DEPTH - 1);
...
} else {
    int f = av_clip_intp2(3 * (q0 - p0), BIT_DEPTH - 1);
...

which can simply be combined to

int f = hev ? av_clip_intp2(p1 - q1, BIT_DEPTH - 1) : 0;
f = av_clip_intp2(3 * (q0 - p0) + f, BIT_DEPTH - 1);

which boils down to masking f by hev after the first
av_clip_intp2().

The second is that two more rows are processed for filter4(),
but one can simply mask the differential there, too.

So one can just combine both filters into one; doing so saves
8457B of .text here and improves performance.

Old benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                77.3
  vp9_loop_filter_h_4_8_8bpp_mmxext:           60.8 ( 1.27x)
  vp9_loop_filter_h_8_8_8bpp_c:                86.3
  vp9_loop_filter_h_8_8_8bpp_mmxext:           82.9 ( 1.04x)
  vp9_loop_filter_h_16_8_8bpp_c:              139.7
  vp9_loop_filter_h_16_16_8bpp_c:             251.9
  vp9_loop_filter_h_16_16_8bpp_sse2:          164.8 ( 1.44x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         164.4 ( 1.49x)
  vp9_loop_filter_h_16_16_8bpp_avx:           155.1 ( 1.59x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        169.2
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      58.0 ( 2.91x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     56.6 ( 2.99x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       57.1 ( 2.96x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        174.9
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      75.4 ( 2.32x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     80.6 ( 2.16x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       75.1 ( 2.28x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        179.7
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      79.5 ( 2.25x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     78.2 ( 2.28x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       76.1 ( 2.32x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        184.5
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      82.5 ( 2.23x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     78.7 ( 2.34x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       74.4 ( 2.48x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        176.0
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      28.4 ( 6.18x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     26.3 ( 6.69x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       26.2 ( 6.72x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        188.7
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      60.4 ( 3.11x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     53.6 ( 3.50x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       51.2 ( 3.64x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        191.7
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      55.2 ( 3.40x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     51.6 ( 3.71x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       50.8 ( 3.77x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        201.1
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      52.1 ( 3.86x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     53.9 ( 3.73x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       51.5 ( 3.90x)
  vp9_loop_filter_v_4_8_8bpp_c:                83.7
  vp9_loop_filter_v_4_8_8bpp_mmxext:           29.7 ( 2.81x)
  vp9_loop_filter_v_8_8_8bpp_c:                93.2
  vp9_loop_filter_v_8_8_8bpp_mmxext:           58.7 ( 1.59x)
  vp9_loop_filter_v_16_8_8bpp_c:              156.5
  vp9_loop_filter_v_16_16_8bpp_c:             292.9
  vp9_loop_filter_v_16_16_8bpp_sse2:          117.8 ( 2.43x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         116.7 ( 2.45x)
  vp9_loop_filter_v_16_16_8bpp_avx:           104.5 ( 2.78x)

New benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                77.5
  vp9_loop_filter_h_4_8_8bpp_mmxext:           54.8 ( 1.41x)
  vp9_loop_filter_h_8_8_8bpp_c:                85.5
  vp9_loop_filter_h_8_8_8bpp_mmxext:           75.0 ( 1.12x)
  vp9_loop_filter_h_16_8_8bpp_c:              139.8
  vp9_loop_filter_h_16_16_8bpp_c:             249.7
  vp9_loop_filter_h_16_16_8bpp_sse2:          154.9 ( 1.54x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         152.9 ( 1.54x)
  vp9_loop_filter_h_16_16_8bpp_avx:           142.1 ( 1.62x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        167.8
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      52.3 ( 3.21x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     52.7 ( 3.18x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       51.9 ( 3.18x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        175.1
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      73.2 ( 2.39x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     74.9 ( 2.28x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       69.0 ( 2.54x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        177.7
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      72.6 ( 2.45x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     71.3 ( 2.48x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       69.3 ( 2.56x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        184.3
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      69.6 ( 2.59x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     70.2 ( 2.62x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       69.3 ( 2.66x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        176.3
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      22.0 ( 7.98x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     21.4 ( 8.23x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       21.6 ( 8.18x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        187.8
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      49.9 ( 3.75x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     47.3 ( 3.96x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       46.1 ( 4.07x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        189.7
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      51.2 ( 3.58x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     48.4 ( 3.90x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       45.9 ( 4.14x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        199.8
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      51.2 ( 3.88x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     50.5 ( 3.93x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       45.4 ( 4.40x)
  vp9_loop_filter_v_4_8_8bpp_c:                83.3
  vp9_loop_filter_v_4_8_8bpp_mmxext:           23.2 ( 3.58x)
  vp9_loop_filter_v_8_8_8bpp_c:                92.6
  vp9_loop_filter_v_8_8_8bpp_mmxext:           52.5 ( 1.76x)
  vp9_loop_filter_v_16_8_8bpp_c:              154.9
  vp9_loop_filter_v_16_16_8bpp_c:             291.3
  vp9_loop_filter_v_16_16_8bpp_sse2:          104.8 ( 2.68x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         107.5 ( 2.64x)
  vp9_loop_filter_v_16_16_8bpp_avx:            97.7 ( 2.90x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274647

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 12 Jul 2026 23:46:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a13c2b1e4f7e39a86a863a7682a6e11cd6d12dba

Comments

avcodec/x86/vp9lpf: Simplify clip_u8(u8+i8)
Just subtract 128 from u8 to make it signed, use paddsb
and convert back to u8. Saves 2477B of .text here.

Old benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                77.5
  vp9_loop_filter_h_4_8_8bpp_mmxext:           54.8 ( 1.41x)
  vp9_loop_filter_h_8_8_8bpp_c:                85.5
  vp9_loop_filter_h_8_8_8bpp_mmxext:           75.0 ( 1.12x)
  vp9_loop_filter_h_16_8_8bpp_c:              139.8
  vp9_loop_filter_h_16_16_8bpp_c:             249.7
  vp9_loop_filter_h_16_16_8bpp_sse2:          154.9 ( 1.54x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         152.9 ( 1.54x)
  vp9_loop_filter_h_16_16_8bpp_avx:           142.1 ( 1.62x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        167.8
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      52.3 ( 3.21x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     52.7 ( 3.18x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       51.9 ( 3.18x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        175.1
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      73.2 ( 2.39x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     74.9 ( 2.28x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       69.0 ( 2.54x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        177.7
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      72.6 ( 2.45x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     71.3 ( 2.48x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       69.3 ( 2.56x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        184.3
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      69.6 ( 2.59x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     70.2 ( 2.62x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       69.3 ( 2.66x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        176.3
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      22.0 ( 7.98x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     21.4 ( 8.23x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       21.6 ( 8.18x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        187.8
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      49.9 ( 3.75x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     47.3 ( 3.96x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       46.1 ( 4.07x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        189.7
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      51.2 ( 3.58x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     48.4 ( 3.90x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       45.9 ( 4.14x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        199.8
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      51.2 ( 3.88x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     50.5 ( 3.93x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       45.4 ( 4.40x)
  vp9_loop_filter_v_4_8_8bpp_c:                83.3
  vp9_loop_filter_v_4_8_8bpp_mmxext:           23.2 ( 3.58x)
  vp9_loop_filter_v_8_8_8bpp_c:                92.6
  vp9_loop_filter_v_8_8_8bpp_mmxext:           52.5 ( 1.76x)
  vp9_loop_filter_v_16_8_8bpp_c:              154.9
  vp9_loop_filter_v_16_16_8bpp_c:             291.3
  vp9_loop_filter_v_16_16_8bpp_sse2:          104.8 ( 2.68x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         107.5 ( 2.64x)
  vp9_loop_filter_v_16_16_8bpp_avx:            97.7 ( 2.90x)

New benchmarks:
  vp9_loop_filter_h_4_8_8bpp_c:                77.1
  vp9_loop_filter_h_4_8_8bpp_mmxext:           52.0 ( 1.48x)
  vp9_loop_filter_h_8_8_8bpp_c:                85.6
  vp9_loop_filter_h_8_8_8bpp_mmxext:           73.3 ( 1.17x)
  vp9_loop_filter_h_16_8_8bpp_c:              139.2
  vp9_loop_filter_h_16_16_8bpp_c:             252.4
  vp9_loop_filter_h_16_16_8bpp_sse2:          149.3 ( 1.60x)
  vp9_loop_filter_h_16_16_8bpp_ssse3:         149.3 ( 1.62x)
  vp9_loop_filter_h_16_16_8bpp_avx:           140.2 ( 1.77x)
  vp9_loop_filter_mix2_h_44_16_8bpp_c:        168.5
  vp9_loop_filter_mix2_h_44_16_8bpp_sse2:      48.4 ( 3.48x)
  vp9_loop_filter_mix2_h_44_16_8bpp_ssse3:     49.6 ( 3.39x)
  vp9_loop_filter_mix2_h_44_16_8bpp_avx:       49.0 ( 3.43x)
  vp9_loop_filter_mix2_h_48_16_8bpp_c:        174.8
  vp9_loop_filter_mix2_h_48_16_8bpp_sse2:      66.0 ( 2.65x)
  vp9_loop_filter_mix2_h_48_16_8bpp_ssse3:     64.9 ( 2.69x)
  vp9_loop_filter_mix2_h_48_16_8bpp_avx:       64.6 ( 2.65x)
  vp9_loop_filter_mix2_h_84_16_8bpp_c:        177.6
  vp9_loop_filter_mix2_h_84_16_8bpp_sse2:      70.5 ( 2.51x)
  vp9_loop_filter_mix2_h_84_16_8bpp_ssse3:     66.8 ( 2.65x)
  vp9_loop_filter_mix2_h_84_16_8bpp_avx:       66.3 ( 2.68x)
  vp9_loop_filter_mix2_h_88_16_8bpp_c:        185.6
  vp9_loop_filter_mix2_h_88_16_8bpp_sse2:      72.1 ( 2.50x)
  vp9_loop_filter_mix2_h_88_16_8bpp_ssse3:     70.7 ( 2.54x)
  vp9_loop_filter_mix2_h_88_16_8bpp_avx:       65.5 ( 2.83x)
  vp9_loop_filter_mix2_v_44_16_8bpp_c:        175.9
  vp9_loop_filter_mix2_v_44_16_8bpp_sse2:      19.2 ( 9.17x)
  vp9_loop_filter_mix2_v_44_16_8bpp_ssse3:     19.2 ( 9.16x)
  vp9_loop_filter_mix2_v_44_16_8bpp_avx:       19.1 ( 9.18x)
  vp9_loop_filter_mix2_v_48_16_8bpp_c:        186.9
  vp9_loop_filter_mix2_v_48_16_8bpp_sse2:      49.6 ( 3.73x)
  vp9_loop_filter_mix2_v_48_16_8bpp_ssse3:     47.7 ( 3.89x)
  vp9_loop_filter_mix2_v_48_16_8bpp_avx:       43.6 ( 4.28x)
  vp9_loop_filter_mix2_v_84_16_8bpp_c:        189.1
  vp9_loop_filter_mix2_v_84_16_8bpp_sse2:      45.8 ( 4.12x)
  vp9_loop_filter_mix2_v_84_16_8bpp_ssse3:     46.6 ( 4.05x)
  vp9_loop_filter_mix2_v_84_16_8bpp_avx:       43.0 ( 4.27x)
  vp9_loop_filter_mix2_v_88_16_8bpp_c:        198.6
  vp9_loop_filter_mix2_v_88_16_8bpp_sse2:      43.5 ( 4.57x)
  vp9_loop_filter_mix2_v_88_16_8bpp_ssse3:     43.3 ( 4.57x)
  vp9_loop_filter_mix2_v_88_16_8bpp_avx:       43.0 ( 4.62x)
  vp9_loop_filter_v_4_8_8bpp_c:                83.2
  vp9_loop_filter_v_4_8_8bpp_mmxext:           21.8 ( 3.81x)
  vp9_loop_filter_v_8_8_8bpp_c:                92.3
  vp9_loop_filter_v_8_8_8bpp_mmxext:           51.5 ( 1.79x)
  vp9_loop_filter_v_16_8_8bpp_c:              155.6
  vp9_loop_filter_v_16_16_8bpp_c:             290.3
  vp9_loop_filter_v_16_16_8bpp_sse2:          105.1 ( 2.68x)
  vp9_loop_filter_v_16_16_8bpp_ssse3:         103.8 ( 2.72x)
  vp9_loop_filter_v_16_16_8bpp_avx:            97.7 ( 2.86x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #274648

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 02:57:26
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	206f2d92db2e67914cb417b9e1efac191c3b0690

Comments

avcodec/exr: bound total decoded pixels by max_pixels
Use the maximum block dimensions decode_block() can assign after clipping
the nominal tile or scanline block to the data window.

Fixes: Timeout
Fixes: 521392254/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EXR_DEC_fuzzer-6740984590565376
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/exr.c

Change #274649

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 02:58:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a39c205ae90f3691a9996a69e7a52c06312ad32

Comments

avcodec/vorbisdec: skip the MDCT for unused channels
Fixes: Timeout
Fixes: 490144337/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VORBIS_DEC_fuzzer-4539724776931328
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libavcodec/vorbisdec.c

Change #274650

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 03:07:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ed4f286a1049bcd394ab9a5ce71e297f6cbba002

Comments

avcodec/ffv1dec: mask the fltmap index on the 8bit remap path
Fixes: out of array access
Fixes: ffv1_poc/test_ffv1_8bit_v4.avi
Fixes: 6p0ahHBxreqG
Found-by: SecBuddyF - Tencent KeenLab

Changed files

libavcodec/ffv1dec.c

Change #274651

Category	ffmpeg
Changed by	Jiale Yao <yaojiale02ohnoyoudont@163.com>
Changed at	Mon 13 Jul 2026 03:08:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	705ff11c2b02d2925fca5c2c78c547ccbd901182

Comments

avcodec/libcodec2: reject packet sample counts that overflow int
Fixes: out of array access
Fixes: 2jy_poc_codec2.zip / poc_codec2.raw
Fixes: jOQASNnOm6O7
Found-by: Jiale Yao <yaojiale02@163.com>

Changed files

libavcodec/libcodec2.c

Change #274652

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 03:08:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9c2aabaa34147b7a3812e5857bcb4f9f6e17a5f7

Comments

avcodec/cavsdec: reject invalid start-code bitreader sizes
Fixes: NULL pointer dereference
Fixes: 4jy_poc_cavs.zip / poc_cavs.avi
Fixes: UH66Y8Om9F8X
Found-by: Jiale Yao <yaojiale02@163.com>

Changed files

libavcodec/cavsdec.c

Change #274717

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b2b0429d153cdbc1af0711af98e18d4ae861d4cf

Comments

swscale/vulkan: fix invalid SPIR-V generation for plane-remapped passes
The CLEAR codegen iterated components by value[i].den, while the
constants pass iterates by clear.mask; unmasked components may hold
leftover values with a nonzero denominator, consuming more constant
IDs than were registered and emitting ID 0 into the instruction
stream.

The image handle arrays were also sized by the number of planes an
op touches, but plane_src/plane_dst contain actual frame plane
indices, so a pass writing only e.g. the alpha plane references
handle 3 while only handle 0 was loaded.

Either results in invalid SPIR-V, which crashes RADV inside
spirv_to_nir when creating the shader object.

Changed files

libswscale/vulkan/ops.c

Change #274718

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4b9c4b9cfb56126dccd9dd4e0c78736d1f5d6ff1

Comments

swscale/ops_dispatch: fix use-after-free when adding opaque ops passes
comp points into p, which is freed before comp->backend is read.
Use the copy taken before the free.

Changed files

libswscale/ops_dispatch.c

Change #274719

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1a4250e9249682bae2c1e888f9d751a5f34ee9c0

Comments

avcodec: remove unneeded vulkan_spirv.h includes
Leftover crud from before the rewrite to compile-time SPIR-V.

Changed files

libavcodec/apv_encode_vulkan.c
libavcodec/proresenc_kostya_vulkan.c

Change #274720

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6219d4fdd1fb61b7fff0c713f92dbdd156feaa35

Comments

vf_scale_vulkan: port to compile-time SPIR-V generation
Same functionality as before, but cleaner.

Changed files

libavfilter/vf_scale_vulkan.c
libavfilter/vulkan/Makefile
libavfilter/vulkan/debayer.comp.glsl
libavfilter/vulkan/scale.comp.glsl

Change #274721

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aa6aa4ec3ef89fcaf1c7f73da4caa2116b3cbea0

Comments

swscale: remove GLSL backend
No longer needed. The SPIR-V version is pretty clean and neat.

Changed files

doc/scaler.texi
libswscale/ops.c
libswscale/options.c
libswscale/swscale.h
libswscale/vulkan/Makefile
libswscale/vulkan/ops.c
libswscale/vulkan/ops.h
libswscale/vulkan/vulkan_glslang.c
libswscale/vulkan/vulkan_shaderc.c

Change #274722

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9e9ac0dcd93d0dfe31827237539cb82b4c7e4dc7

Comments

vulkan: remove runtime shader compilation support
No longer needed.

Changed files

libavcodec/apv_encode_vulkan.c
libavcodec/ffv1enc_vulkan.c
libavcodec/proresenc_kostya_vulkan.c
libavcodec/vulkan_apv.c
libavcodec/vulkan_dpx.c
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_prores.c
libavcodec/vulkan_prores_raw.c
libavfilter/vf_avgblur_vulkan.c
libavfilter/vf_blackdetect_vulkan.c
libavfilter/vf_blend_vulkan.c
libavfilter/vf_bwdif_vulkan.c
libavfilter/vf_chromaber_vulkan.c
libavfilter/vf_flip_vulkan.c
libavfilter/vf_gblur_vulkan.c
libavfilter/vf_interlace_vulkan.c
libavfilter/vf_nlmeans_vulkan.c
libavfilter/vf_overlay_vulkan.c
libavfilter/vf_scale_vulkan.c
libavfilter/vf_scdet_vulkan.c
libavfilter/vf_transpose_vulkan.c
libavfilter/vf_v360_vulkan.c
libavfilter/vf_xfade_vulkan.c
libavfilter/vsrc_testsrc_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h
libswscale/vulkan/ops.c

Change #274723

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	92f2e6374dfbe9bf1c8e4d879f8810deafff3710

Comments

configure: remove libshaderc and libglslang support
All of it is now gone.

Changed files

.forgejo/workflows/test.yml
configure
doc/filters.texi
libavfilter/Makefile
libavfilter/vulkan_glslang.c
libavfilter/vulkan_shaderc.c
libavutil/Makefile
libavutil/vulkan_glslang.c
libavutil/vulkan_shaderc.c
libavutil/vulkan_spirv.h

Change #274724

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 11:57:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	90436de5e10ebeac98a3c5e1fb8b2f509e3f96e5

Comments

vf_libplacebo: add missing avstring.h include
Otherwise compilation fails due to vulkan.h no longer including it.

Changed files

libavfilter/vf_libplacebo.c

Change #274795

Category	ffmpeg
Changed by	Cloud-LHY <securityohnoyoudont@clouditera.com>
Changed at	Mon 13 Jul 2026 18:13:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fd3ee52fab34d98a95b787d0b5ff45685766200c

Comments

avcodec/tdsc: unref the reference frame before reallocating on size change
Fixes: out of array access
Fixes: tdsc_poc/ffmpeg-tdsc-linesize-report/poc.avi / gen_poc.py
Fixes: tdsc_resize_jpeg_oob.avi / tdsc-resize-stale-linesize-jpeg-oob-generate-poc.py
Fixes: p9xG4xGf9P7H
Fixes: HQL7a1WgTdHZ
Found-by: Cloud-LHY / Clouditera Security, Z.ai Security, NSFOCUS
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/tdsc.c

Change #274800

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 18:13:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0c29cabc97639ccf2f3d676540c3f4d4a54c423e

Comments

doc/infra.txt: add fairy

Changed files

doc/infra.txt

Change #274803

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 18:13:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	01a9a98f353bcf9ffcdfdcc62533ef063231384d

Comments

doc/developer.texi: update reviews section
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

doc/developer.texi

Change #274806

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 18:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	569674ac8dc00c11eaa96d86870b186cfb076639

Comments

vulkan_encode: free CBS context, fragment and SEI buffers on close
The CBS context, its current access unit fragment, and the SEI
buffers were never freed, leaking the refcounted parameter set
clones stored in the CBS private data on every encoder teardown.

Changed files

libavcodec/vulkan_encode_av1.c
libavcodec/vulkan_encode_h264.c
libavcodec/vulkan_encode_h265.c

Change #274809

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 18:23:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ddfa842088cd86bd48bc4533f10a4b5badcbd02d

Comments

vulkan_encode: destroy image views of queued pictures on uninit
ff_hw_base_encode_close() frees any pictures still in the encode
queue without invoking the encoder's free callback, leaking their
image views when an encoder is closed without fully draining it.
Also uninit the packet buffer pool before the base encode context,
rather than after.

Changed files

libavcodec/vulkan_encode.c

Change #274810

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 13 Jul 2026 18:23:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b672ae39053759557a42f3a54f31f8413a37d871

Comments

vulkan_encode: fix leak and swallowed errors in init_base_units()
The encoded session parameters buffer was never freed, and both
GetEncodedVideoSessionParametersKHR error paths returned err, which
is always 0 at that point, reporting success on failure.

Changed files

libavcodec/vulkan_encode_av1.c
libavcodec/vulkan_encode_h264.c
libavcodec/vulkan_encode_h265.c

Change #274811

Category	ffmpeg
Changed by	Araz Iusubov <Primeadviceohnoyoudont@gmail.com>
Changed at	Mon 13 Jul 2026 18:40:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fc4b5235963441cf29fa3c9f0f6d6ad1e61b8749

Comments

avfilter/vf_vqe_amf: Add AMF Video Quality Enhancer filter

Changed files

Changelog
configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_vqe_amf.c

Change #274812

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 19:14:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0089694c060bf25b628d16ba6078f083f7522bae

Comments

avfilter/vf_lut3d: compute size2 after the 3DLUTSIZE directive
Fixes: out of array access
Fixes: lut3d_poc/poc_3dlut2.dat
Fixes: 6p0ahHBxreqG
Found-by: SecBuddyF - Tencent KeenLab

Changed files

libavfilter/vf_lut3d.c

Change #274813

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 19:14:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	24c322fdb232d0a3f3790d544dcb64e5c2138e79

Comments

avfilter/vf_floodfill: size the point stack for the current frame
Fixes: out of array access
Fixes: 8aj_floodfill_dynamic_size.pgm / 8aj_generate_floodfill_dynamic_size_pgm.py
Fixes: 3MleMXjGZvu3
Found-by: Adrian Junge (vurlo) <adjun37@gmail.com>

Changed files

libavfilter/vf_floodfill.c

Change #274814

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 19:14:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f186c50cf53aec20e9a29059cb22ca3f2d59201c

Comments

avfilter/vf_floodfill: remove unneeded variables
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavfilter/vf_floodfill.c

Change #274815

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 20:16:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cff4a692949a0c4b5ef36c1438abaccc5a1588de

Comments

avfilter: add latticepal filter
Convert packed 24/32-bit RGB/BGR/RGBA/BGRA input to PAL8 using a per-frame palette
whose colors are placed on a face-centered cubic lattice (realized as the
scaled D3/D4 checkerboard lattice), with a user-supplied density controlling
the number of lattice steps spanning one color axis.

Only lattice points actually used by a frame enter its palette; if a frame
needs more than 256 of them, the filter will itearatively drop palettte
entries and reassign affected pixels until 256 color remain

lookup uses the Conway-Sloane rounding algorithm. Supported dithering
modes: none, ordered 8x8 bayer (swscale), Cluster & Void blue noise and
Floyd-Steinberg error diffusion.

Co-Authored-by: Fable-5
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

Changelog
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/version.h
libavfilter/vf_latticepal.c
tests/fate/filter-video.mak
tests/ref/fate/filter-latticepal
tests/ref/fate/filter-latticepal-batched
tests/ref/fate/filter-latticepal-bayer
tests/ref/fate/filter-latticepal-maxcolors
tests/ref/fate/filter-latticepal-refine
tests/ref/fate/filter-latticepal-residual

Change #274816

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 20:22:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7e38b617b32f996beaa371bbf04b39907d7a527

Comments

avfilter/vf_swaprect: size the temp row buffer for the widest plane
Fixes: out of array access
Fixes: 7aj_swaprect_odd17_nv12.nut / 7aj_generate_swaprect_odd17_nv12.py
Fixes: VRAXYvKtmKa8
Found-by: Adrian Junge (vurlo) <adjun37@gmail.com>

Changed files

libavfilter/vf_swaprect.c

Change #274817

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 20:29:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	de771bd52774a52d45b0e2c82e56995a1ef40df7

Comments

avformat/ty: don't let the Series2 AC3 trim underflow the packet size
Fixes: negative-size-param
Fixes: ty-s2-ac3-negative-size-single-file.ffconcat / create_poc.py
Fixes: g0qeE6KvrjZi
Found-by: Adrian Junge (vurlo)

Changed files

libavformat/ty.c

Change #274818

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 20:36:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ba55f425fdd889a7998fce068367e9d3a6c40af4

Comments

avcodec/hevc: Fix integer overflow in WPP entry point range check
Fixes: out of array access
Fixes: poc.hevc / make_poc.py
Fixes: 83WY4xnR87vb
Fixes: dad354f38d (avcodec/hevc: Check max ctb addresses for WPP)
Found-by: VulnForge Security Research Team <haoyuliu@clouditera.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavcodec/hevc/hevcdec.c

Change #274819

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e1ef95c779a3757d21ab970c06aac3e9e1b71974

Comments

avcodec/d3d12va_av1: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_av1.c

Change #274820

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7309ad9fd78d875e60427daa52260834e27ef2ea

Comments

avcodec/d3d12va_av1: check slice bitstream size against available buffer size

Changed files

libavcodec/d3d12va_av1.c

Change #274821

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	19e81f9260e64c43604556cbd4079fadee4ac4f1

Comments

avcodec/d3d12va_h264: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_h264.c

Change #274822

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fea3f13a1cc5395b0ab6db95dd20e7cf6023c311

Comments

avcodec/d3d12va_hevc: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_hevc.c

Change #274823

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	26a9f9b3aeb054901b69518223af0d6f4f991fd9

Comments

avcodec/d3d12va_mpeg2: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_mpeg2.c

Change #274824

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1307db3d3c95223b1d2164844e0542cddbb2a401

Comments

avcodec/d3d12va_vc1: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_vc1.c

Change #274825

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 13 Jul 2026 21:13:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a633df56a455ffb75cd97af784cb1f2d2ec2bca9

Comments

avcodec/d3d12va_vp9: check size of frame bitstream against available buffer size

Changed files

libavcodec/d3d12va_vp9.c

Change #274826

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 13 Jul 2026 21:35:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	597036b692b8c39198fe572aad027eeb4c13da35

Comments

avformat: Add recursion limit
This is a generic recursion limit that is not specific to a demuxer
it does change public API which makes this unsuitable for backporting

Fixes: self_ref.ffconcat

Reported-by: Yuhao Jiang <danisjiang@gmail.com>

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/avformat.c
libavformat/avformat.h
libavformat/options_table.h
libavformat/version.h

Change #274831

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 14 Jul 2026 04:52:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	58d39e99e26d1116f43a67d66ecfe71dba316875

Comments

avformat/concatdec: Check recursion depth
This variant is easy backportable but is concatdec specific

Fixes: self_ref.ffconcat
Reported-by: Yuhao Jiang <danisjiang@gmail.com>
Rediscovered-by: Anxious Koisi
Fixes: RWHkT0ASFMNI
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/concatdec.c

Change #274844

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 14 Jul 2026 10:01:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f769e433a3486252143e8d912c45fe44a47d8c04

Comments

Revert "tests/checkasm/vp9dsp: Allow to run only a subset of tests"
This reverts commit 7bc35b8426cee8114b5d807b4d0e37fefe9995fa.
The aim of this patch can now be achieved via checkasm's
--function pattern matching.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/checkasm.c
tests/checkasm/checkasm.h
tests/checkasm/vp9dsp.c

Change #274845

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 14 Jul 2026 10:01:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8bea614d987e481c5e912588c50dbdc9920f87f7

Comments

avcodec/x86/sbcdsp: Avoid shift
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/sbcdsp.asm

Change #274859

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2dbcec5222775b9afd82c6379f20da5ebe91e728

Comments

swscale/ops_dispatch: remove outdated comment
Missed by 228ef8d97b.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_dispatch.h

Change #274861

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	670324675fed1ea22f98e5ad07d5fcf4375e0187

Comments

swscale/uops: add extra convenience metadata for the backends
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.c
libswscale/uops.h

Change #274862

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	17a9a59333469cf98dc7459f47968b160422dca3

Comments

swscale/x86/ops: avoid direct reliance on `ops` metadata
Instead, use the new metadata introduced in the previous commit. The
motivating reason here is a desire to stop taking a direct SwsOpList from
the backends altogether.

In theory, we could still recover this information from the `uops` list
itself, but that would require far more boilerplate than simply having
uops.c generate it using existing SwsOpsList helpers.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c

Change #274863

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	33d183a4735f170d91456e0ce59ebcaf4b4a9480

Comments

swscale/ops_dispatch: add SwsOpBackend.compile_uops()
This can be used as an alternative to the existing compile() backend, that
takes SwsUOpList instead of SwsOpList.

Down the line, I plan on also moving the uops translation to the dispatch
layer itself, but this requires some further changes which have not yet
been merged, so this commit represents a stopgap solution.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_dispatch.h
libswscale/uops_backend.c

Change #274864

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bb553f91605d963173312474210f0a9444479c7f

Comments

swscale/x86/ops: implement compile_uops()
This also splits off the mmsize determination to a separate helper,
to make the code a bit more reusable.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c

Change #274865

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b3a998699c53aa3be37919becb6496dddb6b0a7b

Comments

swscale/x86/ops: add micro-op dump to compile_uops_x86() as well
This was only done in compile_uops_c(). Add it here as well for consistency.
Long-term I want to move this code to ops_dispatch.c, but that requires
moving the packed shuffle solver there as well, which is still WIP.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c

Change #274866

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	44e1b60c6acf0511222f4e0892f3545696bb7769

Comments

swscale/uops: consolidate SWS_UOP_{PERMUTE,COPY,MOVE}
Rather than having two uops for the C reference code and a separate
uop for the x86 backend, we can just implement the C reference
using the same register-register assignment logic.

This results in only slightly worse codegen, since the compiler is
smart enough to eliminate the unnecessary copies.

The main downside is the fact that we need to allocate extra stack space
to back-fill the remaining (unused) pointers. In theory, we could do this
more efficiently by re-using unused *input* registers directly, but that
would require a lot of reliance on compile-time constant folding teasing
apart the entire register index tracking loop and duplicate check, so this
is a useful middle ground.

On platforms like x86, we still don't care about the PERMUTE vs
COPY distinction, but it's also harmless (2-3 extra lines to
instantiate the underlying MOVE macro on both uop variants).

Finally, we need to add a mask to the move uops to keep track of which
outputs are actually needed (especially for checkasm). Though it does result
in the rather unfortunate downside that it bloats the names of the uop
names.

The op lists change to reflect the new representation, e.g.:

 rgb24 16x16 -> bgr24 16x16:
   [ u8 +++X] SWS_OP_READ         : 3 elem(s) packed >> 0
     min: {0 0 0 _}, max: {255 255 255 _}
   [ u8 +++X] SWS_OP_SWIZZLE      : 2103
     min: {0 0 0 _}, max: {255 255 255 _}
   [ u8 XXXX] SWS_OP_WRITE        : 3 elem(s) packed >> 0
     ('X' unused, 'z' byteswapped, '=' copied, '$' const, '+' integer, '0' zero)
  translated micro-ops:
     u8_read_packed_xyz
-    u8_permute_zyxw
+    u8_permute_xyzw_txz_xzt
     u8_write_packed_xyz

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.c
libswscale/uops.h
libswscale/uops_list.h
libswscale/uops_macros.h
libswscale/uops_macros_gen.c
libswscale/uops_tmpl.c
libswscale/x86/ops.c
libswscale/x86/ops_int.asm
libswscale/x86/uops_macros.asm.h
tests/checkasm/sw_ops.c
tests/ref/fate/sws-ops-list

Change #274867

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	909ca1d7ce77f9a188c979902acc223096b5a5ec

Comments

swscale/x86/ops: add missing non-FMA SWS_UOP_LINEAR fallback
Needed for platforms without SSE4.1 support, as well as checkasm.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c
libswscale/x86/ops_float.asm
libswscale/x86/uops_macros.asm.h

Change #274868

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e65dab2b5919ede3a4f308d723e46a6f65643ca8

Comments

swscale/tests/sws_ops: rewrite to test uops directly
Instead of awkwardly reverse-engineering each uop to reflect
it back to the corresponding SwsOp, this leverages the new
compile_uops() API to directly test the uop in question.

This is not only vastly simpler but also fixes several major
shortcomings with the previous design, such as always having
to keep the reflection code in sync for any newly added uop,
or not testing uops that get translated to a packed shuffle
fast path.

The major downside is that arch-specific uops can now no longer
be tested against the C reference unless we specifically add
a C reference function for each relevant uop; and also, that the
packed shuffle fast path is currently no longer tested by checkasm.
(But this will be fixed by my upcoming pshufb uop series)

Another downside is that this effectively disables testing for e.g.
aarch64, since it does not yet use the uops architecture.

$ time checkasm --test=sw_ops --repeat=10 # 1.85s -> 1.39s

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.h
tests/checkasm/sw_ops.c

Change #274869

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 14 Jul 2026 14:24:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	150f7d15df726bdcf49a7c5ef9d8d67c46ed3d41

Comments

swscale/uops_macros_gen: drop checkasm planar write hack
No longer needed since the new uops-based rewrite.
Here's hoping we won't need to revert it if these ops come up again
in the future.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops_macros.h
libswscale/uops_macros_gen.c

Change #274876

Category	ffmpeg
Changed by	lck <luckyskohnoyoudont@foxmail.com>
Changed at	Tue 14 Jul 2026 15:54:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9287fc3bc90e6c2cbdd5a5955eb46153501695fe

Comments

lavc/mediacodecdec: rename MediaCodecH264DecContext to MediaCodecContext
The context structure was named specifically for H.264, but it's actually
used for multiple codecs (H.264, H.265, MPEG-4, etc.). Renaming it to
MediaCodecContext better reflects its actual usage and makes the code
more consistent.

Changed files

libavcodec/mediacodecdec.c

Change #274877

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Tue 14 Jul 2026 16:26:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	312c8309162eab790d6bed069db7eb1a52263e1e

Comments

avfilter/dnn: auto-initialize nireq for Torch backend
Initialize ctx->nireq with av_cpu_count() / 2 + 1 when
unset, matching the TensorFlow and OpenVINO backends.

Create ctx->nireq THRequestItems in a loop instead of
hardcoding a single request, enabling concurrent async
inference requests for improved throughput.

Signed-off-by: Raja Rathour <imraja729@gmail.com>

Changed files

libavfilter/dnn/dnn_backend_torch.cpp

Change #274893

Category	ffmpeg
Changed by	Vignesh Venkat <vigneshvohnoyoudont@google.com>
Changed at	Tue 14 Jul 2026 20:50:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a302c9ae44ec07cbe12417f14c1852868f903f60

Comments

avformat/mov: Add itut-t35 data codec support
The specification for the T35MetaDataSampleEntry ('it35') box is
available in the "Draft text of ISO/IEC 14496-12 10th edition"
which can be found on the MPEG website:
https://www.mpeg.org/standards/MPEG-4/12/

Support the t35 track in mov and export the `t35_identifier`
as extradata and human readable sample description as the stream
description.

Signed-off-by: Vignesh Venkat <vigneshv@google.com>

Changed files

libavformat/isom.c
libavformat/mov.c

Change #274894

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Tue 14 Jul 2026 20:50:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8a3e9147b697f093ef59675906ceb8f64cd78d19

Comments

avformat/mov: parse cdsc track references
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/mov.c

Change #274895

Category	ffmpeg
Changed by	Vignesh Venkat <vigneshvohnoyoudont@google.com>
Changed at	Tue 14 Jul 2026 20:50:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5002b55ce79fda51e1d329a28d112495aab9f8cb

Comments

fate/mov: Add fate test for t35 cdsc track
Signed-off-by: Vignesh Venkat <vigneshv@google.com>

Changed files

tests/fate/mov.mak
tests/ref/fate/mov-t35-cdsc-track

Change #274896

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c7546ce0ea6e451c33bf16c8ea1a341b917652a5

Comments

avutil/pixfmt: add AV_PIX_FMT_CUARRAY for opaque CUDA block-linear surfaces
Add a new hardware pixel format representing opaque block-linear CUDA
arrays (CUarray). This format is used by NVDEC when decoding into
opaque surfaces registered with cuvidRegisterDecodeSurfaces, and can
be consumed directly by NVENC as CUDA array input without requiring
a pitch-linear copy.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

doc/APIchanges
libavutil/pixdesc.c
libavutil/pixfmt.h

Change #274897

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ef3dcf4ea6139f2b8cad29e67b48a39f094fa3bc

Comments

avutil/hwcontext_cuda: add CUarray surface pool support
Extend the internal CUDA frames context (CUDAFramesContext) to manage
block-linear CUarray surfaces alongside the existing pitched-linear
allocations.

When the frames context format is AV_PIX_FMT_CUARRAY, the caller
(e.g. NVDEC) can fill cuarray_desc and cuarray_num_surfaces before
calling av_hwframe_ctx_init(), which then allocates the CUarray
surfaces.

Also add cuda_transfer_data_from_cuarray() to support hwdownload
from CUarray frames to host memory via cuMemcpy2DAsync.

Co-authored-by: Diego de Souza <ddesouza@nvidia.com>
Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

doc/APIchanges
libavutil/hwcontext_cuda.c
libavutil/hwcontext_cuda.h
libavutil/hwcontext_cuda_internal.h
libavutil/version.h

Change #274898

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ca13ad0c2873f682f99ae52cfe7c8b611044d63b

Comments

avcodec/nvdec: add opaque CUARRAY output and zero-copy decode support
Add support for NVDEC opaque output using block-linear CUarray surfaces
registered with cuvidRegisterDecodeSurfaces (Video Codec SDK 13.1+).

Two output modes are implemented:

  - Zero-copy (unsafe_output): decode surfaces are exposed directly
    to the consumer (e.g. NVENC). The surface is locked via
    surface_in_use[] and released when the downstream AVBufferRef
    is freed.

  - Safe copy (!unsafe_output): the frame is set up pointing to the
    decode CUarray, then av_frame_make_writable() copies it into a
    new CUarray from the output hwframes pool via cuda_transfer_data().
    This decouples the decode and encode surface lifetimes.

All opaque-specific code is guarded by NVDEC_HAVE_OPAQUE_OUTPUT_SUPPORT,
with a AVERROR(ENOSYS) fallback when building against older SDK versions.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavcodec/nvdec.c
libavcodec/nvdec.h

Change #274899

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	acbaf420aade95d5965260ede91945b194093b32

Comments

avcodec/nvdec: register CUARRAY hwaccel variants for all supported codecs
Add CUARRAY hwaccel structs (ff_<codec>_nvdec_cuarray_hwaccel) and the
corresponding AV_PIX_FMT_CUARRAY entries in each decoder's pixel format
negotiation list. This allows decoders to advertise CUARRAY as a
supported hardware output format alongside AV_PIX_FMT_CUDA.

Codecs: H.264, HEVC, AV1, MPEG-1, MPEG-2, MPEG-4, VC-1, WMV3, VP8, VP9.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

configure
libavcodec/Makefile
libavcodec/av1dec.c
libavcodec/h263dec.c
libavcodec/h264_slice.c
libavcodec/h264dec.c
libavcodec/hevc/hevcdec.c
libavcodec/hwaccels.h
libavcodec/hwconfig.h
libavcodec/mpeg12dec.c
libavcodec/mpeg4videodec.c
libavcodec/nvdec_av1.c
libavcodec/nvdec_h264.c
libavcodec/nvdec_hevc.c
libavcodec/nvdec_mpeg12.c
libavcodec/nvdec_mpeg4.c
libavcodec/nvdec_vc1.c
libavcodec/nvdec_vp8.c
libavcodec/nvdec_vp9.c
libavcodec/vc1dec.c
libavcodec/vp8.c
libavcodec/vp9.c

Change #274900

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0c0580e5bf97fc5792af0b635ce5a50a501526ee

Comments

avcodec/cuviddec: add CUARRAY opaque output and zero-copy support
Implement opaque block-linear output for the CUVID decoder, mirroring
the NVDEC hwaccel path but within the standalone CUVID demuxer/decoder.

Key additions:
  - output_format=cuarray option and zero_copy toggle
  - CUarray surface creation and cuvidRegisterDecodeSurfaces
  - Zero-copy path: frames reference decode CUarray surfaces directly
  - Copy path: output surfaces allocated from the hwcontext_cuda pool
  - Deferred decoder cleanup (CuvidDecoderCleanup) to handle the case
    where surfaces outlive the decoder due to encoder buffering
  - 444 chroma sw_pix_fmt remapping to semi-planar (NV24, P410, P412,
    P416) to match the native CUarray layout
  - Per-plane linesize derived from CUarray plane descriptors via
    cuArrayGetPlane / cuArray3DGetDescriptor
  - Crop, resize, and deinterlace forced off (with warnings) for
    cuarray output, as the decode surfaces are used directly

All opaque-specific code is guarded by NVDEC_HAVE_OPAQUE_OUTPUT_SUPPORT.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavcodec/cuviddec.c

Change #274901

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a127ad2da64e7337efa850a5186ead9ee70b546d

Comments

avcodec/nvenc: add CUARRAY (block-linear) input support
Allow NVENC to accept AV_PIX_FMT_CUARRAY frames by registering them
as NV_ENC_INPUT_RESOURCE_TYPE_CUDAARRAY. This enables zero-copy
transcode when paired with NVDEC CUARRAY output, avoiding the
pitched-linear intermediate copy. Also accept semi-planar 444
formats (NV24, P410, P412, P416) and map them to the corresponding
NVENC buffer formats when registered as CUarray input resources.

Guarded by NVENC_HAVE_CUDA_ARRAY_INPUT_SUPPORT (SDK 13.1+).

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavcodec/nvenc.c
libavcodec/nvenc.h
libavcodec/version.h

Change #274902

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1eb47126215761abf89ff9db8fd9164bd596d206

Comments

fftools/ffmpeg_dec: improve hwaccel format negotiation for CUARRAY
Adjust get_format() to prefer the exact hwaccel_output_format requested
by the user when multiple hardware formats are available (e.g. both
AV_PIX_FMT_CUDA and AV_PIX_FMT_CUARRAY). Previously, the loop would
break on the first matching device type, ignoring the user's preferred
output format.

When the preferred format is not available, the code now falls back
gracefully and emits a warning.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

fftools/ffmpeg_dec.c

Change #274903

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Tue 14 Jul 2026 21:45:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	07020f5ce8dbac1bbde8cb9fa1cad1accb9a5042

Comments

avfilter/hwupload_cuda: add CUARRAY output and semi-planar 444 format support
Add an output_format option (cuda | cuarray) so the filter can produce
either pitched-linear (AV_PIX_FMT_CUDA, default) or block-linear
(AV_PIX_FMT_CUARRAY) device frames.  This enables host-to-device
upload directly into CUarray surfaces for the CUARRAY transcode
pipeline.

Also extend the accepted input pixel format list with semi-planar 444
formats (NV24, P410, P416) and the planar MSB variants (YUV444P10MSB,
YUV444P12MSB).  Semi-planar 444 formats are rejected with a clear
error when output_format=cuda, as they are only supported by the
CUarray allocation path.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/version.h
libavfilter/vf_hwupload_cuda.c

Change #274917

Category	ffmpeg
Changed by	zhanghongyuan <zhanghongyuanohnoyoudont@uniontech.com>
Changed at	Tue 14 Jul 2026 23:25:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ac91f1c0f7001cd62528a2130a41379fd2b6d176

Comments

avdevice/pulse_audio_dec: validate sample spec before use
ff_codec_id_to_pulse_format() returns PA_SAMPLE_INVALID for
unsupported codecs (e.g. AAC, FLAC, Opus). Without validation,
the invalid format propagates to pa_frame_size() and other PA
calls, causing assertion failures or division-by-zero.

Add pa_sample_spec_valid() check, matching the existing guard
in pulse_audio_enc.c.

Signed-off-by: zhanghongyuan <zhanghongyuan@uniontech.com>

Changed files

libavdevice/pulse_audio_dec.c

Change #274925

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 15 Jul 2026 00:48:26
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	96bce2956546468bc1019fa772eae8820fe48c68

Comments

avformat/mov: fix error log for cdsc track references
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/mov.c

Change #274933

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 15 Jul 2026 02:17:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1ad50bb62ced9c3a8479ebc57dc3c78d89c25eda

Comments

configure: gate CUARRAY support on ffnvcodec capabilities
The CUARRAY implementation unconditionally references CUDA array
formats and dynamic-loader entries added in nv-codec-headers 13.1.
This breaks builds with older ffnvcodec versions that configure
still accepts.

Probe for the complete CUARRAY header support and use the result
to disable only CUARRAY hwaccels and code paths. Ordinary CUDA,
NVDEC, NVENC, and hwupload_cuda remain available with older headers.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

configure

Change #274934

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Wed 15 Jul 2026 02:24:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d7fe91e9c38fa968034d33d15c3b6c79d0f9c405

Comments

avutil/hwcontext_cuda: fix build with old cuda headers

Changed files

libavutil/hwcontext_cuda.c

Change #274935

Category	ffmpeg
Changed by	Thomas Guilbert <tguilbertohnoyoudont@chromium.org>
Changed at	Wed 15 Jul 2026 03:36:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	005e963c85689f4714ea9948a6be4739fc68ebc9

Comments

Guard against int overflow when discarding samples
In `discard_samples()`, the rescaling operation can evaluate to
AV_NOPTS_VALUE if the rescaling operation overflows. This commit
prevents adjusting timestamp and durations by the invalid value.

Changed files

libavcodec/decode.c

Change #274936

Category	ffmpeg
Changed by	Mirkó Visontai <mirkovohnoyoudont@google.com>
Changed at	Wed 15 Jul 2026 05:47:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1588bce21b6d5d9064fd9576c038fd8d6ec8beac

Comments

avdevice/android_camera: fix OOB read in metadata parsing
ACAMERA_SCALER_AVAILABLE_STREAM_CONFIGURATIONS metadata is an
int32[n*4] array (one 4-tuple per stream config: format, width,
height, input/output flag). ACameraMetadata_const_entry.count is the
total number of int32_t elements, not the number of tuples. The loop
bound must be count/4 to avoid iterating past the end of the array.

Similarly, ACAMERA_CONTROL_AE_AVAILABLE_TARGET_FPS_RANGES is an
int32[n*2] array (min/max pairs). The loop bound must be count/2.

Without this fix, both loops over-iterate and read heap memory
beyond the metadata array bounds.

Signed-off-by: Mirko Visontai <mirkov@google.com>

Changed files

libavdevice/android_camera.c

Change #274955

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 15 Jul 2026 12:46:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a5e6c0175a7245fc1a9f4639da2810156128aa8b

Comments

swscale/uops: mark unneeded rows as zero in translate_linear_op()

Changed files

libswscale/uops.c
libswscale/uops_macros.h

Change #275023

Category	ffmpeg
Changed by	Hans-Kristian Arntzen <postohnoyoudont@arntzen-software.no>
Changed at	Thu 16 Jul 2026 09:19:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0661ef6bb3e50f502e79d3c5fea1c83e860f1d09

Comments

avcodec/h264: Ensure that private HW data is freed early for pictures.
Vulkan video decode holds a reference to a VkSemaphore that is waited on
in ff_vk_decode_free_frame.
The bug manifests when a H264 stream changes resolution (or similar)
where the stream needs to be reset in-flight.
In this case, the wait_semaphores call ends up waiting on a destroyed
semaphore, leading to segfaults in drivers. FFVulkanDecodePicture does
not hold an owning reference to the semaphore, so destruction order
matters in this case.

Vulkan validation flags the error as waiting on an invalid semaphore.

Changed files

libavcodec/h264_picture.c

Change #275024

Category	ffmpeg
Changed by	Hans-Kristian Arntzen <postohnoyoudont@arntzen-software.no>
Changed at	Thu 16 Jul 2026 09:19:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d34492955221cc2633a9fe26145e099a702b9d90

Comments

avcodec/hevc: Ensure that private HW data is freed early for pictures.
Lynne suggested that HEVC would have a similar bug.
I couldn't reproduce a crash or VVL error in this area.

Comment copied from the h264 fix:

Vulkan video decode holds a reference to a VkSemaphore that is waited on
in ff_vk_decode_free_frame.
The bug manifests when a H264 stream changes resolution (or similar)
where the stream needs to be reset in-flight.
In this case, the wait_semaphores call ends up waiting on a destroyed
semaphore, leading to segfaults in drivers. FFVulkanDecodePicture does
not hold an owning reference to the semaphore, so destruction order
matters in this case.

Vulkan validation flags the error as waiting on an invalid semaphore.

Changed files

libavcodec/hevc/refs.c

Change #275025

Category	ffmpeg
Changed by	Hans-Kristian Arntzen <postohnoyoudont@arntzen-software.no>
Changed at	Thu 16 Jul 2026 09:19:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	977db8c2cbfc835401f972dc15161fc01155b189

Comments

avcodec/av1: Ensure that private HW data is freed early for pictures.
Comment copied from the h264 fix:

Vulkan video decode holds a reference to a VkSemaphore that is waited on
in ff_vk_decode_free_frame.
The bug manifests when a H264 stream changes resolution (or similar)
where the stream needs to be reset in-flight.
In this case, the wait_semaphores call ends up waiting on a destroyed
semaphore, leading to segfaults in drivers. FFVulkanDecodePicture does
not hold an owning reference to the semaphore, so destruction order
matters in this case.

Vulkan validation flags the error as waiting on an invalid semaphore.

Changed files

libavcodec/av1dec.c

Change #275026

Category	ffmpeg
Changed by	Hans-Kristian Arntzen <postohnoyoudont@arntzen-software.no>
Changed at	Thu 16 Jul 2026 09:19:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0b0870dc8cfe099f51a4836c3ff5b34d77e6b5cb

Comments

avcodec/vp9: Ensured that private HW data is freed early for pictures.
Comment copied from the h264 fix:

Vulkan video decode holds a reference to a VkSemaphore that is waited on
in ff_vk_decode_free_frame.
The bug manifests when a H264 stream changes resolution (or similar)
where the stream needs to be reset in-flight.
In this case, the wait_semaphores call ends up waiting on a destroyed
semaphore, leading to segfaults in drivers. FFVulkanDecodePicture does
not hold an owning reference to the semaphore, so destruction order
matters in this case.

Vulkan validation flags the error as waiting on an invalid semaphore.

Changed files

libavcodec/vp9.c

Change #275029

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:28:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b801236856a1bed30052e4ca008a3cc88976ddb0

Comments

avutil/bprint: fix pointer signedness in return value for av_bprint_get_buffer()
This is usually silenced by -Wno-pointer-sign, but a good fix nonetheless.

Changed files

libavutil/bprint.c

Change #275030

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:28:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	61939878a48dcd04ab0431cd440397f1434e7d14

Comments

swscale/aarch64/rasm: use AVBPrint instead of custom string functions
AVBPrint can be #include'd directly into aarch64/ops_asmgen so it can
keep working standalone. This move will help implementing runtime JIT
support for aarch64.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/rasm.h
libswscale/aarch64/rasm_print.c

Change #275031

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:30:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5e7983d7fc34985a8750a66ad5b727ae803b3d6b

Comments

swscale/aarch64/ops: fix SWS_OP_CLEAR parameter extraction
The mask at op->clear.mask already contains only the components which
are necessary. The values in op->clear.value may contain stale values
for unneeded components.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_impl_conv.c

Change #275032

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:30:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3d5ab8093b1c163f8362a6ab57a818e46035de97

Comments

swscale/aarch64/ops: perform pass splitting while generating ops_entries.c
There was a mismatch between the ops generated by sws_ops_aarch64 and
the aarch64 backend, leading to missed optimizations.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_entries.c
libswscale/tests/sws_ops_aarch64.c

Change #275033

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:30:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	136ecf554d69f1c65edc354391b818e0dbda9e93

Comments

swscale/aarch64/ops: remove struct SwsAArch64BackendContext and aarch64_optimize() stubs
The purpose of this struct and function was to perform aarch64-specific
optimizations, but the backend will eventually be converted to the new
uops architecture, which has the ability to perform arch-specific
optimizations.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops.c

Change #275034

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:30:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	91f4552d9f633ad341659208ee4752ffa0f2cdb5

Comments

swscale/aarch64/ops: move SwsOpExec/SwsOpImpl offset value checks to ops.c
This prevents build warnings for ops_asmgen on non-aarch64 hosts.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops.c
libswscale/aarch64/ops_impl.h

Change #275035

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:30:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a09558e5cca19177efbdd9105f9d5ba9a6d50b5

Comments

swscale/uops: move some simple functions to header file
So that they may be used without uops.c.

This will be useful for aarch64/ops_asmgen.c which is built for the
host system, and not the target system.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/ops.c
libswscale/uops.h

Change #275036

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:32:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4d9be5f57c0e3153c1741bcfc54a13d0d3943f43

Comments

swscale/jit: move jit-related code to its own file
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/Makefile
libswscale/jit.c
libswscale/jit.h
libswscale/utils.c

Change #275037

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:32:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	771999278027b1930b0f2948ec8a9f24659446e6

Comments

swscale/jit: implement ff_sws_jit_protect() for win32
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/jit.c

Change #275040

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Thu 16 Jul 2026 10:46:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1dae8f89b24ff23565be29b689c061a2d499bbeb

Comments

swscale/uops: generate more specific masks for unpack when translating
Unused components weren't being taken into consideration.

Changed files

libswscale/uops.c
libswscale/uops_macros.h
tests/ref/fate/sws-ops-list

Change #275041

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Thu 16 Jul 2026 10:54:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	86c03a14ee81536f19e7cecf9d9dc028a3bba896

Comments

avcodec/d3d12va_encode_av1: add user-configurable tile support
Add a "tiles" option (WxH format) to the D3D12VA AV1 encoder. When unset, the encoder falls
back to the minimum number of tile columns/rows that satisfies AV1 spec constraints.

Multi-tile support is also a prerequisite for 8K encoding.

Example usage:
  # default (minimum tiles, auto-computed):
  ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4 -c:v av1_d3d12va output.mp4
  # 2x2 tile  (4 tiles):
  ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4 -c:v av1_d3d12va -tiles 2x2 output.mp4
  # 4x4 tile - required for 8K (7680x4320):
  ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input_8k.mp4 -c:v av1_d3d12va -tiles 4x4 output_8k.mp4

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavcodec/d3d12va_encode.c
libavcodec/d3d12va_encode.h
libavcodec/d3d12va_encode_av1.c

Change #275043

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 16 Jul 2026 11:17:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ceabc9b306f5385d92efdd9cd18d210deb3055b3

Comments

avcodec/x86/vp9lpf: Avoid load
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9lpf.asm

Change #275068

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Thu 16 Jul 2026 15:02:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7f0b6476b6ef2d07d163a7d3229f8c9e250112b5

Comments

avcodec/nvdec: pass hw format explicitly to frame params
avcodec_get_hw_frames_parameters() may invoke frame_params from an
application's get_format() callback. ff_get_format() clears the current
hwaccel before calling get_format(), so avctx->hwaccel is NULL at that
point.

The CUARRAY frame-parameter path dereferenced avctx->hwaccel to
distinguish CUDA from CUARRAY, causing callers such as mpv to crash.
Add per-variant callbacks that pass the registered pixel format
explicitly instead.

Fixes issue #23826.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavcodec/nvdec.c
libavcodec/nvdec.h
libavcodec/nvdec_av1.c
libavcodec/nvdec_h264.c
libavcodec/nvdec_hevc.c
libavcodec/nvdec_mjpeg.c
libavcodec/nvdec_mpeg12.c
libavcodec/nvdec_mpeg4.c
libavcodec/nvdec_vc1.c
libavcodec/nvdec_vp8.c
libavcodec/nvdec_vp9.c

Change #275073

Category	ffmpeg
Changed by	Vignesh Venkat <vigneshvohnoyoudont@google.com>
Changed at	Thu 16 Jul 2026 21:35:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e0947fcb655c8663a7fc3351254f6378b9a8305f

Comments

avcodec/hevcdec: Support SMPTE-2094-50 metadata
Handle SMPTE-2094-50 metadata that appears as a SEI message.

Signed-off-by: Vignesh Venkat <vigneshv@google.com>

Changed files

libavcodec/hevc/hevcdec.c

Change #275074

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Thu 16 Jul 2026 21:49:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8e081d3c31c84429a117e14b7aa947b7c4bf6e90

Comments

checkasm: sw_ops: Fix building with older versions of Clang
Older versions of Clang (17 and older) errored out on the previous
form like this:

    ./src/tests/checkasm/sw_ops.c:210:9: error: expected expression
      210 |         static_assert(sizeof(src0[1]) >= sizeof(uint32_t[256]), "palette plane too small");
          |         ^
    /usr/include/assert.h:158:24: note: expanded from macro 'static_assert'
      158 | # define static_assert _Static_assert
          |                        ^

While newer versions of Clang merely warned about it:

    ./src/tests/checkasm/sw_ops.c:210:9: warning: label followed by a declaration is a C23 extension [-Wc23-extensions]
      210 |         static_assert(sizeof(src0[1]) >= sizeof(uint32_t[256]), "palette plane too small");
          |         ^
    /usr/include/assert.h:158:24: note: expanded from macro 'static_assert'
      158 | # define static_assert _Static_assert
          |                        ^

Changed files

tests/checkasm/sw_ops.c

Change #275077

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Thu 16 Jul 2026 23:39:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7f6b35d6c804ed8d1bc517c33044f075d72852ba

Comments

arm: Avoid warnings about redefining HWCAP constants
Before ced4a6ebc9e7cd92d0ca9b9fb8f9d1013d23cbfa, we used constants
in the form HWCAP_NEON. In that commit, they were renamed to
HWCAP_ARM_NEON, to match the pattern used for aarch64 and PPC.

On PPC our naming is entirely custom to us, upstream headers
(from e.g. glibc) use an entirely different naming.

On aarch64, we use names similar to the system headers, where they
are named e.g. HWCAP_CRC32, but we add an extra prefix, naming them
HWCAP_AARCH64_CRC32, to avoid any potential conflict and redefinition
issues. (In practice, this isn't needed - both glibc, musl and
BSDs provide the aarch64 HWCAP_ defines with the same names and
same literal expansion.)

However, on ARM, the landscape of the HWCAP defines is much less
consistent. On glibc, <bits/hwcap.h> (which gets included by
<sys/auxv.h>) provides most constants with an HWCAP_ARM_ prefix.
(The couple newest constants lack that prefix though.)
The kernel headers, in <asm/hwcap.h>) define them with just a
HWCAP_ prefix, i.e. HWCAP_ARM_NEON vs HWCAP_NEON.

For the actual expansion, <asm/hwcap.h> defines HWCAP_NEON as
(1 << 12), while glibc <bits/hwcap.h> defines HWCAP_ARM_NEON
as 4096. Musl's <bits/hwcap.h> defines both HWCAP_NEON and
HWCAP_ARM_NEON, as (1 << 12).

The BSDs define HWCAP_NEON in <elf.h>, defined as 0x00001000.

Thus, all system headers define the constant with differing
expansions. And they differ in which name they expose for the
constants. But we provide our definitions in any case.

Due to all this, wrap our definitions in #ifndefs, to avoid warnings
about redefining the constants on glibc.

Changed files

libavutil/arm/cpu.c

Change #275081

Category	ffmpeg
Changed by	deng.zewen <deng.zewenohnoyoudont@zte.com.cn>
Changed at	Fri 17 Jul 2026 04:44:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8d394252d80d045bd5ad473f25e85dc55556105d

Comments

avcodec/riscv/hevcdsp_init: decouple add_residual from RVB guard

Changed files

libavcodec/riscv/hevcdsp_init.c

Change #275110

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	89cc079b302752dd2b870752cbe746be7703886e

Comments

swscale/ops_optimizer: add ff_sws_uop_list_optimize() placeholder
This will be populated with logic in the near future. I decided to
also include the skeleton of the loop here, guarded by #if 0 to
avoid triggering unused variable warnings until we actually use it.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_optimizer.c
libswscale/uops.c
libswscale/uops.h

Change #275111

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	755be6f8422fa59c7372b161783e16cdc3f39cbe

Comments

swscale/uops: add ff_sws_uop_list_remove_at()
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.c
libswscale/uops.h

Change #275112

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5c864087246655cf102e614c2096055f9afc78a2

Comments

swscale/uops: add SWS_UOP_RW_SHUFFLE
This directly encodes the "pshufb" loop style processing seen in the x86
backend (and, presumably, other backends down the line). The major
advantage of encoding this as a separate UOP is that it allows the macro
generator to exhaustively enumerate all needed variants of the read/write
chunk sizes, which so-far have been hand-maintained through trial and error.

I decided to also go ahead and pre-emptively generalize the clear
value to allow us to encode RGBA clears in addition to RGB0 clears,
which makes this UOP useful for e.g. Gray -> RGBA32 as well; something
that the current pshufb loop did not handle.

This commit merely adds the UOP defition; it does not yet port
the packed shuffle solver or x86 implementation.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.c
libswscale/uops.h
libswscale/uops_list.h
libswscale/uops_macros.h
libswscale/uops_macros_gen.c

Change #275113

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	535ee06f6d050a1957f17fc255a6b487a9c77133

Comments

swscale/uops: add SWS_UOP_FLAG_PSHUFB
Exposes the SWS_UOP_RW_SHUFFLE capability. While this translates
to the exact `pshufb` instruction on x86, we don't expose a separate
capability for e.g. `vpermb` on AVX-512, since the underlying semantics
are essentially equivalent.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops.h
libswscale/uops_macros_gen.c
libswscale/x86/ops.c

Change #275114

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	784f4c10cd1f05f23f197af2456a4f461ca24167

Comments

swscale/uops: add ff_sws_shuffle_mask() helper
This takes an arbitrary shuffle mask and expands it out to the needed number
of bytes by repeating pixels.

I decided to have the implementation live in ops_optimizer.c to locate
it closer to the code where these shuffle masks are generated, to aid
in understanding, and because the optimizer will itself use it (in the
following commit).

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_optimizer.c
libswscale/uops.h

Change #275115

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b97d65aae23e5184fdaf699e427feba96d796e4b

Comments

swscale/uops: rewrite ff_sws_solve_shuffle() in terms of uops
This allows the uop optimizer stage to directly generate RW_SHUFFLE
uops, which will replace the original ff_sws_solve_shuffle() over the next
couple of commits.

This way of solving it has a number of additional upsides, including but not
limited to:

- The ability to move certain low-level optimizations (like expanding
  conversions) out of the higher-level abstraction and have them arise
  purely during uops translation / optimization

- The ability to stop passing SwsOpList to backends at all

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_optimizer.c
libswscale/uops_macros.h

Change #275116

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8fdafc8198c7aac078b4a9c550ac50e32ae3ec2b

Comments

swscale/x86/ops: rewrite packed_shuffle() in terms of uops macro
This is mostly the same, with the added complication of now needing to handle
the case of clearing to a nonzero value, which we do using `pblendb` on SSE4,
and respectively `vpblendb` / `vpblendmb` on AVX-2 and AVX-512.

Since `pblendb` hard-codes XMM0 implicitly, we have to flip the order of `m0`
and `m1` in the rest of the code as well; though that's not a huge deal.

The bigger complication comes from the fact that the macro gives us the size
for a size 16 mask, which may need to be rounded up for the lane-aligned
AVX-2 / AVX-512 paths, so we need to recreate this logic in the NASM macro
itself. Fortunately, it's fairly straightforward.

The big upside is that we now gain a fast path for e.g. rgb24 -> rgba, which
is arguably more common than rgb24 -> rgb0:

  rgb24 1920x1080 -> rgba 1920x1080, speedup=7.106x faster

It's worth pointing out that, because checkasm sometimes generates weird ops
that don't occur in real pixel formats, it's not a guarantee that we will
actually implement all paths that the optimizer spits out.

This should be improved by a future checkasm refactor; but for now, we need
to return ENOTSUP in the case that we encounter a shuffle mask combination
that doesn't actually exist. (In practice, this only happens for degenerate
no-op tests like 1->1 shuffles)

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c
libswscale/x86/ops_common.asm
libswscale/x86/uops_macros.asm.h

Change #275117

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a785ff4ca2a73d151d5a8b7ce5d101063fe3dc3c

Comments

swscale/x86/ops: add AVX512 vpermb packed shuffle path
We can quite easily generalize the existing AVX512 pushfb implementation to
support vpermb when the shuffle mask is cross-lane.

  rgb24 1920x1080 -> rgba 1920x1080, speedup=1.372x faster

This does require some finesse w.r.t. the MOVSIZE macro, to avoid reading or
writing too many bits. In particular, for gray->rgba64le, the pattern even
devolves into 8/64 which would trigger an illegal operand error when combining
movq with zmm registers.

Other than that, we can use {k1}{z} zero masking to cleanly recreate the
semantics of pushfb's implicit zero masking, except for non-zero clear
values where we continue doing what we already do.

Lastly, while EVEX-encoded vpermb can take the source operand directly, this
somewhat defeats the purpose of the READ size optimization; in practice it
was better to keep the read and vpermb separate.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/x86/ops.c
libswscale/x86/ops_common.asm

Change #275118

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 17 Jul 2026 15:23:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0869e710e6876792fbcebccb536ad620d8e65b97

Comments

swscale/ops_optimizer: remove now-unused ff_sws_solve_shuffle()
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_internal.h
libswscale/ops_optimizer.c

Change #275148

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 04:56:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c60ac8f256cd76341df4fb5abdfd38d9c47c0a6e

Comments

avfilter/dnn: implement batching for Torch backend
Add batch processing support to the LibTorch DNN backend, following
the same pattern used by the OpenVINO backend.

Key changes:
- Add batch_size AVOption (range 1-1000, default 1) to DnnContext
- Accumulate lltasks in the queue, trigger inference when batch_size
  lltasks are ready (matching the OpenVINO batch execution pattern)
- Pre-allocate a single contiguous memory buffer for the entire batch
  in fill_model_input_th() to avoid unnecessary tensor copies
- Split batched output in infer_completion_callback() and dispatch
  each slice to its corresponding task

Tested with:
ffmpeg -f lavfi -i testsrc=duration=5:size=640x480:rate=25 -vf format=rgb24,dnn_processing=dnn_backend=torch:model=dummy_model.pt:batch_size=4 -f null -

Signed-off-by: Raja Rathour <imraja729@gmail.com>

Changed files

libavfilter/dnn/dnn_backend_torch.cpp
libavfilter/dnn/dnn_interface.c
libavfilter/dnn_interface.h

Change #275151

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 16:59:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9dcb14d9825415c35d675b4e22d36db5d9729756

Comments

tests/fate/id3v2.mak: add test for XXX lang in COMM and USLT frames, fix comment for -sort test.

Changed files

tests/fate/id3v2.mak
tests/ref/fate/id3v2-comm-foo-lang-as-descriptor
tests/ref/fate/id3v2-comm-sort-
tests/ref/fate/id3v2-comm-sort-eng
tests/ref/fate/id3v2-lang-und
tests/ref/fate/id3v2-lang-xxx
tests/ref/fate/id3v2-lang-xxx-remux

Change #275152

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 16:59:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9998dc7cc0c4b74d48f2bad16d7b205afce77001

Comments

fftools/ffmpeg_mux_init.c: Add comment-* version of the filtered iTunes id3v2 tags that are actually encoded as comments.

Changed files

fftools/ffmpeg_mux_init.c

Change #275153

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 16:59:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eb107d87723508907f1690552b29f803f241068e

Comments

libavformat/id3v2.c: Factor out support for reading lyrics and comment tags.

Changed files

libavformat/Makefile
libavformat/id3v2.c
tests/ref/fate/id3v2-lang-xxx
tests/ref/fate/id3v2-lang-xxx-remux
tests/ref/fate/id3v2-wma-comm
tests/ref/fate/id3v2-wma-comm-asf_o

Change #275154

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 16:59:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b0c702aec28b00aa08cce05c6066e535a5c210a5

Comments

libavformat/id3v2: deprecate legacy COMM descriptor-as-key behavior

Changed files

doc/APIchanges
libavformat/avformat.h
libavformat/id3v2.c
libavformat/version.h
libavformat/version_major.h
tests/ref/fate/id3v2-reenc-remux-keep
tests/ref/fate/id3v2-wma-comm
tests/ref/fate/id3v2-wma-comm-asf_o

Change #275155

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 16:59:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b79eae1af40f5831d5529eae6b919c43a713be00

Comments

libavformat/id3v2enc.c: add write support for COMM/USLT descriptor+lang keys

Changed files

libavformat/avformat.h
libavformat/id3v2.c
libavformat/id3v2enc.c
tests/ref/fate/generic-tags-remux-mp3
tests/ref/fate/id3v2-comm
tests/ref/fate/id3v2-comm-dashed-descriptor
tests/ref/fate/id3v2-comm-descriptor
tests/ref/fate/id3v2-comm-descriptor-no-lang
tests/ref/fate/id3v2-comm-foo-lang-as-descriptor
tests/ref/fate/id3v2-comm-invalid-lang
tests/ref/fate/id3v2-comm-lang-as-descriptor
tests/ref/fate/id3v2-comm-long-descriptor
tests/ref/fate/id3v2-comm-multi-invalid-lang
tests/ref/fate/id3v2-comm-raw-key
tests/ref/fate/id3v2-comm-sort
tests/ref/fate/id3v2-comm-sort-
tests/ref/fate/id3v2-comm-sort-eng
tests/ref/fate/id3v2-lang-und
tests/ref/fate/id3v2-lang-xxx-remux
tests/ref/fate/id3v2-lyrics
tests/ref/fate/id3v2-reenc-remux-keep

Change #275156

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 17:48:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	de038122a37c7c4f4938d4c0340584bf25789447

Comments

tests/fate/ogg-vorbis.mak: add test to show packets initial padding and side data.

Changed files

tests/fate/ogg-vorbis.mak
tests/ref/fate/ogg-vorbis-bear
tests/ref/fate/ogg-vorbis-initial-padding-chained
tests/ref/fate/ogg-vorbis-libvorbis-1s
tests/ref/fate/ogg-vorbis-libvorbis-30s
tests/ref/fate/ogg-vorbis-ts-offset
tests/ref/fate/ogg-vorbis-ts-seek

Change #275157

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 17:48:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	63b7e4cefa91aac517e3012d64f75c754e680674

Comments

lavf/oggparsevorbis: rework timestamp offset and end trimming
Replace the first-page packet scan with a direct use of the first
packet's duration as the timestamp offset (the Vorbis I spec
guarantees it produces no PCM output).

Track last_page_granule to correctly anchor EOS trimming regardless
of any initial granule offset. Track bos_pos to reinitialize parser
state at stream boundaries without resetting at page boundaries.

Restore the granule=0 workaround from ticket #3710 and handle
chained streams by clearing bos_pos when new headers are collected.

Vorbis I spec: https://xiph.org/vorbis/doc/Vorbis_I_spec.html

Changed files

libavformat/oggdec.c
libavformat/oggdec.h
libavformat/oggparsevorbis.c
tests/ref/fate/limited_input_seek
tests/ref/fate/limited_input_seek-copyts
tests/ref/fate/mov-mp4-chapters
tests/ref/fate/ogg-vorbis-bear
tests/ref/fate/ogg-vorbis-chained-meta.txt
tests/ref/fate/ogg-vorbis-initial-padding-chained
tests/ref/fate/ogg-vorbis-libvorbis-1s
tests/ref/fate/ogg-vorbis-libvorbis-30s
tests/ref/fate/ogg-vorbis-ts-offset
tests/ref/fate/ogg-vorbis-ts-seek
tests/ref/fate/webm-dash-chapters
tests/ref/lavf-fate/vp8.ogg

Change #275158

Category	ffmpeg
Changed by	张永鹏 <roc63ohnoyoudont@outlook.com>
Changed at	Sat 18 Jul 2026 19:04:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	629a5b4b0fcc0e574e4c35068d81445ec399df37

Comments

fftools: reject negative -t durations
Reject negative recording durations for input and output -t before they reach streamcopy or trim handling. Keep -t 0 accepted.

Signed-off-by: 张永鹏 <roc63@outlook.com>

Changed files

fftools/ffmpeg_demux.c
fftools/ffmpeg_mux_init.c
tests/fate/ffmpeg.mak

Change #275159

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 18 Jul 2026 19:06:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	70a5df6d274400b186a1166551be3aa304b1d195

Comments

fftools: fix manual merge missing changes
Signed-off-by: Romain Beauxis <romain.beauxis@gmail.com>

Changed files

fftools/ffmpeg_demux.c
fftools/ffmpeg_mux_init.c

Change #275165

Category	ffmpeg
Changed by	Kirill A. Korinsky <kirillohnoyoudont@korins.ky>
Changed at	Sun 19 Jul 2026 01:09:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	65c6383f899be853e6bed30797909073956e4745

Comments

configure/x86: test 6- and 7-register inline asm
HAVE_6REGS and HAVE_7REGS should describe whether the compiler can
allocate the inline asm register sets they guard. The old x86-32 logic
inferred that from EBX/EBP availability, while x86-64 enabled both
macros without a configure test.

Probe the 6- and 7-register asm constraints directly for all x86 targets
and let asm.h use the configure results. Drop the EBP availability
probe, which is no longer used. Keep EBX availability, as other inline
asm still uses it to decide whether named EBX can be clobbered.

Changed files

configure
libavutil/x86/asm.h

Change #275166

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 19 Jul 2026 02:48:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6d300b473266cda5926d86fe5a9d126c9c10b3c6

Comments

avformat/mov: avoid overflow/negative discard sample duration
This also fixes a corner case with st->duration = INT64_MAX

This assumes pkt->duration >= 0

Fixes: 525566001/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-5952332261818368
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/mov.c

Change #275168

Category	ffmpeg
Changed by	Omkhar Arasaratnam <omkharohnoyoudont@linkedin.com>
Changed at	Sun 19 Jul 2026 05:56:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b96701098fd8bc33a7726b895ca92102da65202d

Comments

avformat/tls_mbedtls: check the certificate verification result in tls_handshake()
tls_mbedtls verifies the peer certificate manually via
mbedtls_ssl_get_verify_result() (it uses MBEDTLS_SSL_VERIFY_OPTIONAL), but
only in tls_open(). On the external-socket path (external_sock=1) tls_open()
skips the handshake; it then runs later through the url_handshake hook,
tls_handshake(), which never checks the verification result, so with verify=1
an untrusted peer certificate would be accepted.

The only in-tree user of this path is the WHIP muxer, which sets verify=0
(WebRTC binds the peer via the SDP a=fingerprint, not a CA), so this is not
reachable today. This change is defence-in-depth: it makes tls_handshake()
honor verify symmetrically with the existing tls_open() check.

Signed-off-by: Omkhar Arasaratnam <omkhar@linkedin.com>

Changed files

libavformat/tls_mbedtls.c

Change #275169

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 19 Jul 2026 12:20:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ccaeb89a643a98717951bbe562fce0f4582b3182

Comments

avformat/shared: use strong CAS for marking PENDING
This was using a weak atomic under the assumption that the `goto retry`
will immediately jump back to the same line. However, that's no longer
true after the introduction of the `read_block` label, which we can
jump to from the error handling fallback path.

In this case, we really ought to use a strong CAS to ensure that we
only re-check the state if it truly changed in the meantime.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #275170

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 19 Jul 2026 12:20:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c764c4b8e47cd3764aff1de33b1d1f710dd34980

Comments

avformat/shared: give up on cache file after too many corrupt blocks
If we got 10 corrupt blocks in a row, chances are, the rest of the file is
completely corrupt. I didn't bother making this user-configurable as it is
already sort of a niche option to work around lack of a reliable filesystem.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #275171

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 19 Jul 2026 12:20:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9bc73ba3449ff91e760b8c4e0a5d2bbc953ab9ac

Comments

avformat/shared: treat unexpected EOF as corruption
This can happen if e.g. the file is truncated (e.g. by a filesystem bug
or self-healing response to an underlying checksum mismatch), in which case
we should treat it similar as a corruption case and retry it under
-retry_corrupt.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #275172

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	26586c898fadd410989fa52f591dce3055e08e53

Comments

avcodec/psymodel: add pair-synced window switching hook and pair state
Divergent block switching between the two channels of a CPE disables
common_window, and with it every joint stereo tool, on exactly the
frames that need them most; the lone short channel then codes its
transient alone, which is audible as one-channel garble at beats.

Add an optional window_pair() model hook so a psy model can decide both
channels of a pair together, and encoder-fed per-pair state: the
joint-tool candidacy fraction EMA and a Schmitt-gated decouple flag,
letting the model fall back to independent per-channel switching on
content whose joint tools are dead (diffuse, decorrelated pairs) where
forcing a common grid only costs quality.

Changed files

libavcodec/psymodel.h

Change #275173

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	646f2796c8e267fd77b6f6dacc60c6e9ea333c95

Comments

avcodec/aacenc: rework NMR rate control, pool CPE budgets, add decision memory
Rate control: replace the integral servo with a stateless pressure
offset (exp2(-K*fill/R)) so a drained reservoir cannot wind up and
crater quality after loud stretches; slew-limit the final operating
lambda per frame (bits deviate instead, the reservoir absorbs); seed
the reservoir full at stream start; and track rate strain explicitly:
a long-frame lambda EMA against anchors that scale up when the achieved
distortion/mask ratio flags noise-class content (whose psy masks are
wholesale violated and lambda reads inflated), plus a lambda min-tracker
separating sustained starvation from transient spikes at a comfortable
operating point. The resulting pressure ramp gates every
pressure-adaptive tool from one place.

CPE budget pooling: solve both channels of a pair jointly under one
shared lambda against a pooled budget (NMRSlot defer/solve/commit)
instead of an equal per-channel split, which starved the mid/carrier
while the side gold-plated. Mono and VBR output are unchanged.

Transients: isolated onsets are coded uniformly finer across the short
run and repaid from steady stretches; dense-beat runs get a
starvation-scaled boost; a transition premask clamps START-frame
thresholds toward the previous long frame (an attack cannot mask
backwards).

Decision memory: marginal per-frame re-decisions oscillate audibly, so
every stateful choice now carries hysteresis - band zeroing, PNS
enter/leave with debounce (and near-masked bands staying noise until a
loudness guard), per-grid stereo mode banks that survive window
switches, and the short-TNS accept state.

TNS-covered bands price distortion by the synthesis filter's
re-amplification gain, so the trellis spends where noise will actually
be heard.

Changed files

libavcodec/aaccoder_nmr.h
libavcodec/aacenc.h
tests/ref/fate/id3v2-reenc-delete-metadata
tests/ref/fate/id3v2-reenc-delete-metadata-keep
tests/ref/fate/id3v2-reenc-delete-metadata-keep-format
tests/ref/fate/id3v2-reenc-delete-metadata-keep-stream
tests/ref/fate/id3v2-reenc-delete-metadata-map-metadata

Change #275174

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	abe157fbf2bc34f99ef87e55abfdf5fe74d5651e

Comments

avcodec/aaccoder: widen PNS candidacy to near-masked bands
Bands whose energy sits between threshold/4 and 2x threshold are
deletion candidates for the trellis: coded, they round to silence under
pressure, and an audible hole is worse than energy-matched noise. Let
them qualify for PNS with a relaxed uniformity requirement, so the
noise substitution keeps the texture the allocator was about to delete.

Changed files

libavcodec/aaccoder.c

Change #275175

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aa88527d27e564348b854045e3bbd35705576abc

Comments

avcodec/aacenc_tns: enable short-block TNS via pooled per-group filters
Per-window filters combined with group-shared scalefactors made
short-block TNS produce decoder-synthesized silence: a weak residual in
one sub-window rounds to zero under the group's scalefactor and the
synthesis filter has nothing to re-amplify, audible as 1.5-3 ms
dropouts on beat content. Fit ONE filter per window group instead, on
the concatenated weighted spectra, and require every window in the
group to pass the prediction-gain bar.

Accepts are Schmitt-gated and run-scoped: within a short run the enter
bar is raised and the hold bar lowered (accept flicker was audible as
gravel), while isolated transient frames - where TNS matters most and
cannot flicker - use the base bar.

Changed files

libavcodec/aacenc_tns.c

Change #275176

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c8dc3cf4003cb09f9fc4b2d3dbce2d01c7b0f77e

Comments

avcodec/aacpsy: sync block switching across channel pairs
Implement the window_pair() hook: detect both channels of a CPE, then
decide together - if either channel attacks, both switch, with one
merged attack map so the grouping matches across the pair and
common_window survives transients. Pairs the encoder has flagged as
decoupled (joint tools dead) keep independent per-channel decisions,
where forcing the steady channel short at every event in the other
only costs quality.

Attacks must also be NOVEL against the trailing high-pass max-envelope:
a steady pitch-pulse train repeats its peak every period and re-fires
the ratio test forever, producing long false short runs on periodic
content (low brass and the like); a genuine onset towers over the
recent past.

The pre-echo threshold relaxation for gentler isolated onsets is kept,
and START frames entering a short sequence clamp their thresholds
toward the previous long frame's per-band values on the mapped long
grid.

Changed files

libavcodec/aacpsy.c

Change #275177

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9d0a4132ff5b53ba2240ed2465f953a0d1a55634

Comments

avcodec/aacenc: rework NMR stereo decisions, retune coded bandwidth
Stereo decisions are made per band before quantization, from the psy
model's spectra, and carry cross-frame memory (EMA-smoothed statistics,
per-grid mode banks, leave-hysteresis) so the image holds instead of
churning:

  * M/S adopts content-driven and rate-free (side under half the mid);
    for mid-dominant bands M/S is simply the better coding at every
    rate, and a wandering L/R fraction reads as image instability.
  * I/S competes with M/S above 6.1 kHz instead of only seeing M/S
    rejects (which are exactly the wide bands it cannot render), and
    engages under SUSTAINED strain only: the pressure ramp gated by the
    lambda floor, so pressure spikes at a comfortable operating point
    cannot flood it onto content where coding the side is affordable
    and strictly better. Unengaged candidates fall back to M/S.
  * Pairs whose joint-tool candidacy fraction stays low decouple:
    block switching goes per-channel and M/S stops, matching how
    independent coding wins on diffuse decorrelated content.
  * PNS in a pair is reserved for clearly-wide bands (it renders
    uncorrelated noise per channel).

Block switching for a pair is decided through the psy window_pair()
hook so common_window survives transients.

The NMR rate-to-bandwidth table is retuned upward at >= 48 kbps/ch to
track the bandwidth strong encoders deliver; both quality metrics
improve on a 16-clip battery and wider regresses. Lower rates are
unchanged.

Changed files

libavcodec/aacenc.c
tests/fate/aac.mak
tests/ref/fate/id3v2-reenc-delete-metadata
tests/ref/fate/id3v2-reenc-delete-metadata-keep
tests/ref/fate/id3v2-reenc-delete-metadata-keep-format
tests/ref/fate/id3v2-reenc-delete-metadata-keep-stream
tests/ref/fate/id3v2-reenc-delete-metadata-map-metadata

Change #275178

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 14:41:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ddc587bad18eb8a46425008b91bd18cb7aa97e46

Comments

aacenc: add an info log to ask users to resample

Changed files

libavcodec/aacenc.c

Change #275179

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	086029215702b95e980898c4e981fe79d2122ce1

Comments

swscale/aarch64/ops: consistently use SWS_COMP() to test and set mask values
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_impl_conv.c

Change #275180

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4e91f49e3db857e71451f7acb41b0a6808d7bad2

Comments

swscale/aarch64/ops: convert swizzle operation to a series of moves
This is similar to the way SWS_UOP_PERMUTE/SWS_UOP_COPY work and will
simplify the move to uops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c

Change #275181

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	be71c4a70f3fec81f3915b3d4ac9f62f4bb2509c

Comments

swscale/aarch64/ops: specialize convert and expand operations
This is similar to the way SWS_UOP_TO_{U8,U16,U32,F32} and
SWS_UOP_EXPAND_{PAIR,QUAD} work and will simplify the move to uops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c

Change #275182

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e66601ac6b410f05d353d63ad4b9f05816fcf18c

Comments

swscale/aarch64/ops: specialize linear operation
This is similar to the way SWS_UOP_LINEAR{,_FMA} work and will simplify
the move to uops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275183

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cc7784205c5b2877c693fafcfe859d6f3caada9d

Comments

swscale/aarch64/ops: specialize clear operation
There are now more specialized functions for components that clear to
zero and one (using `movi` instead of `dup`).

This is similar to the way SWS_UOP_CLEAR works and will simplify the
move to uops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c

Change #275184

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5778cbb11fe00d30d2d9cb4e666a60eec1961257

Comments

swscale/aarch64/ops: remove overly-complicated generated lookup function
... and replace it by a simple linear lookup over ops_entries.

The previous approach was faster, but for an init function this hardly
matters.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/Makefile
libswscale/aarch64/ops.c
libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.c
libswscale/aarch64/ops_lookup.h
libswscale/tests/sws_ops_aarch64.c

Change #275185

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f4edbf0b38bb88f3ad950754063266f979bc4ebb

Comments

swscale/aarch64/ops: remove overly-complicated field-specific functions
... and replace them by simple switches.

Since the last commit, op_fields is no longer used by aarch64/ops_asmgen,
so it can be simplified into its only user.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_impl.c
libswscale/tests/sws_ops_aarch64.c

Change #275186

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	000bc6c85a8a6deb4fc4d4a91a3f8599924ff8f9

Comments

swscale/tests/sws_ops_aarch64: use AVBPrint instead of custom string functions
buf_appendf() was used when this code still needed to run standalone in
aarch64/ops_asmgen. Since this is no longer the case, use AVBPrint
instead.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/tests/sws_ops_aarch64.c

Change #275187

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a839f444b002cf0cfef1013672f633d135ece5b3

Comments

swscale/aarch64/ops: use SwsUOpType/SwsPixelType instead of SwsAArch64OpType/SwsAArch64PixelType
Also use ff_sws_pixel_type_size() instead of aarch64_pixel_size().

This is a step towards the move to uops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275188

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:02
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3332e5e27895051a2bacfe63064d19831281768b

Comments

swscale/aarch64/ops: use SwsShiftUOp for SwsAArch64OpImplParams.shift
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275189

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	57ca4d2ae330e5acfd0486b01c1dc6c5f817ba6d

Comments

swscale/aarch64/ops: use SwsClearUOp for SwsAArch64OpImplParams.clear
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275190

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c0e5d4682cdb9daeb2e359c41b44e5921a447cd

Comments

swscale/aarch64/ops: use SwsMoveUOp for SwsAArch64OpImplParams.move
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275191

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e45e0e708420294995a67f2013b5a64218f1436d

Comments

swscale/aarch64/ops: use SwsPackUOp for SwsAArch64OpImplParams.pack
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275192

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4a33d2ce2c5cc08988e357b394fbce5ec2b4ac4a

Comments

swscale/aarch64/ops: use SwsLinearUOp for SwsAArch64OpImplParams.linear
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops.c
libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275193

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	353b8297a5c81c08049b43845ff21876e360b3f7

Comments

swscale/aarch64/ops: use SwsDitherUOp for SwsAArch64OpImplParams.dither
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275194

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0536c93b3289c98f90a2dfae7cc9045891a9252c

Comments

swscale/aarch64/ops: use SwsCompMask for SwsAArch64OpImplParams.mask
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275195

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	19cc66f02f2926fc2b9246a221015cebc7d5f501

Comments

swscale/aarch64/ops: use SwsUOpParams for SwsAArch64OpImplParams's parameter fields
This is one more step to eventually replace the parameter fields in
SwsAArch64OpImplParams by generic structs from libswscale/uops.h.

The function names and ordering in ops_entries.c is maintained to
simplify the gradual move to uops.h.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops.c
libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_entries.c
libswscale/aarch64/ops_impl.h
libswscale/aarch64/ops_impl_conv.c
libswscale/tests/sws_ops_aarch64.c

Change #275196

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Sun 19 Jul 2026 15:07:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	82c1a01142177282beee13382f8b57c8cb2c9a0f

Comments

swscale/tests/sws_ops_aarch64: pretty-print entries in ops_entries.c
This makes them all neatly aligned.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_entries.c
libswscale/tests/sws_ops_aarch64.c

Change #275200

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 19 Jul 2026 17:50:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3a8a1a7c6a20cb2738bc3721624b7d10d1fec19d

Comments

avcodec/aacpsy: fix the attack lookback
This fixes a 15+ year old bug.

The novelty check read its previous-frame envelope from
prev_energy_subshort after the sub-block loop had already overwritten
it with current-frame values, so early sub-blocks were compared
against the frame's own future instead of the past. That accident
suppressed short-period pitch trains well enough, but also suppressed
genuine onsets, and could not see pulse periods beyond ~12ms at all:
on quiet slow pulse trains (engine-idle buzz and the like) the ratio
test leaks sporadic isolated short excursions, each an audible click -
reported against this encoder since 2015, at every bitrate, and
immune to disabling every coding tool.

Keep a rolling two-frame history of HP sub-block peaks instead and
require an attack to tower over it: within ~12ms unconditionally
(pitch-rate trains), and within ~44ms when coming from steady
long-window state (slow trains, where an isolated short excursion can
only click). Dense irregular transients tower locally and reset
frames_since_short, so their block switching is untouched.

Changed files

libavcodec/aacpsy.c
tests/fate/aac.mak

Change #275201

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Sun 19 Jul 2026 18:10:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e2acef488f20098576cfbcd99c0c256313e913e8

Comments

forgejo/workflows: add workflow for automatic backport PRs via labels

Changed files

.forgejo/workflows/backports.yml

Change #275202

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Sun 19 Jul 2026 18:29:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	162c2784f90969ae53c1f4aa36d22ef93945a293

Comments

forgejo/workflows: fix backport condition compatibility

Changed files

.forgejo/workflows/backports.yml

Change #275227

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Sun 19 Jul 2026 22:47:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1b0c9e3dce64423052e204ca5669a21f9f9a2eac

Comments

forgejo/workflows: block concurrent backport runs for the same pull request

Changed files

.forgejo/workflows/backports.yml

Change #275249

Category	ffmpeg
Changed by	sfan5 <sfan5ohnoyoudont@live.de>
Changed at	Mon 20 Jul 2026 04:35:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8fe61b7a6446b6a687b9056ba34664a2de168e5a

Comments

libavcodec/mediacodecdec_common: improve logging for failing to get codec
The logging should obviously happen before returning an error.
Also include the mime type and detected profile.

Signed-off-by: sfan5 <sfan5@live.de>

Changed files

libavcodec/mediacodecdec_common.c

Change #275250

Category	ffmpeg
Changed by	sfan5 <sfan5ohnoyoudont@live.de>
Changed at	Mon 20 Jul 2026 04:35:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c23123630e6a7e645c199599b8ade3fe7e9ab3db

Comments

libavcodec/mediacodecdec_common: support AV_HWACCEL_FLAG_ALLOW_PROFILE_MISMATCH
Signed-off-by: sfan5 <sfan5@live.de>

Changed files

libavcodec/mediacodecdec_common.c

Change #275322

Category	ffmpeg
Changed by	StachowiakDawid <dawidos234ohnoyoudont@gmail.com>
Changed at	Mon 20 Jul 2026 19:49:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c182b1c09c086deb62545393cc7b4a829d541508

Comments

Add new mode to mpdecimate video filter

Changed files

doc/filters.texi
libavfilter/vf_mpdecimate.c
tests/fate/filter-video.mak
tests/ref/fate/filter-mpdecimate-mode-1
tests/ref/fate/filter-mpdecimate-mode-1-min-3

Change #275326

Category	ffmpeg
Changed by	David Rosca <david.roscaohnoyoudont@amd.com>
Changed at	Tue 21 Jul 2026 06:25:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c349297c163151c2576882887ae12e4cf2bdae4

Comments

avcodec/vulkan_encode_av1: fix level auto-detection using zero dimensions
Move level guessing from init_profile() to init_sequence_headers(),
where base_ctx->surface_width/height are already initialized.

init_profile() is called from ff_vulkan_encode_init() before the
surface dimensions are set, so ff_av1_guess_level() always received
zeros and selected the lowest level (2.0). init_sequence_headers()
runs after ff_vulkan_encode_init() returns, matching how H.264 and
H.265 Vulkan encoders handle level detection.

Co-Authored-By: Claude Opus 4

Changed files

libavcodec/vulkan_encode_av1.c

Change #275327

Category	ffmpeg
Changed by	David Rosca <david.roscaohnoyoudont@amd.com>
Changed at	Tue 21 Jul 2026 06:25:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ccc57378b37d9129396a037df02c83a877d8eef0

Comments

avcodec/vulkan_encode_av1: fix level and tier option handling
Copy opts.level into avctx->level (matching H.264/H.265 encoders),
so the -level command line option is actually used. Also initialize
enc->seq_tier from opts.tier so the -tier option takes effect.

The -level option values (0-23) already match the StdVideoAV1Level
enum directly, so assign avctx->level to seq_level_idx without going
through ff_vk_av1_level_to_vk() which expects AV1 spec-style level
numbers (20, 21, 30, ...).

Co-Authored-By: Claude Opus 4

Changed files

libavcodec/vulkan_encode_av1.c

Change #275378

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 21 Jul 2026 20:48:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	44889493e3b11a90a9a816ed350cfada17f01c84

Comments

tools/target_dec_fuzzer: use av_fallthrough to avoid warning
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/target_dec_fuzzer.c

Change #275379

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 21 Jul 2026 20:48:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	95d5815003258be03929e0fcbd6d5a504c5ccd2b

Comments

tools/target_enc_fuzzer: mark flags64 as unused
Still read, to not change byte stream offsets. Technically we could skip
those bytes, but this way it's clear for te future that we can use those
flags for anything needed.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/target_enc_fuzzer.c

Change #275380

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 21 Jul 2026 20:48:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b96a1ff373e5442707c88a10f6c1941fbeb6a121

Comments

tools/target_enc_fuzzer: don't access removed AVCodec::pix_fmts
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/target_enc_fuzzer.c

Change #275403

Category	ffmpeg
Changed by	Leo Izen <leo.izenohnoyoudont@gmail.com>
Changed at	Tue 21 Jul 2026 23:41:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b506fafec9a19fcbc2be5271875fd4a63d6615bc

Comments

avcodec/pngenc: fix overrun caused by exif size discrepancy
If the estimated EXIF size based on the attached frame data is lower
than the actual EXIF size after sanitizing the IFD, then an overrun
might occur. Instead, we parse the IFD and use the parsed size as the
estimated exif size so there won't be any discrepancy between the two
values.

Signed-off-by: Leo Izen <leo.izen@gmail.com>
Reported-by: Adrian Junge <adjun37@gmail.com>

Changed files

libavcodec/pngenc.c

Change #275408

Category	ffmpeg
Changed by	KawaiiEngine (Sashimiso) <209421934+KawaiiEngineohnoyoudont@users.noreply.github.com>
Changed at	Wed 22 Jul 2026 00:39:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0f1e9710896108e8aa7f60f8b1c6d16333eab2ae

Comments

avutil/hwcontext_cuda: drain queued copies on transfer error
cuda_transfer_data() can enqueue copies for earlier planes before a
later cuArrayGetPlane() or cuMemcpy2DAsync() fails. Returning at that
point can let queued work outlive caller-owned AVFrame buffers.

Track whether a copy was successfully queued and synchronize the stream
on these later failures. Synchronization is best-effort: its errors and
cuCtxPopCurrent() errors stay logged without replacing the primary
failure.

Successful H2D and D2D transfers remain asynchronous, and the normal
host-download synchronization path is unchanged.

Signed-off-by: KawaiiEngine (Sashimiso) <209421934+KawaiiEngine@users.noreply.github.com>

Changed files

libavutil/hwcontext_cuda.c

Change #275409

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 22 Jul 2026 04:50:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f0f634b6585fdc7bbb43ab3ae461499bfca9ad2e

Comments

avfilter/vf_hqdn3d: reject unsupported frame parameter changes
Fixes: out of array access
Fixes: 9aj_hqdn3d_dynamic_res.mjpg / 9aj_generate_hqdn3d_dynamic_res_mjpg.py
Fixes: wWDsy2oDvMuR
Found-by: Adrian Junge (vurlo) <adjun37@gmail.com>

Changed files

libavfilter/vf_hqdn3d.c
libavfilter/vf_hqdn3d.h

Change #275410

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 22 Jul 2026 04:50:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5d7112c60e6f0f0742ce47d448e6da0718a70f4c

Comments

avfilter/vf_hqdn3d: support dynamic frame sizes

Changed files

libavfilter/avfilter.c
libavfilter/vf_hqdn3d.c

Change #275411

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Wed 22 Jul 2026 05:18:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1b1f6026990bd081ec6e2cef8cd88f60ddbfea66

Comments

avcodec/d3d12va_encode: set extradata for global header when muxing AV1 to MKV
When the muxer sets AV_CODEC_FLAG_GLOBAL_HEADER (as MKV does), the codec
must populate avctx->extradata with the sequence header so the muxer can
write the av1C codec private entry. Otherwise, the mkv muxer will return
AVERROR_INVALIDDATA on the first packet (Invalid data found when
processing input).

Example usage:
  # use mkv container:
  ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4 -c:v av1_d3d12va output.mkv

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavcodec/d3d12va_encode.c

Change #275473

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 22 Jul 2026 14:52:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f8c4828c0f1b5b15c404042d390dbfe38fc4d023

Comments

avutil/x86/asm: Remove HAVE_[67]REGS, use HAVE_X86_[67]REGS directly
This allows to remove the asm.h inclusion from x86/mathops.h
and therefore from hundreds of files which don't need it.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/dirac_arith.h
libavcodec/msmpeg4.c
libavcodec/x86/cabac.h
libavcodec/x86/h264_cabac.c
libavcodec/x86/mathops.h
libavcodec/x86/mlpdsp_init.c
libavcodec/x86/mpeg4videodsp.c
libavcodec/x86/mpegaudiodsp.c
libavcodec/x86/snowdsp_init.c
libavcodec/x86/vc1dsp.h
libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_mmx.c
libavcodec/x86/vpx_arith.h
libavfilter/x86/vf_noise.c
libavutil/x86/asm.h
libswscale/x86/rgb2rgb.c
libswscale/x86/swscale.c
libswscale/x86/swscale_template.c

Change #275474

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 22 Jul 2026 14:52:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	68c94ed330a5b00d6b655e93679f0f189c9bcf00

Comments

av{codec,filter}/x86: Remove unnecessary asm.h inclusions
asm.h is only for inline assembly.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/ac3dsp_init.c
libavcodec/x86/apv_dsp_init.c
libavcodec/x86/h264dsp_init.c
libavcodec/x86/h26x/h2656dsp.c
libavcodec/x86/h26x/h2656dsp.h
libavcodec/x86/hevc/dsp_init.c
libavcodec/x86/huffyuvdsp_init.c
libavcodec/x86/mpegvideoenc.c
libavcodec/x86/utvideodsp_init.c
libavfilter/x86/vf_atadenoise_init.c
libavfilter/x86/vf_bwdif_init.c
libavfilter/x86/vf_eq_init.c
libavfilter/x86/vf_maskedclamp_init.c
libavfilter/x86/vf_tinterlace_init.c
libavfilter/x86/vf_transpose_init.c
libavfilter/x86/vf_w3fdif_init.c
libswscale/utils.c

Change #275475

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 22 Jul 2026 14:52:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1cf582b00ce14b6c8639e62a154497066f2c3bf6

Comments

configure: Remove unused check_exec_crash
Unused since 65c6383f899be853e6bed30797909073956e4745.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

configure

Change #275477

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	53f1b3dd0917ea4645e553510bbc94c5574ebf56

Comments

swscale/aarch64/rasm: fill AArch64VecViews by value
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/rasm.c
libswscale/aarch64/rasm.h

Change #275479

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e418d44c1e19580731699716ba35a30be23b7c48

Comments

swscale/aarch64/rasm: add extra helpers for instructions that take all 16b operands
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/rasm.h

Change #275481

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d34ed6ce404bf4a6e73a8a9f9e36c0029e275c3a

Comments

swscale/aarch64/ops_asmgen: use and/orr arrangement specifier helpers from rasm
This way, we don't depend on the RasmOp having been correctly reshaped
prior to being used.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275483

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a9665b7c9696fe29844d1b5f4c85e060f5ea084

Comments

swscale/aarch64/ops_asmgen: use arrangement specifiers for vector registers with known size
This way, we don't depend on the RasmOp having been correctly reshaped
prior to being used.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275485

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	af0d56b00e33a6c55d32d743d9e812d73d28a2fb

Comments

swscale/aarch64/ops_asmgen: use RasmOp directly when generating masks for SWS_UOP_UNPACK
There is no need to pack the vector register usage for this function.
This change creates holes in the usage of const vector registers, but
populates the RasmOp const array directly, which will help for the
implementation of the JIT compiler.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275487

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	154791b150eb3b9ce2c646df1d95c785401b99d0

Comments

swscale/aarch64/ops_asmgen: add IMPL_PRIV helper for offset operand
This reduces the line length for instructions that access impl->priv.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275488

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	97590dbd48103b4742d767fa3eaf1e040ab02436

Comments

swscale/aarch64/ops_asmgen: split constant vector registers out of temp
This makes each vector register's purpose clearer, and will let the JIT
compiler later factor loading of constant values out of the main loop.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275489

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	98e536f4d3a9eb114079f803c5a281ad8f447d1d

Comments

swscale/aarch64/ops_asmgen: split reshape_all_vectors() into vector register classes
This reduces some unnecessary vector reshaping, and also makes the
intention clearer at the call-sites.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275490

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d99e656ee022f9000a705ef7ffd459d95c487544

Comments

swscale/aarch64/ops_asmgen: pass SwsAArch64OpRegs as a separate argument to asmgen_op_*()
This will help the JIT compiler by letting us provide a separate set of
registers for each operation.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275491

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7ee8947b2ca0bc1a3bf001c9a83749868bf94ba8

Comments

swscale/aarch64/ops_asmgen: split asmgen_setup_*() out of asmgen_op_*()
This paves the way for performing operation setup differently for the
JIT backend, which will have constant values factored out of the loop.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275492

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	90a8749bf35b25e7e61b9413db4c3a9ef67d4c92

Comments

swscale/aarch64/ops_asmgen: split cps code out of asmgen_process()
The asmgen_process() function is now more generic, which will allow us
to use it from the JIT compiler.

The naming of the function, along with the setup code and main loop,
are moved to asmgen_process_cps().

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275493

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	28a47aea66fba67a21c38fce8d9817441b6ff792

Comments

swscale/aarch64/ops_asmgen: pass both input and output masks to asmgen_process()
Currently, for CPS code, both masks are the same, similar to how the C
and x86 backends do it.

But the JIT compiler will have more precise input and output masks.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275494

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	25ee9a4218d522f780c8076c9e035656a81d213c

Comments

swscale/aarch64/ops_asmgen: use more (but simpler) loops for exec load generation
Same behaviour, just looks neater.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275495

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0096b488257cdfe9ac5acf51637c7974973e22a3

Comments

swscale/aarch64/ops_asmgen: load values from exec before performing setup
This changes little for the current CPS code, which has fairly simple
setup code. But in JIT we will factor out much more code from the main
loop into the setup section.

This also frees up the register used by exec (x0) earlier on, so it may
be reused earlier in JIT code.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275496

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	12a01917449ec8bb09683cc0e0ca9126b999ea81

Comments

swscale/aarch64/ops_asmgen: setup frame for each function
Currently this is redundant, since all CPS functions share the same
frame.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275497

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	940e4b29e150761c2ca9dea1970438fe29355c4e

Comments

swscale/aarch64/ops_asmgen: split vector init from reshaping
This changes nothing for the CPS code, but will allow the JIT compiler
to allocate and reorder vector registers without them being affected by
reshaping.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275498

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1b66342df27499271654b4b888c8e2e4e3b31272

Comments

swscale/aarch64/ops_asmgen: split input/output vector banks into separate register variables
For CPS these will continue being the same vector register numbers,
because of the fixed ABI, but for JIT we will be able to use different
input/output vector registers to implicitly perform swizzles.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275499

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	132187cdbba4230005d96705751dff95e2316e79

Comments

swscale/aarch64/ops_asmgen: increase the number of temp vectors to 12
This change is done separately from the next commit (which uses the new
temp vectors in the linear op) to make it easier to track changes to
ops_neon.gen.S.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275500

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ecbb883014676bf25ce62b50f056f9ec16424305

Comments

swscale/aarch64/ops_asmgen: move save_mask setup out of asmgen_op_linear() and into asmgen_setup_linear()
The save_mask setup makes copies of input data that would be clobbered
by the linear operation itself. This setup is not entirely necessary
with JIT, since we may have separate input and output vector registers,
so the code is moved into the setup function which will be CPS-only.

This doubles the amount of temp registers needed, since they must all
be copied before both loops.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275501

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1fcb227aec266d67712296bd21664d61ca13f94

Comments

swscale/aarch64/ops_asmgen: reword annotations for linear operation
Refer to offset[i] or coeff[i][j] instead of position of the value from
the packed data.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275502

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bc06af02c62c538ddd5024513271d5abf46a18fe

Comments

swscale/aarch64/ops_asmgen: use different value for vtmp register index
Use vtmp register index relative to the current matrix column instead
of reusing the position of the value from the packed data.

The i_coeff and derived values will be removed in the following commit.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275503

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e69c1d87f208cfcbacc5c55c175368c91d7a1e3e

Comments

swscale/aarch64/ops_asmgen: populate operands matrix in asmgen_setup_linear()
This will serve multiple purposes:
- Allow JIT code to populate operands from a deduplicated data pool;
- Simplify the future implementation of integer linear operation, since
  mul/mla take full vector registers as coefficients, and not elements.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/ops_asmgen.c

Change #275504

Category	ffmpeg
Changed by	Ramiro Polla <ramiro.pollaohnoyoudont@gmail.com>
Changed at	Wed 22 Jul 2026 16:07:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c61690153c9089e082e1b1c36ae48959d542b7e8

Comments

swscale/aarch64: split ops_static.c out of ops_asmgen.c
ops_asmgen.c contains common aarch64 asm generation code that will be
used by CPS backend and the JIT backend.

The standalone tool with CPS-specific code is now ops_static.c.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>

Changed files

libswscale/aarch64/.gitignore
libswscale/aarch64/Makefile
libswscale/aarch64/ops.c
libswscale/aarch64/ops_asmgen.c
libswscale/aarch64/ops_asmgen.h
libswscale/aarch64/ops_static.c

Change #275511

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 22 Jul 2026 16:08:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c450bf833caf2a35392e073cbf33b3598e40fe09

Comments

Revert "avformat/concatdec: Check recursion depth"
This reverts commit 58d39e99e26d1116f43a67d66ecfe71dba316875.
Superseded by the generic recursion limit introduced
in 597036b692b8c39198fe572aad027eeb4c13da35.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/concatdec.c

Change #275543

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Wed 22 Jul 2026 23:42:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2cf3f4d64de0efa5ccb4021f7245e93b041dbd9e

Comments

avutil/hwcontext_cuda: remove format allowlist, accept any pixel format
Remove the static supported_formats[] allowlist from hwcontext_cuda.
cuda_frames_get_constraints() now iterates all registered pixel formats
and includes every non-hwaccel, non-palette format. cuda_frames_init()
validates with av_pix_fmt_desc_get() instead of checking against the
allowlist.

Palette formats have a special-cased use of plane[1] for the palette itself
and it's not worth the hassle of supporting in an otherwise clean generic
copy loop.

Also add a roundtrip test that uploads a deterministic byte pattern to
CUDA, downloads it back, and verifies the data match. This tests all
non-hwaccel pixel formats. The test is gated on CONFIG_CUDA.

Changed files

libavutil/Makefile
libavutil/hwcontext_cuda.c
libavutil/tests/.gitignore
libavutil/tests/hwcontext_cuda.c
tests/fate/hw.mak

Change #275558

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 01:50:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	642ae726ff3a0d8d8c9416c0f1968b936df7e9c4

Comments

avformat/matroska: rescale default duration derived block durations to stream timebase
Block timestamps and durations are in Track Ticks, meaning they are scaled with both
Segment timescale and Track timescale.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/matroskadec.c

Change #275559

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 01:50:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9c8fc66ce196d40b12259458e045436cedc0d69e

Comments

tests/fate/matroska: add a test for Track Timescale handling
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/matroska.mak
tests/ref/fate/matroska-track-timescale

Change #275561

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 02:14:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	afb74faff6fb20cca0c15477872583a7c3d54977

Comments

tests/ref: update references
Missed in 642ae726ff3a0d8d8c9416c0f1968b936df7e9c4 after a faulty rebase.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/ref/fate/atrac-matroska-remux
tests/ref/fate/ffmpeg-streamloop-transcode-av
tests/ref/fate/h264-dts_5frames
tests/ref/fate/hevc-bsf-dovi-split-bl
tests/ref/fate/hevc-bsf-dovi-split-bl-rpu
tests/ref/fate/hevc-bsf-dovi-split-el
tests/ref/fate/hevc-bsf-dovi-split-el-rpu
tests/ref/fate/matroska-dovi-hvce-mkv-to-mkv
tests/ref/fate/matroska-dovi-write-config7
tests/ref/fate/matroska-flac-channel-mapping
tests/ref/fate/matroska-mastering-display-metadata
tests/ref/fate/matroska-ms-mode
tests/ref/fate/matroska-qt-mode
tests/ref/fate/mkv
tests/ref/fate/mkv-1242
tests/ref/fate/mov-dovi-hvce-mkv-to-mp4

Change #275572

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Thu 23 Jul 2026 05:26:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	80eb9e99b93491753fc9d5bf1f8718c361131998

Comments

avutil/hwcontext_cuda: fix unused variable 'i' warning
The variable i is only used inside the HAVE_FFNVCODEC_CUARRAY block.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavutil/hwcontext_cuda.c

Change #275634

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 15:39:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e1c1b6c55835d13647636340d478d370aa48fb8d

Comments

avcodec/get_buffer: use frame pixel format instead of context
The documentation for get_buffer2() states:

"This callback must use the frame values, not the codec context values, to
calculate the required buffer size."

Fixes wrong output on hbd LCEVC enhanced samples where the base stream is
8bit.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/get_buffer.c

Change #275647

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 18:56:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b3d4ce20a4608fe72af0b035b4007bba46a15279

Comments

avformat/lcevc: add a log context parameter to all functions
Printing to NULL should be avoided when possible.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/codecstring.c
libavformat/lcevc.c
libavformat/lcevc.h
libavformat/movenc.c

Change #275648

Category	ffmpeg
Changed by	Marcos Ashton Iglesias <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 23 Jul 2026 19:58:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9347affa3dc3aa35e90661a4a7564328f3db7972

Comments

avcodec/x86/hevc: add SSE2 and AVX transform_4x4_luma
Implements the 4x4 luma DST-VII inverse transform for 8 and 10 bit,
using pmaddwd on (src0,src2)/(src1,src3) pairs, mirroring the
existing 4x4 IDCT structure.

checkasm --bench on a Core Ultra 7 155H:
hevc_transform_luma_4x4_8_c:        151.0
hevc_transform_luma_4x4_8_sse2:      35.9 ( 4.20x)
hevc_transform_luma_4x4_8_avx:       34.2 ( 4.42x)
hevc_transform_luma_4x4_10_c:       146.5
hevc_transform_luma_4x4_10_sse2:     35.0 ( 4.19x)
hevc_transform_luma_4x4_10_avx:      33.7 ( 4.35x)

Part of #23022.

Changed files

libavcodec/x86/hevc/dsp_init.c
libavcodec/x86/hevc/idct.asm

Change #275667

Category	ffmpeg
Changed by	Jun Zhao <barryjzhaoohnoyoudont@tencent.com>
Changed at	Fri 24 Jul 2026 04:36:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2f209337fc66b58bf0495265880bb37580c3f981

Comments

avformat/mov: read DTS metadata from ddts box
The MOV demuxer only handled the low four bits of ChannelLayout and
did not expose the DTS profile carried by StreamConstruction. It also
discarded the sample-entry channel count when ChannelLayout was zero.

Parse StreamConstruction, RepresentationType, and all 16 ChannelLayout
bits according to ETSI TS 102 114. Map StreamConstruction to the
corresponding FFmpeg DTS profile, and preserve the sample-entry channel
count for zero layouts unless RepresentationType explicitly signals two
encoded channels.

This also provides profile and channel metadata for encrypted samples
whose frame headers cannot be inspected.

Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Changed files

libavformat/mov.c

Change #275709

Category	ffmpeg
Changed by	Guo Yejun <yejun.guoohnoyoudont@intel.com>
Changed at	Fri 24 Jul 2026 16:06:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6095372a70d17e48b67678532a492f943d206d00

Comments

avfilter/dnn: use batch_size in DnnContext instead of ov_option
batch_size in DnnContext is a common variable for all dnn backends

Changed files

libavfilter/dnn/dnn_backend_openvino.c
libavfilter/dnn_interface.h

Change #275718

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 24 Jul 2026 21:09:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	92cd5c97817be98361137e12fdba1f13db20f5c9

Comments

avfilter/vf_lut3d: do not compute size*size before the size is validated
parse_dat(), parse_cube(), and parse_cinespace() multiply an untrusted LUT size
before allocate_3dlut() validates it, which can overflow int.
Use the validated lutsize2 computed by allocate_3dlut() instead.

Fixes: signed integer overflow

Changed files

libavfilter/vf_lut3d.c

Change #275789

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	607862dd6685f53bdb26b6c92e4d04057d1c18f8

Comments

avcodec/x86/h264_intrapred: SSE2 impl. of pred8x8l_vertical_8
Deprecate MMX. Remove the SSSE3 impl. since we no longer use palignr.

pred8x8l_vertical_8_mmxext:                              7.4 ( 3.61x)
pred8x8l_vertical_8_ssse3:                               6.6 ( 4.03x)
pred8x8l_vertical_8_sse2:                                6.3 ( 4.19x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275790

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c7e532e81e44c5687270dfffa8c05e393bd94830

Comments

avcodec/x86/h264_intrapred: SSE2 impl. of pred8x8l_horizontal_8
Deprecate MMX. Remove the SSSE3 impl. since we no longer use palignr. No
change in performance.

pred8x8l_horizontal_8_mmxext:                           16.2 ( 1.44x)
pred8x8l_horizontal_8_ssse3:                            15.6 ( 1.50x)
pred8x8l_horizontal_8_sse2:                             15.8 ( 1.49x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275791

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3333082cd79fb4ad4b8cb4134a2350d63f0dc87a

Comments

avcodec/x86/h264_intrapred: SSE2 impl. of pred8x8l_horizontal_up_8
Deprecate MMX. Remove the SSSE3 impl. since we no longer use palignr and
SSE2 is faster.

pred8x8l_horizontal_up_8_mmxext:                        27.1 ( 2.30x)
pred8x8l_horizontal_up_8_ssse3:                         23.9 ( 2.60x)
pred8x8l_horizontal_up_8_sse2:                          20.9 ( 2.96x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275792

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	14dfce5eaf18b7630922117a2541d8646db21289

Comments

avcodec/x86/h264_intrapred: SSE2 impl. of pred8x8_dc_rv40_8
Deprecate MMX. No change in performance. Also add test coverage in
checkasm.

pred8x8_dc_rv40_mmxext:                                 10.0 ( 1.24x)
pred8x8_dc_rv40_sse2:                                   10.0 ( 1.24x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c
tests/checkasm/h264pred.c

Change #275793

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d03788af83260044c6ef0e21a1c24245920cb5c4

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred8x8l_down_right_8
Convert use of MMX to SSE2, also remove the SSSE3 impl. since we no
longer use palignr

Before:
 pred8x8l_down_right_8_sse2:                             26.4 ( 3.26x)
 pred8x8l_down_right_8_ssse3:                            23.0 ( 3.74x)

After:
 pred8x8l_down_right_8_sse2:                             19.7 ( 4.33x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275794

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ea2bf7e7757b6528dd5a83b0eb750e10dda86df2

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred8x8l_vertical_right_8
Convert use of MMX to SSE2, also remove the SSSE3 impl. since we no
longer use palignr

Before:
 pred8x8l_vertical_right_8_sse2:                         24.1 ( 3.51x)
 pred8x8l_vertical_right_8_ssse3:                        20.8 ( 4.07x)

After:
 pred8x8l_vertical_right_8_sse2:                         19.5 ( 4.38x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275795

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	093fda3c21679c4a81902f0baeaa361fc3b2a1cd

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred8x8l_vertical_left_8
Convert use of MMX to SSE2, also remove the SSSE3 impl. since we no
longer use palignr

Before:
 pred8x8l_vertical_left_8_sse2:                          17.3 ( 4.89x)
 pred8x8l_vertical_left_8_ssse3:                         15.0 ( 5.63x)

After:
 pred8x8l_vertical_left_8_sse2:                          11.8 ( 6.80x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275796

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1a0747633c25eed7592e36a94ce1b12ea7e067b1

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred8x8l_horizontal_down_8
Convert use of MMX to SSE2, also fix checkasm test.

Before:
 pred8x8l_horizontal_down_8_sse2:                        33.2 ( 2.53x)
 pred8x8l_horizontal_down_8_ssse3:                       27.8 ( 3.02x)

After:
 pred8x8l_horizontal_down_8_sse2:                        24.3 ( 3.45x)
 pred8x8l_horizontal_down_8_ssse3:                       22.6 ( 3.71x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
tests/checkasm/h264pred.c

Change #275797

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9e3d688314d84968e9da6d1d64a351e254defdd0

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred16x16_dc_8
Convert use of MMX to SSE2; no change in performance.

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #275798

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c87b51a93e72ae0b70abc507666b5615c70f6934

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from all pred4x4 functions
Convert all use of MMX to XMM; no change in performance.

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275799

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 04:16:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9e598c5188d5d279eeeb08a128854d70094d8597

Comments

avcodec/x86/h264_intrapred: Deprecate MMX from pred8x8l_down_left_8
Convert use of MMX to SSE2, also remove the SSSE3 impl. since we no
longer use palignr

Before:
  pred8x8l_down_left_8_sse2:               18.0 ( 4.04x)
  pred8x8l_down_left_8_ssse3:              16.2 ( 4.49x)

After:
  pred8x8l_down_left_8_sse2:               14.2 ( 5.14x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libavcodec/x86/h264_intrapred_init.c

Change #275800

Category	ffmpeg
Changed by	wang-bin <wbsecg1ohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 07:30:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a06abd2d7cf32b5dc2e0dd0c4578afcd845a9ba

Comments

hwcontext_vaapi: fix undefined close() on windows

Changed files

libavutil/hwcontext_vaapi.c

Change #275822

Category	ffmpeg
Changed by	Guo Yejun <yejun.guoohnoyoudont@intel.com>
Changed at	Sat 25 Jul 2026 13:39:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d09db42d8f6b1665fb4bdbdd92f78c1e6be908f7

Comments

configure: add manual check for openvino v2

Changed files

configure

Change #275945

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Sat 25 Jul 2026 17:08:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b52c71e43813952650a74c57b2f31bd0166600ba

Comments

avfilter/dnn: fix async teardown race condition in all backends
When the filter graph is torn down early (e.g. at EOF with -frames:v),
the main thread calls dnn_free_model and destroys the request queue
while async inference threads may still be running. When a detached
thread finishes and tries to push its result back to the destroyed
queue, it triggers a segmentation fault.

Add ff_dnn_wait_requests() to dnn_backend_common which blocks
until all allocated request items (ctx->nireq) have been safely
returned to the request_queue. Call it from dnn_free_model in the
torch, tensorflow, openvino, and onnx backends before destroying the queue.

Changed files

libavfilter/dnn/dnn_backend_common.c
libavfilter/dnn/dnn_backend_common.h
libavfilter/dnn/dnn_backend_onnx.c
libavfilter/dnn/dnn_backend_openvino.c
libavfilter/dnn/dnn_backend_tf.c
libavfilter/dnn/dnn_backend_torch.cpp

Change #275946

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Sun 26 Jul 2026 03:05:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	275e217b10057e812039c722d420b3ba996ae69a

Comments

avformat/hlsenc: Fix heap buffer overflow in parse_playlist()
Fix: vulnerability:019f3b84-903b-75fb-a8de-fe6f84d6bc32

When parsing IV=0x... followed by a comma, end - ptr was passed directly
to av_strlcpy() as the destination size. Since iv_string is only 33 bytes
(KEYSIZE*2 + 1), a long IV token could overflow into adjacent heap data.

Use FFMIN(end - ptr + 1, sizeof(buf)) to cap the copy size to the actual
buffer size. Apply the same fix to key_uri parsing for consistency.

Fixes a heap buffer overflow in append_list mode when reading an existing
playlist with a crafted IV or URI token length.

Found-by: depthfirst
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/hlsenc.c

Change #275947

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Sun 26 Jul 2026 03:05:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	23602df9cd1b485c45ba6f533d3b85569de3f323

Comments

avformat/hls: Enforce protocol checks when opening child playlists
Fix: vulnerability:019f3b84-903b-75fb-a8de-fad1c2d7e99c
The parse_playlist() path for opening a new child playlist (in == NULL)
directly called c->ctx->io_open(), bypassing the protocol whitelist and
file-extension checks that open_url() enforces. This allowed child URLs
from a master playlist to use protocols other than file/http/data or to
have disallowed extensions.

Route the !in branch through open_url() so that all child playlists are
subject to the same security policy as segment URLs.

Fixes a security bypass in HLS master playlist parsing.

Found-by: depthfirst
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/hls.c

Change #275948

Category	ffmpeg
Changed by	Steven Liu <lqohnoyoudont@chinaffmpeg.org>
Changed at	Sun 26 Jul 2026 03:05:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	601d9ee881fbd9d9ff44466c561c480ff244eb9f

Comments

avformat/hlsenc: Handle extensionless URIs in extract_segment_number()
fix: vulnerability:019f3b84-90fb-737a-ae7a-b3e0219266fc
extract_segment_number() used strrchr(filename, '.') and immediately did
dot - 1 without checking for NULL. A segment URI with no extension would
cause undefined behavior (NULL pointer arithmetic) when resuming an
append_list playlist with subtitle variants.

Add a NULL check and return -1 early when no dot is found.

Fixes a crash/UB in HLS muxer resume path.

Found-by: depthfirst
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

Changed files

libavformat/hlsenc.c

Change #275972

Category	ffmpeg
Changed by	Nathan Lucas <nlucasgitohnoyoudont@gmail.com>
Changed at	Sun 26 Jul 2026 04:12:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2db563fac2ad9e8509d4c5a699462eead5c32b8c

Comments

avformat/spdifenc: preserve TrueHD MAT padding across branches
Some TrueHD streams, particularly remuxes of seamless branching Blu-rays,
contain input_timing gaps that require more padding than the current
MAT_FRAME_SIZE / 2 limit allows.  spdifenc treats this padding as
invalid and drops it, breaking MAT timing and IEC 61937 carrier cadence,
causing TrueHD/Atmos dropouts on some receivers.

Allow larger valid padding gaps and queue completed MAT buffers so the
muxer can preserve the carrier cadence while still writing at most one
MAT buffer per input packet.

Some branches also have discontinuous timing, so input_timing cannot be
directly used to compute padding at those boundaries.  When output_timing
is available from a TrueHD restart header, use it to detect those
discontinuities and compute padding from the change in
output_timing - input_timing offset across the boundary instead.

Fixes: https://trac.ffmpeg.org/ticket/9569
Fixes: https://trac.ffmpeg.org/ticket/10948
See also: https://github.com/mpv-player/mpv/issues/9659
See also: https://github.com/mpv-player/mpv/issues/13943

Signed-off-by: Nathan Lucas <nlucasgit@gmail.com>

Changed files

libavformat/spdifenc.c

Change #275973

Category	ffmpeg
Changed by	Nathan Lucas <nlucasgitohnoyoudont@gmail.com>
Changed at	Sun 26 Jul 2026 04:12:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7002e01c192e0cd44119cf3e3790c642cb4d0942

Comments

tests/fate/spdif: add TrueHD branch padding test
Add a small synthetic TrueHD sample that exercises large MAT padding in
spdifenc. The sample covers both the input_timing path with padding above
MAT_FRAME_SIZE / 2 and the output_timing discontinuity path.

It also verifies queued MAT frame emission when one input packet completes
more than one MAT frame.

Signed-off-by: Nathan Lucas <nlucasgit@gmail.com>

Changed files

tests/fate/spdif.mak
tests/ref/fate/spdif-truehd-branch-padding

Change #276042

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 27 Jul 2026 05:33:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	705890061467ad550ecc1dad5eea07f28ccfb43e

Comments

avcodec/screenpresso: reject deflate output shorter than the frame
Fixes: use of uninitialized memory
Fixes: screenpresso_short_zlib_heap_disclosure.avi
Fixes: ksUBwBOjJodq
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/screenpresso.c

Change #276073

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:27:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd9d265344b1d39c7b65eb018acac0478c5d736a

Comments

Revert "avfilter/avfiltergraph: always retry format negotiation after auto-filters"
This reverts commit 978a0821eeb909d7b2cb777bed03a8fc380e9503.

The stability of the negotiation process has never been studied
with multiple rounds.

It has always been documented to users that scale was the filter
to insert to manage format conversions. A second filter for
a specific type of conversion should never have been added:
the proper fix for the issue that this commit tried to fix is
to give scale the ability to handle premultiplication too.

Changed files

libavfilter/avfiltergraph.c
tests/fate/filter-video.mak
tests/ref/fate/filter-scale-premultiply

Change #276074

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:27:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f3431169daab3f1ca26638de6d4f6fca0f090c15

Comments

lavfi: do not send premultiplied alpha to filters not ready for it
The arithmetic to process colors in premultiplied alpha is
completely different from the arithmetic for straight alpha.
Running a filter using straight arithmetic on premultiplied
will result in incorrect and sometimes invalid output.
Therefore, premultiplied alpha should only be selected for
filters that either explicitly support it or only use
elementary color arithmetic or none at all.
For other filters, automatic conversion will do its work.

Changed files

libavfilter/formats.c
libavfilter/formats.h

Change #276075

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:27:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a757b708ae7d43fdec89545a55cbc11ae2967b19

Comments

lavfi: cosmetic: rename variable

Changed files

libavfilter/formats.c

Change #276098

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:51:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cf6abefe83e8faa06f8f7e033a67716f1c1f937d

Comments

lavfi/vf_format: report error for invalid names

Changed files

libavfilter/vf_format.c

Change #276099

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:51:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ad1f8a09a66c1126f77b07610f4fbe00f7860504

Comments

lavfi/vf_format: factor parsing the lists

Changed files

libavfilter/vf_format.c

Change #276100

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 10:51:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ef82d0789afe9783cf2aad186addbe81fa30d00a

Comments

fftools/ffmpeg: remove extra \n

Changed files

fftools/ffmpeg_mux_init.c

Change #276116

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 13:38:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	acf6b520c1e73707190c0a591735ee8c7fbe8f58

Comments

fftools/ffmpeg_sched: do not print error code
The error codes are meaningless to users, especially
the ones that do not come from standard errno codes.

Changed files

fftools/ffmpeg_sched.c

Change #276146

Category	ffmpeg
Changed by	Nicolas George <georgeohnoyoudont@nsup.org>
Changed at	Mon 27 Jul 2026 16:07:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	336b175c9f5d1194f4377f34aacfe4a85ca628b6

Comments

fate: disable tests that depend on premultipled conversion
They can be re-enabled onece automatic conversion of
premultiplied alpha is fixed.

Changed files

tests/fate/image.mak
tests/fate/lavf-image.mak

Change #276149

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c6e2ae51073120b063a46ec49748bfb3534afeb3

Comments

avformat: add parent AVFormatContext to URLContext
Add a URLContext.fmt_ctx back-reference, set by io_open_default before
url_open() runs. Will be used to access shared resources per avfc.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/avio.c
libavformat/avio_internal.h
libavformat/internal.h
libavformat/options.c
libavformat/url.h

Change #276150

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dd2554331668097f59e4a47a353dd16b4ae90fe9

Comments

avformat/libcurl: add libcurl protocol skeleton
configure changes, and protocol with ENOSYS impl.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

configure
libavformat/Makefile
libavformat/libcurl.c
libavformat/protocols.c

Change #276151

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	15e07fbcd1739acffbcd9be21df29d2901b9df1e

Comments

avformat/libcurl: implement event loop and read path
Add the per-AVFormatContext curl_multi worker thread, the command queue,
write/header/xferinfo callbacks and a FIFO with pause/unpause
backpressure. url_open2 probes the response on a dedicated thread.
url_read drains the FIFO and yields to the avio layer so the interrupt
callback is honoured. An explicit "libcurl:" URL prefix forces the
protocol.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

configure
libavformat/avformat.c
libavformat/internal.h
libavformat/libcurl.c

Change #276152

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	363f04fee9ef7fa0699d6d484966719c44703f2d

Comments

avformat/libcurl: add seek, size and retry
Probe Accept-Ranges/Content-Range/Content-Encoding to determine
seekability and total size, issue ranged requests via CURLOPT_RANGE,
implement url_seek (including AVSEEK_SIZE) as a reconnect at the new
offset, and resume seekable transfers from the next missing byte after a
recoverable error.

Note that we prefer the compression over seekability, as servers are
unlikely to compress media files in practice, as this would be huge
performance cost for zero gain. However downloading text playlists
benefits a lot from compression. YouTube for example can serve text
files that have 30MB+ which is huge bandwidth usage when not compressed.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276153

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d0972a725d1f91f069a5e044d220fe45a573bb3e

Comments

avformat/libcurl: add options, TLS, proxy, cookies and headers
Add the AVOption table and map it onto libcurl. Verbose curl output is
routed to av_log at debug level. This mirrors subset of http.c options.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

doc/protocols.texi
libavformat/libcurl.c

Change #276154

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7346f917e4219888ffdefc80cc5b5803b4d939a3

Comments

avformat: add prefer_libcurl option to route http(s) via libcurl
Register a prefer_libcurl IO option that, when set, opens http(s) URLs
with the libcurl protocol instead of the native one. The native http
protocol stays the default.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

doc/protocols.texi
libavformat/avio.c
libavformat/url.h

Change #276155

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0a3cf17d5ff18cd6a69e62985f88e48b572fef88

Comments

avformat/libcurl: set minimum buffer size to curl write size
Otherwise, there will never be enough space in the FIFO to put the data
written, resulting in the write callback being stuck/paused indefinitely.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276157

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a910a30326fdcaeef1ed6b1ba4ea56d935200b6d

Comments

avformat/libcurl: also allow forcing streams to be seekable
The documentation states this as a valid option, but the code only allows
force-disabling atm.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276158

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4127eb2d6ab0008b4c7a347fffef42b467c88f5b

Comments

avformat/libcurl: avoid TOCTOU in libcurl_seek()
If this triggers while an asynchronous header_callback() (e.g. from a previous
seek or redirect) is still running, the seekable/content_size fields can
be written to while libcurl_seek() is still reading from them.

I'm not convinced this is an actual bug in practice, but tsan at least is
bound to complain about it.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276159

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6c338a659551c8f395954698f8e0c8607c3e3f4c

Comments

avformat/libcurl: attach a CURLSH shared handle to all requests
This allows sharing cookies, HSTS state etc. between requests inside a single
AVFormatContext. This is analogous to the status quo of http.c, which also
re-uses cookies across all requests.

Note that the connection state, DNS cache, SSL session etc. are shared by
default when using a multi handle, so we only need to explicitly share the
rest.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276160

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c70d777f7fb1023adba169dd2e037b4900e96af7

Comments

avformat/libcurl: enable cookie engine unconditionally
With the previous commit, this allows cookies to persist across requests.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276161

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e11cec1059cd2e183be2ce0b7ab4b29a173667f7

Comments

avformat/libcurl: respect format whitelist
And also set a reasonable default whitelist. This is especially important
because libcurl supports a much wider range of protocols than FFmpeg,
including e.g. smtps, telnet, gopher, etc.

We have to include "libcurl" in libcurl's own default whitelist, otherwise
it will refuse te recursively open itself for e.g. HLS.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276162

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b737bddaf197be1cd6b846b4d8810fb20a2366d0

Comments

avformat/avio: re-use av_opt_set() to parse prefer_libcurl
Otherwise, this fails to parse values like "true" or "yes". This does
require stack-allocating a dummy object, but that's not a big deal, as
we only ever set this single option on it, so nothing should ever
get allocated.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/avio.c

Change #276163

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	45c4f59eb2da4a084fd2cda52ec362e6185839b8

Comments

avformat/aviobuf: add prefer_libcurl to ffurl_copy_url_options()
Otherwise this doesn't reach e.g. the inner HTTP requets made by HLS etc.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/aviobuf.c

Change #276164

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b5bf0a1a87c4fc572a899234d63e1db3e2b0dafa

Comments

avformat/hls: check if http protocol is native or not
When using -prefer_libcurl, http might be served by libcurl instead of the
native HTTP backend; in this case we shouldn't use `ff_http_do_new_request2`
and instead create a new request.

I added a few assertions to safeguard some of the assumptions this code
clearly relies on.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/hls.c

Change #276165

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	835ebef5351dd7bd60ce59070c24099678c93021

Comments

avformat/libcurl: keep track of owning AVFormatContext
For avfc-attached curl loops.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276166

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cee0613cf7787a9dce9a256027dd4484deea21e6

Comments

avformat/libcurl: track and report connection/request statistics
In theory, we could expand this a lot more, by adding more verbose
statistics about individual request latency etc, but I don't think that's
as helpful (to me) as at least tracking the number of requests made.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276167

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d993f37a1bb8ac6ba4094b82767c596c031edc49

Comments

avformat/avio: propagate prefer_libcurl preference from parent URLContext
Since this is consumed by e.g. protocols which open inner protocols.
This would normally be propagated by the `av_opt_copy` call, but as this
check happens *before* the inner URLContext is even allocated, we're
forced to propagate the options manually.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/avio.c

Change #276168

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	90a08d3657654603d45b54d9f4a6b71846578c76

Comments

avformat/libcurl: early exit on no-op seek
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276169

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	34015e179d947d27475defe2dea46457698f9bbd

Comments

avformat/libcurl: fully parse Content-Range header
Logic copied from http.c

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276170

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d754fb37a572148830625f7a4d8006c3836822b1

Comments

avformat/libcurl: simplify continuation decision logic
Instead of trying to statically guess whether or not we expect more data
to be available by exactly recreating the logic, we can consult the parsed
content-range header directly to determine if there are more bytes in the
file.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276171

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	946a450cecaededc1a44d779176684e3412ceb71

Comments

avformat/libcurl: implement -initial_request_size
See corresponding feature in http.c.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

doc/protocols.texi
libavformat/libcurl.c

Change #276172

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7530aa12ee9e0ee33d5dfc2336411acae378ebf1

Comments

avformat/libcurl: avoid integer overflow in connect_timeout
This is multiplied by 1000, which can overflow on systems with 32-bit
long (i.e. most of them).

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276173

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	615032302929256507093419f8637d3f81140367

Comments

avformat/libcurl: use int64_t consistently
Instead of randomly mixing in uint64_t. As far as I can tell, we don't
depend on defined overflow behavior anywhere, except where such would be a
bug anyways, and I'm also pretty sure we won't exhaust the 63-bit file size
limit of int64_t any time soon, so this is just a footgun prevention
mechanism.

We already have a few dodgy places of mixing int64_t and uint64_t values
in the code haphazardly.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276174

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a6514b7c4bbc67b1458caf306e742a121088976

Comments

avformat/libcurl: harden against malicious underflow
If the server sends a value of INT64_MIN here, we would correctly parse
it and then subtract off -1, underflowing to a huge positive value
(and triggering undefined behavior in the process).

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276175

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fea5d3ccc3b13bd60bdd4316309856b8043aec7c

Comments

avformat/libcurl: guard against overflow on seek()
Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276176

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	337b884a4ea56a84e269a40154328f99731f0d9c

Comments

avformat/libcurl: guard against integer overflow on start_request()
The failure mode here is to just request the rest of the file, which is
explicitly safe.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276177

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	488061b9cb2e29a22b64962b64e2424b732d10a8

Comments

avformat/libcurl: guard against overflow from over-reading data
In theory a client could be coaxed by a malicious server to seek to a
large 64-bit offset, which would then trigger an overflow of the position.

Better safe than sorry and just error out with EIO in such cases.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276178

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cbed9900216d4885e089a59eedd6893104c4d6bc

Comments

avformat/libcurl: guard against unexpected reply content-range
Harden against malicious or badly misbehaving servers.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/libcurl.c

Change #276179

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	723a855e52cabe7784193b7cd63772bd8916ee55

Comments

avformat/libcurl: validate that the server honored a requested start offset
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276180

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aec0c7a0cb4f5dae75c39db56c1a33b1f469668f

Comments

avformat/libcurl: only offer compression for full unbounded requests
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276181

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	be6b6743b81109b865964325b1ffce88c92888fb

Comments

avformat/libcurl: improve content size tracking
Track the size for non-seekable replies too, don't reset a known size when
a reply omits it, and don't store the encoded length of a compressed body.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276182

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c2b0152ad3468ffe35ecd6eb63afdee30e5d15c5

Comments

avformat/libcurl: keep the seekable user override across replies
It was applied once after the probe, from the wrong thread, and any
follow-up reply re-evaluating seekability discarded it.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276183

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	798fb1f9aae45bcdde0496950a7fdc7625585d22

Comments

avformat/libcurl: clear the pause flag when unpausing
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276184

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6d626c1a3c104602fa9ea6ea0d8db0b669ec8f78

Comments

avformat/libcurl: fix AVBPrint leak on setup_protocols() error paths
The API is not clear to me if AVBPrint has to be finalized on errors,
but there is nothing wrong to be defensive here

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276185

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d097f9385fb91b36b82a39033a4ca0ce739f054c

Comments

avformat/libcurl: limit buffer_size to INT_MAX
Larger values silently truncate on 32-bit, which can make the resulting
FIFO too small to ever satisfy the unpause threshold.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276186

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 27 Jul 2026 19:05:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	298f67fa2a8ec97af323e92c76a65b5f3552ed7a

Comments

avformat/libcurl: fix comments
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/libcurl.c

Change #276212

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 27 Jul 2026 20:13:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7e1f4dc31864b4bde2b43ed80aedaa2b6600ff81

Comments

Reapply "avfilter/avfiltergraph: always retry format negotiation after auto-filters"
This reverts commit cd9d265344b1d39c7b65eb018acac0478c5d736a.

Changed files

libavfilter/avfiltergraph.c
tests/fate/filter-video.mak
tests/ref/fate/filter-scale-premultiply

Change #276213

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 27 Jul 2026 20:32:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	80f14b251b54d259b18b0ba55ca06d23da2c20f0

Comments

Revert "lavfi: do not send premultiplied alpha to filters not ready for it"
This reverts commit f3431169daab3f1ca26638de6d4f6fca0f090c15, which
introduced various regressions and hangs with at least exr, as
reported by fate, and potentially other codecs that are not as
thouroughly tested.

Changed files

libavfilter/formats.c
libavfilter/formats.h

Change #276214

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 27 Jul 2026 20:34:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2be4a694a3893d2ec741ab8c51c2217f95411273

Comments

Revert "fate: disable tests that depend on premultipled conversion"
This reverts commit 336b175c9f5d1194f4377f34aacfe4a85ca628b6.

Changed files

tests/fate/image.mak
tests/fate/lavf-image.mak

Change #276215

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 28 Jul 2026 00:53:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	42970ac7e2a18334e1cc645f6a39194778c4e35a

Comments

tests/checkasm/checkasm: Map rnd to checkasm_rand_uint32
Before the switch to libcheckasm, rnd returned an unsigned value;
now it only returns a signed value in the range 0..INT_MAX,
so that the sign bit is not random at all. This means that when
rnd is used to initialize four bytes at a time, one bit is not random.
Fix this by mapping rnd to checkasm_rand_uint32().

Reviewed-by: Niklas Haas <ffmpeg@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/checkasm.h

Change #276216

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 28 Jul 2026 00:54:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c6309b5c63add7ad0ec221fafefc32bdcd6f8b91

Comments

tests/checkasm/h264pred: Avoid declare_func_emms
This is possible after the recent MMX->XMM conversions
and preferable as it makes the test stricter.

Reviewed-by: Niklas Haas <ffmpeg@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/h264pred.c

Change #276235

Category	ffmpeg
Changed by	Christopher Decker <chris.decker08ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 06:59:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	94ff335d762ad08f2557a320def57f287d74e2a5

Comments

avformat/os_support: fix return value of win32_rename
The return value of MoveFileExW was not being correctly interpreted,
see https://learn.microsoft.com/en-us/windows/win32/api/winbase/nf-winbase-movefileexw.

On Windows a failed rename over the file: protocol now surfaces as failed to rename file %s to %s:
Operation not permitted (ff_rename, libavformat/avio.c:867) plus an AVERROR(EPERM) return, where previously the muxer
reported success and the user was left with a missing or stale output file and no diagnostic. This affects the
write-to-temp-then-rename paths in hlsenc, dashenc, hdsenc, smoothstreamingenc, segment, and img2enc (e.g. HLS/DASH
playlist updates, -write_temp_file, segment list finalization).

Also, added unit tests to exercise the rename function.

Signed-off-by: Christopher Decker <chris.decker08@gmail.com>

Changed files

libavformat/Makefile
libavformat/os_support.h
libavformat/tests/rename.c
tests/fate/libavformat.mak

Change #276245

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 10:39:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5dbf1dccadd51718a4553879d1eaf6fd9ee962b0

Comments

avcodec/dxva2: map 4:2:0 12-bit to P012 surfaces
Select P012 as the D3D11VA/DXVA2 frames sw_format for YUV420P12 content
and map it to DXGI_FORMAT_P016, so 12-bit 4:2:0 hardware decoding can
allocate a usable surface.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/dxva2.c

Change #276247

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 10:39:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	04d2790ab75127216f00316f3c8a15ff20ffc0eb

Comments

avcodec/dxva2: add AV1 Profile 2 decoder GUIDs
Add the AV1 Professional (Profile 2) decoder GUIDs, including the
dedicated 12-bit 4:2:0 mode, so D3D11VA/DXVA2 can match a decoder for
Profile 2 streams.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/dxva2.c

Change #276248

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 10:39:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	099aa719021a09d2373dec6cadb6b3d4c280095d

Comments

avcodec/av1dec: allow D3D11VA for 12-bit 4:2:0
Offer the D3D11VA hwaccel pixel formats for YUV420P12 (AV1 Profile 2),
enabling hardware decoding of 12-bit 4:2:0 streams on capable devices.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/av1dec.c

Change #276249

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 10:39:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fe953596e9f53e3d61c465bce7a29834cae3375b

Comments

avcodec/d3d12va: add AV1 Profile 2 support
Accept the AV1 Professional profile in the D3D12VA hwaccel and select the
matching decode profile GUID, offer D3D12 for YUV420P12 in the decoder,
map the frames context to P012, and add P016 (12-bit) support to the
d3d12va hwcontext.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/av1dec.c
libavcodec/d3d12va_av1.c
libavcodec/d3d12va_decode.c
libavutil/hwcontext_d3d12va.c

Change #276299

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 17:39:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cac354352929816748f7fcb99ccf2e1740d7955f

Comments

avcodec/codec_desc: sync long_name with pcm-dvd.c
It was outdated for a long time.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/codec_desc.c

Change #276300

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 17:39:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	88c8100e22d538ea318a505efa93c9f2d353249f

Comments

avcodec/pcm-dvd: rename long codec name to DVD-Video
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/pcm-dvd.c
libavcodec/pcm-dvdenc.c

Change #276316

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 17:54:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	047eb18451180c5b4933050a4453274d81ae4127

Comments

avcodec: add PCM DVD-Audio decoder
Decodes the LPCM variant found in DVD-Audio AOB streams. Unlike
DVD-Video LPCM it supports up to 192 kHz sample rates and splits
channels in two channel groups, which may use different quantization.
Samples are grouped in sets of 2 samples over all channels, with the
second channel group's data stored first within each set, as 16-bit
big-endian most significant parts followed by the remaining bits.

Based on information from expired US 6,580,671 patent et al.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

Changelog
doc/APIchanges
doc/decoders.texi
doc/general_contents.texi
libavcodec/Makefile
libavcodec/allcodecs.c
libavcodec/codec_desc.c
libavcodec/codec_id.h
libavcodec/pcm-dvda.c
libavcodec/version.c
libavcodec/version.h

Change #276317

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 17:55:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ff818b87bf1cfee7a10196e90d0df8de44e9a771

Comments

avformat/mpeg: fix demuxing of DVD-Audio LPCM
Substream 0xa0 packets without the DVD-Video dynamic range control
marker byte were assumed to be MLP, while in DVD-Audio AOBs substream
0xa0 always carries LPCM (MLP uses 0xa1), so hi-res LPCM streams were
misdetected and decoded as garbage. Classify them as PCM_DVDA and keep
the private stream header in the packet for the decoder to parse, as
its length is variable and it carries the audio format.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/mpeg.c

Change #276318

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 17:55:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b449f86287f173802abadcf3fa3e37f88d9fb685

Comments

tests/fate/pcm: add fate-pcm_dvda test
Input is cut down version of https://samples.mplayerhq.hu/DVD-Audio/ats.AOB

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tests/fate/pcm.mak
tests/ref/fate/pcm_dvda

Change #276319

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e822dd253b281e24a6a25cfe422ba5323ae9b961

Comments

avformat/hls: remove else after break
To reduce unnecessary nesting.

Changed files

libavformat/hls.c

Change #276320

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2526ac4771d6fff24506e5871d842771e2bbd59c

Comments

avformat/hls: use av_rescale_q instead of doing only denominator

Changed files

libavformat/hls.c

Change #276321

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ede944b3622dd2bcd414f6e522ce7fbeb5f7974a

Comments

avformat/hls: allow seeking within live playlists
Live (and EVENT) playlists are seekable within the currently known
segment list. Seeks before the list start clamp to the oldest segment,
past the end to the live edge. Live playlist can evict old segments, we
track that by reading EXT-X-MEDIA-SEQUENCE. Previusly the code assumed
that the live playlist, don't have past segments, however some servers
has very deep usable timeshift.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c

Change #276322

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	84bc57637130c320a8bdbdce9bd438b8210fe980

Comments

avformat/hls: set stream start_time for all playlist types
Rebase the first packet's timestamp onto the start of the segment list
regardless of playlist type, and report it as start_time. Previously
only EVENT playlists did this and live streams took start_time from the
first loaded segment at the live edge. Which is not correct as some
servers provide very deep (hours even) time shift in live playlist. The
evicted segments are already tracked by EXT-X-MEDIA-SEQUENCE. This
allows us to use all available segments, not only the future ones on
live playlist.

Also prefer PTS over DTS, as EXTINF durations and start_time are
presentation time.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c
tests/fate/hlsenc.mak
tests/ref/seek/hls

Change #276323

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	848fb754dd8ac26e1fa42a94d1034af3eabcea0d

Comments

avformat/hls: don't discard target segment's first keyframe when seeking
Segment starts on the seek timeline are derived from EXTINF durations,
so the first keyframe of the target segment can have a slightly lower
DTS than the snapped seek position, in which case the packet discard
loop would skip it and resume playback one segment late. Make sure we
start at the very beginning of the stream in this case.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c
tests/ref/seek/hls

Change #276324

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 28 Jul 2026 18:01:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f6fa0d3fdafc8cd8224c5fe9dc974bd5ad2c9a5e

Comments

avformat/hls: refine first_timestamp from the first segment's packets
first_timestamp is taken from the first packet in mux order, which is
not necessarily the earliest timestamp in the segment. Lower it while
packets of that segment are still arriving, so the seek timeline is
anchored at the true segment start.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c

Change #276330

Category	ffmpeg
Changed by	Kenan Alghythee <kalghy2ohnoyoudont@uic.edu>
Changed at	Tue 28 Jul 2026 20:04:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e2bc6d88cd68184288c16570e04881c3a426a05a

Comments

avcodec/cbs_av1: pad the ITU-T T.35 payload buffer
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavcodec/cbs_av1_syntax_template.c

Change #276373

Category	ffmpeg
Changed by	Kenan Alghythee <kalghy2ohnoyoudont@uic.edu>
Changed at	Wed 29 Jul 2026 02:42:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	06fd4fd0e787e819fd697dc687efe7f7eae798c8

Comments

avfilter/vf_xpsnr: avoid a zero block size on small frames
Fixes: division by zero
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavfilter/vf_xpsnr.c

Change #276375

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 29 Jul 2026 04:03:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2b7e5012424a52998cd6a1fe3556272313cb7527

Comments

avformat/codec2: avoid integer overflow in packet size and duration
Fixes: signed integer overflow
Fixes: 2jy_poc_codec2.zip / poc_codec2.raw
Fixes: jOQASNnOm6O7
Found-by: Jiale Yao <yaojiale02@163.com>

Changed files

libavformat/codec2.c

Change #276376

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 29 Jul 2026 06:37:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f3b13867076d616de36d7d8c50298c113b1c62ba

Comments

avcodec/videotoolboxenc: check PTS validity before timestamp conversion

Changed files

libavcodec/videotoolboxenc.c

Change #276377

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 29 Jul 2026 06:37:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6f82db5ad5a10a6b49d1460a7ed4acace5e2c6a9

Comments

avcodec/videotoolboxenc: respect CMTime.timescale when computing timestamps
vtenc_cm_to_avpacket converts output CMSampleBuffer timestamps back
to codec timebase by dividing pts.value by avctx->time_base.num.
This relies on the output CMTime having the same timescale as the
one passed to VTCompressionSessionEncodeFrame. Apple does not
guarantee this, and the output timescale can differ from the input.

Use av_rescale_q with the actual CMTime timescale so that the
conversion is correct regardless of what timescale VideoToolbox
returns.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/videotoolboxenc.c

Change #276381

Category	ffmpeg
Changed by	Julius Bairaktaris <juliusohnoyoudont@bairaktaris.de>
Changed at	Wed 29 Jul 2026 10:52:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d43b1efd2e948f44cfac91f7a4325a3d927d6718

Comments

avfilter/vf_amf_common: fix component Init() format with hwaccel input frames
amf_init_filter_config() returned the hardware pixel format (e.g.
AV_PIX_FMT_D3D11) as in_format when the input link carries D3D11VA or
DXVA2 hw frames. av_av_to_amf_format() maps hw formats to
AMF_SURFACE_UNKNOWN, so every AMF filter failed its component Init()
with AMF_INVALID_ARG (VQEnhancer, FRC) or AMF_NOT_SUPPORTED
(Converter, HQScaler). Return the underlying software format instead,
which is already validated against AMF a few lines above. Apply the
same mapping to the default output sw format, which otherwise ends up
as the hw pixel format in hwframes_out->sw_format when the output link
uses a hwaccel format.

Fixes e.g.:
  ffmpeg -hwaccel d3d11va -hwaccel_output_format d3d11 -i in.mp4          -vf frc_amf -c:v hevc_amf out.mp4

Note: with D3D11VA input, the compute-based components (VQEnhancer,
HQScaler, VideoConverter) additionally require decode textures created
with D3D11_BIND_SHADER_RESOURCE, e.g.:
  -init_hw_device d3d11va=dx11:,SHADER=1 -hwaccel_device dx11
FRC and DXVA2 input work with default decode textures.

Changed files

libavfilter/vf_amf_common.c

Change #276439

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Wed 29 Jul 2026 14:37:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a234fc1130e6b17d550371dede849918202a303a

Comments

avfilter/dnn: prevent crash on parameterless LibTorch models
When loading a TorchScript model that does not contain any learnable
parameters (e.g., a purely functional model), the Torch backend would
crash during inference. This occurred because the code attempted to
dereference the first iterator of the model's parameter list
`parameters().begin()` to determine the device, which results in
Undefined Behavior when the parameter list is empty.
This commit fixes the issue by determining the inference device directly
from the user-configured `ctx->device` string instead of probing the
model parameters, allowing parameterless models to execute safely.

Testing:
1. Generate a parameterless model:
cat << 'EOF' > generate_model.py
import torch
class DummyModel(torch.nn.Module):
    def forward(self, x):
        return x
scripted_model = torch.jit.script(DummyModel())
scripted_model.save("dummy_model.pt")
EOF
python3 generate_model.py
2. Run inference (previously crashed, now succeeds):
./ffmpeg -y -i input.mp4 -vf 'format=rgb24,dnn_processing=dnn_backend=torch:model=dummy_model.pt' -frames:v 5 -f null -
Signed-off-by: Raja Rathour <imraja729@gmail.com>

Changed files

libavfilter/dnn/dnn_backend_torch.cpp

Change #276475

Category	ffmpeg
Changed by	Omkhar Arasaratnam <omkharohnoyoudont@linkedin.com>
Changed at	Wed 29 Jul 2026 22:37:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	83c692282631e6d035fa52d2e2c5118b1d2a27d7

Comments

avformat/tls_openssl: bind peer identity for numeric-IP verify
tls_open() installed a peer-identity target only when the URL host was
non-numeric. For a numeric-IP target the whole block was skipped, so
with verify=1 OpenSSL validated the chain to a trusted CA but bound no
identity, and any publicly-trusted certificate for any name was accepted
(e.g. tls://203.0.113.10?verify=1).

Run the identity block for every non-listening connection. Classify
s->host with the same getaddrinfo(AI_NUMERICHOST) rule tls.c uses. A
numeric host is pinned to the certificate's iPAddress SAN by handing
OpenSSL the parsed binary address via X509_VERIFY_PARAM_set1_ip();
everything else, including a verifyhost=<name> override, binds by name
via SSL_set1_host(). Passing the binary address rather than re-probing
the ASCII form pins legacy numeric spellings such as 2130706433 as IPs
instead of letting them fall back to hostname matching. SNI is still
suppressed for numeric transport hosts (RFC 6066 sec. 3) via
s->numerichost.

The identity target is installed unconditionally for non-listening
connections; whether the certificate is actually verified stays
controlled separately by SSL_CTX_set_verify() under s->verify, so
disabling verification keeps the prior behaviour. Sets AVERROR_EXTERNAL
with an explicit log line on failure.

dtls_open() sets s->is_dtls and then calls tls_open(), so DTLS client
connections run through the same identity block and are covered here.

Found-by: Claude (Anthropic). Human-verified and reported by
Omkhar Arasaratnam <omkhar@linkedin.com>.
Signed-off-by: Omkhar Arasaratnam <omkhar@linkedin.com>

Changed files

libavformat/tls_openssl.c

Change #276479

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 29 Jul 2026 23:07:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	93f2a525ec6c7b467bae68322720d10188fc6e30

Comments

avcodec/dvbsub_parser: avoid signed overflow in the capacity check
Fixes: signed integer overflow
Fixes: out of array access
Fixes: poc.wtv
Fixes: fJeEU9JwKwsR
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/dvbsub_parser.c

Change #276480

Category	ffmpeg
Changed by	almogyalin <almog.yalinohnoyoudont@gmail.com>
Changed at	Wed 29 Jul 2026 23:08:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	04ecca792fe8c4bfb9a8a2f580e6f3740258feee

Comments

avformat/iamf_parse: fix inverted subblock duration validation
The check on variable subblock durations had an inverted comparison, rejecting valid multi-subblock parameters and failing to catch subblock durations exceeding the remaining time.

Changed files

libavformat/iamf_parse.c

Change #276481

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 29 Jul 2026 23:20:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd1f545cf27ba08f6f5b31b1e92665d7874d4fd7

Comments

avcodec/rscc: do not leave uninitilized data when the input is too short
Fixes: use of uninitialized memory
Fixes: rscc_short_deflate_heap_disclosure.avi
Fixes: plB80py3i3Bu
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/rscc.c

Change #276501

Category	ffmpeg
Changed by	panboxiaosa <panboxiaosaohnoyoudont@gmail.com>
Changed at	Thu 30 Jul 2026 01:49:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b57a3d028dbefc95f5f4bb70f2b2f5f8debb9a0c

Comments

avformat/mov: set video packet durations in presentation order
STTS sample deltas follow decode order, while AVPacket.duration is
defined as the interval to the next PTS in presentation order. Assigning
the deltas directly therefore produces incorrect packet durations for
VFR video with reordered frames.

After index construction and edit-list processing, sort samples by PTS
and set every duration with a following PTS from adjacent presentation
timestamps. This reorders STTS durations where possible and derives
intervals produced by the STTS/CTTS combination when necessary. Keep the
original timing table if allocation fails or timestamps are invalid.

Add FATE coverage for the official VFR H.264 sample and for a generated
MPEG-4 case whose presentation intervals are not a permutation of its
STTS deltas. Update the HEVC dts2pts CRA reference for the corrected
presentation-order durations.

Signed-off-by: panboxiaosa <panboxiaosa@gmail.com>

Changed files

libavformat/mov.c
tests/fate/mov.mak
tests/ref/fate/hevc-bsf-dts2pts-cra
tests/ref/fate/mov-vfr-bframes-derived-duration
tests/ref/fate/mov-vfr-bframes-duration

Change #276502

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 30 Jul 2026 02:12:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2f234ea34c81288e3840fca632dd16481d8de39f

Comments

avcodec/tiff: reject inflate output shorter than the strip
Fixes: use of uninitialized memory
Fixes: tiff_short_deflate_heap_disclosure.tiff
Fixes: 1cRIkpUVMQtn
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/tiff.c

Change #276503

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 30 Jul 2026 02:13:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a100d346dafd012bfc0c20098a764c6d01a05714

Comments

avformat/mov: bound sgpd sync entry_count by the atom size
Fixes: unbounded allocation
Fixes: ffmpeg-sgpd-sync-entry-count-alloc.mp4
Fixes: 4f7OaHqbUKh2
Found-by: Brian Lee, Youngjoon Kim (Georgia Tech SSLab)

Changed files

libavformat/mov.c

Change #276519

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Thu 30 Jul 2026 06:47:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	05be98e3a2ea5c8dfc102860bd96a12a84b93346

Comments

doc/nut.texi: point at the latest spec in the git repository
subversion in 2026 is a bit odd

Changed files

doc/nut.texi

Change #276537

Category	ffmpeg
Changed by	wangnov <wangnov6ohnoyoudont@gmail.com>
Changed at	Thu 30 Jul 2026 08:51:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2ae24134889e3684327196a2e6354c36d967051a

Comments

avcodec/vp8, vp9: warn when native decoders ignore WebM alpha
WebM VP8/VP9 alpha is a second bitstream attached via a Matroska
BlockAdditional element (BlockAddID 1). Only the libvpx wrapper
decoders merge it into the output frame; the native vp8/vp9 decoders
have no code path for it and silently return an opaque frame.

Warn once per decoder instance instead of failing silently, pointing
at the decoder that does support it. The av_log_once() state is
synchronized across frame-thread contexts to avoid duplicate warnings
with frame threading.

Reported in https://trac.ffmpeg.org/ticket/11165 and
https://github.com/renpy/renpy/issues/1402.

Signed-off-by: wangnov <wangnov6@gmail.com>

Changed files

libavcodec/vp8.c
libavcodec/vp8.h
libavcodec/vp9.c
libavcodec/vp9dec.h

Change #276592

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 30 Jul 2026 15:19:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c77c0e580a39580c085b7182885601015b7f3da4

Comments

avcodec/bsf/eac3_core: clear profile value on init()
The output stream no longer has Atmos metadata, so the
AV_PROFILE_EAC3_DDP_ATMOS profile, if it was set, is no longer valid.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/bsf/eac3_core.c

Change #276593

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 30 Jul 2026 15:19:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	86940d45aff7d59810794df3ab2b39b7b83b478c

Comments

avcodec/bsf/truehd_core: clear profile value on init()
The output stream no longer has Atmos metadata, so the
AV_PROFILE_TRUEHD_ATMOS profile, if it was set, is no longer valid.

Fixes issue #23195.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/bsf/truehd_core.c

Change #276604

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 30 Jul 2026 23:31:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c6bd6abb59926385a454a4485f9c28d08a20f775

Comments

avcodec/x86/h264_intrapred: Report accurate number of registers used
The pred{8x8,16x16}_plane functions only use five xmm register.
The inflated number of seven (which leads to saving and restoring
xmm6 on Win64) has been introduced in
d20f133ef962da71326bc3635e086696f45ab64e, probably because
the mmx version of these functions used seven mmx register.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #276605

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 30 Jul 2026 23:31:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	27a90f6ad4c62f1bbbc6cf74e78f4a5e0c9ccd37

Comments

avcodec/x86/h264_intrapred: Reduce number of regs in pred4x4_tm_vp8_8
Avoids saving and restoring volatile xmm registers on Windows;
no change in benchmarks here (on Unix64).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #276606

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 30 Jul 2026 23:31:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5f832b7b3fd4cb8daf9bb64373230809c246bd84

Comments

avcodec/x86/h264_intrapred: Avoid shuffles
We only need the shuffle a byte in the lower quadword;
also notice that we don't need the clamping provided
by packuswb as these values are averages of byte values.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #276609

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 06:00:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e38b5d15bd93585f96c75e31e2dd1e1fe9294d05

Comments

avfilter/af_arnndn: pad the DCT input buffers to the read length
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavfilter/af_arnndn.c

Change #276610

Category	ffmpeg
Changed by	Kenan Alghythee <kalghy2ohnoyoudont@uic.edu>
Changed at	Fri 31 Jul 2026 06:00:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	534f16d866c732a85c34ac576d66d578669578f1

Comments

avcodec/bsf/dovi_rpu: handle update_rpu() returning no RPU
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavcodec/bsf/dovi_rpu.c

Change #276611

Category	ffmpeg
Changed by	Kenan Alghythee <kalghy2ohnoyoudont@uic.edu>
Changed at	Fri 31 Jul 2026 06:00:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	372a611913e6e2ba3d0fe859a00934f974f7d003

Comments

avcodec/dovi_rpuenc: validate vdr_rpu_id from the input metadata
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavcodec/dovi_rpuenc.c

Change #276612

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 06:00:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	79e10e51960f848b82c597c540087bcc66aa6e17

Comments

avcodec/dovi_rpudec: bound num_x/y_partitions
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavcodec/dovi_rpudec.c

Change #276613

Category	ffmpeg
Changed by	Kenan Alghythee <kalghy2ohnoyoudont@uic.edu>
Changed at	Fri 31 Jul 2026 06:00:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f1ed338541d7596f5cd7721288049745ec5265f5

Comments

avcodec/dovi_rpuenc: validate the data mapping before generation
Fixes: out of array access
Fixes: SUcVEyk7r3Gc
Found-by: Kenan Alghythee <kalghy2@uic.edu>

Changed files

libavcodec/dovi_rpuenc.c

Change #276614

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 06:00:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2c18311d59ac8b1a4ca2255750836b892179d27a

Comments

avcodec/dovi_rpuenc: normalize vdr_dm_metadata_present to 0/1

Changed files

libavcodec/dovi_rpuenc.c

Change #276615

Category	ffmpeg
Changed by	Abdessamie <try70457ohnoyoudont@gmail.com>
Changed at	Fri 31 Jul 2026 06:01:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a441a2eb383960d76632eb5dc42639ec52d46bd8

Comments

avformat/rtsp: clear authentication on cross-origin redirects
RTSP redirects retain URL credentials and authentication state when a
Location URI changes to another origin. This can forward reusable
credentials to a server with a different scheme, host, or port.

Clear the stored credentials and authentication state when the redirect
crosses an origin boundary. Preserve them for same-origin redirects.

Fixes: cross-origin credential disclosure
Fixes: rtsp_redirect_auth_leak_poc.py
Fixes: VaKaPOnfN02z

Changed files

libavformat/rtsp.c

Change #276616

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Fri 31 Jul 2026 07:41:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	43efff09ee84c95341a945097e467dcbbf9dbbe3

Comments

libavcodec/x86/h264_intrapred: xmmN -> mN, movdqa -> mova
Use m aliases instead of xmm registers directly so that certain macros
can work (e.g. SWAP); likewise replace movdqa with mova

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #276617

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Fri 31 Jul 2026 07:41:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a83c6ad94619ab8f7f4f877463c9532a4f10bb78

Comments

libavcodec/x86/h264_intrapred: Base-4 numbers for shuffling immediates
0xee -> q3232 for example for better readability

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred.asm

Change #276618

Category	ffmpeg
Changed by	shadowcaster3 <vasyl.samoilovohnoyoudont@gmail.com>
Changed at	Fri 31 Jul 2026 08:25:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ad53728984531cebdb027e70e78d7165a6bdfe20

Comments

avformat/mpegtsenc: add mpegts_pcr_pid option for separate PCR PID
Add a new muxer option mpegts_pcr_pid that generates PCR on a dedicated
adaptation-field-only PID, separate from any elementary stream.

ISDB-T 1seg receivers require PCR on a standalone PID. When PCR shares
the video PID, these receivers acquire the service but refuse to decode
video and audio. This is standard practice in Japanese digital
terrestrial broadcasting per ARIB TR-B14.

When mpegts_pcr_pid is set to a valid PID value:
- PCR packets are generated as adaptation-field-only packets on the
  specified PID at the interval set by pcr_period
- The video stream adaptation fields do not carry PCR flags
- The PMT PCR_PID field is set to the specified PID

When mpegts_pcr_pid is -1 (default): existing behavior is unchanged.

The existing mpegts_insert_pcr_only function is refactored into a
parameterized mpegts_insert_pcr_only_pid(s, pid, cc, discontinuity)
to avoid code duplication.

Fixes: https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/23649
Signed-off-by: shadowcaster3 <vasyl.samoilov@gmail.com>

Changed files

doc/muxers.texi
libavformat/mpegtsenc.c

Change #276648

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 15:18:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	db05df9d135fb56a4babb836d5e9f5c1d984e087

Comments

avcodec/cfhd: reject transform-2 output wider than the plane
Fixes: out of array access
Fixes: cfhd_transform2_output_width_oob.avi
Fixes: MimvoaEVpKow
Found-by: Adrian Junge (vurlo)

Changed files

libavcodec/cfhd.c

Change #276649

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Fri 31 Jul 2026 15:21:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0dd50ec5353ce61f7e2c75a44036de8b8338b055

Comments

avfilter/dnn: initialize DNNData at the config_input call sites
vf_dnn_processing.c and vf_dnn_detect.c both declare an uninitialized
DNNData model_input before using it. This commit initialize it.

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavfilter/vf_dnn_detect.c
libavfilter/vf_dnn_processing.c

Change #276650

Category	ffmpeg
Changed by	Dan Dennedy <danohnoyoudont@dennedy.org>
Changed at	Fri 31 Jul 2026 15:29:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	66a8c3758216736cb0cc0b9123f0514eb2627d2f

Comments

scale_d3d11: Fix hw_frame_ctx reference leak
This fixes #20995. The reference is taken twice and assigned to the same pointer (see above in the same function). Only the latter is needed.

Changed files

libavfilter/vf_scale_d3d11.c

Change #276651

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 15:34:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5d4d3bdc61412641883a45e060e810f80ea7f4b5

Comments

avcodec/pgssubdec: always give an output rect a palette
Fixes: NULL pointer dereference
Fixes: poc_null_deref.sup
Fixes: Fobbab5Vtlr3
Found-by: VRI with 图龙锋

Changed files

libavcodec/pgssubdec.c

Change #276664

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8ab8b4fabe8440d32a966f355dc2102d86336673

Comments

avcodec/ffv1dec: mark the slice damaged when its remap fails

Changed files

libavcodec/ffv1dec.c

Change #276665

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	56743898d214decccf219aea4233da7aca9dc71c

Comments

avcodec/ffv1dec: reject a remap that produces zero entries

Changed files

libavcodec/ffv1dec.c

Change #276666

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	533a3c254d864a7cd4066cd055c2e2d038e95576

Comments

avcodec/vulkan/ffv1_dec_setup: act on the slice header rejection

Changed files

libavcodec/vulkan/ffv1_dec_setup.comp.glsl

Change #276667

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cc82fdfbd4cacef2018f1dadb29330bf80c4fe87

Comments

avcodec/vulkan/ffv1_dec_setup: reject a remap that produces zero entries

Changed files

libavcodec/vulkan/ffv1_dec_setup.comp.glsl

Change #276668

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c91b50659b9ae66a966b74246ead91418be53d21

Comments

avcodec/vulkan/ffv1_dec_setup: test mul_count as unsigned

Changed files

libavcodec/vulkan/ffv1_dec_setup.comp.glsl

Change #276669

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Fri 31 Jul 2026 21:37:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	946272b79a325e9bce613b260e50e4e4fe7f3159

Comments

avcodec/vulkan/ffv1_dec_setup: bound the fltmap write

Changed files

libavcodec/vulkan/ffv1_dec_setup.comp.glsl

Change #276701

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 01 Aug 2026 16:45:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4f8043e658e195cc4e44cb79e864c4a6466b72a7

Comments

avcodec/cbs_h266: size vps_direct_ref_layer_flag for the full layer range
Fixes: out of array access
Fixes: poc_vvc_vps.266
Fixes: 1z6RugPwxp39
Found-by: TainAn Bill

Changed files

libavcodec/cbs_h266.h

Change #276706

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b5483446d33c6a883d7eec247498a3cc067bfeba

Comments

avcodec/x86/vc1dsp_init: Avoid fpel wrappers
One can just reuse the already existing SIZExSIZE fpel
wrappers.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/fpel.asm
libavcodec/x86/vc1dsp_init.c

Change #276707

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	933b195bcb2cf1cd8e2c7ba0967452074f01e77d

Comments

avcodec/vc1dsp: Avoid negating unnecessarily
Due to the earlier "if (a1 < a0 || a2 < a0)" check
it is guaranteed that min(a1,a2) is < a0, allowing
simplifications.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/vc1dsp.c

Change #276708

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f30ef7eec41dce9ff0f134dcd579b5ade95b5b1b

Comments

avcodec/x86/vc1dsp_loopfilter: Avoid imul
Multipliying a GPR by 0x01010101 splats a byte to a dword;
the intention was to avoid vector shuffles lateron, yet
this is not true because the byte splat necessitates
a byte->word conversion lateron.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276709

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8434869cfdf3c708c048ce7db1519d4a27f20e1d

Comments

avcodec/x86/vc1dsp_loopfilter: Reduce number of GPRs used
Reuse the register that held pq for this instead of using
another one. This saves a push+pop on 32bit systems and avoids
REX prefixes on Unix64.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276710

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f2035a4605734b15847ebf6ffb0d6bd0b1507de3

Comments

avcodec/x86/vc1dsp_loopfilter: Reorder instructions slightly
In particular, apply the rounding before adding the result
of the multiplication. No change in performance here.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276711

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	55bdc109a1911044d11c2a8d8b447a55f6bbe059

Comments

avcodec/x86/vc1dsp_loopfilter: Avoid unnecessary PABSW
The loop filter is only active if min(a1,a2)<abs(a0)
which is done via masking. Ergo the sign of min(a1,a2)-abs(a0)
is known (always negative) and one does not need to use
PABSW to get its absolute value.

This gives a small speedup. Old benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:            3.1
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:       33.6 ( 0.09x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:          43.0
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:      33.4 ( 1.29x)
  vc1dsp.vc1_h_loop_filter8_bestcase_c:            6.0
  vc1dsp.vc1_h_loop_filter8_bestcase_sse2:        38.4 ( 0.16x)
  vc1dsp.vc1_h_loop_filter8_bestcase_ssse3:       37.2 ( 0.16x)
  vc1dsp.vc1_h_loop_filter8_bestcase_sse4:        37.6 ( 0.16x)
  vc1dsp.vc1_h_loop_filter8_worstcase_c:          87.5
  vc1dsp.vc1_h_loop_filter8_worstcase_sse2:       38.4 ( 2.28x)
  vc1dsp.vc1_h_loop_filter8_worstcase_ssse3:      37.6 ( 2.32x)
  vc1dsp.vc1_h_loop_filter8_worstcase_sse4:       37.6 ( 2.32x)
  vc1dsp.vc1_h_loop_filter16_bestcase_c:          11.7
  vc1dsp.vc1_h_loop_filter16_bestcase_sse2:       42.0 ( 0.28x)
  vc1dsp.vc1_h_loop_filter16_bestcase_ssse3:      41.4 ( 0.28x)
  vc1dsp.vc1_h_loop_filter16_bestcase_sse4:       44.8 ( 0.26x)
  vc1dsp.vc1_h_loop_filter16_worstcase_c:        167.2
  vc1dsp.vc1_h_loop_filter16_worstcase_sse2:      41.8 ( 4.00x)
  vc1dsp.vc1_h_loop_filter16_worstcase_ssse3:     41.7 ( 3.92x)
  vc1dsp.vc1_h_loop_filter16_worstcase_sse4:      45.3 ( 3.69x)
  vc1dsp.vc1_v_loop_filter4_bestcase_c:            3.6
  vc1dsp.vc1_v_loop_filter4_bestcase_ssse3:       17.1 ( 0.21x)
  vc1dsp.vc1_v_loop_filter4_worstcase_c:          49.3
  vc1dsp.vc1_v_loop_filter4_worstcase_ssse3:      17.0 ( 2.89x)
  vc1dsp.vc1_v_loop_filter8_bestcase_c:            6.3
  vc1dsp.vc1_v_loop_filter8_bestcase_sse2:        17.7 ( 0.36x)
  vc1dsp.vc1_v_loop_filter8_bestcase_ssse3:       16.8 ( 0.38x)
  vc1dsp.vc1_v_loop_filter8_worstcase_c:          85.8
  vc1dsp.vc1_v_loop_filter8_worstcase_sse2:       17.7 ( 4.83x)
  vc1dsp.vc1_v_loop_filter8_worstcase_ssse3:      16.7 ( 5.12x)
  vc1dsp.vc1_v_loop_filter16_bestcase_c:          12.7
  vc1dsp.vc1_v_loop_filter16_bestcase_sse2:       26.4 ( 0.48x)
  vc1dsp.vc1_v_loop_filter16_bestcase_ssse3:      24.0 ( 0.53x)
  vc1dsp.vc1_v_loop_filter16_worstcase_c:        170.1
  vc1dsp.vc1_v_loop_filter16_worstcase_sse2:      25.7 ( 6.63x)
  vc1dsp.vc1_v_loop_filter16_worstcase_ssse3:     25.2 ( 6.75x)

New benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:            3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:       32.4 ( 0.09x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:          42.3
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:      32.7 ( 1.29x)
  vc1dsp.vc1_h_loop_filter8_bestcase_c:            6.1
  vc1dsp.vc1_h_loop_filter8_bestcase_sse2:        37.1 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_bestcase_ssse3:       36.4 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_bestcase_sse4:        36.6 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_worstcase_c:          87.2
  vc1dsp.vc1_h_loop_filter8_worstcase_sse2:       36.9 ( 2.36x)
  vc1dsp.vc1_h_loop_filter8_worstcase_ssse3:      35.8 ( 2.35x)
  vc1dsp.vc1_h_loop_filter8_worstcase_sse4:       36.7 ( 2.38x)
  vc1dsp.vc1_h_loop_filter16_bestcase_c:          12.0
  vc1dsp.vc1_h_loop_filter16_bestcase_sse2:       40.0 ( 0.30x)
  vc1dsp.vc1_h_loop_filter16_bestcase_ssse3:      39.8 ( 0.30x)
  vc1dsp.vc1_h_loop_filter16_bestcase_sse4:       44.4 ( 0.27x)
  vc1dsp.vc1_h_loop_filter16_worstcase_c:        166.4
  vc1dsp.vc1_h_loop_filter16_worstcase_sse2:      39.9 ( 4.17x)
  vc1dsp.vc1_h_loop_filter16_worstcase_ssse3:     39.7 ( 4.19x)
  vc1dsp.vc1_h_loop_filter16_worstcase_sse4:      43.8 ( 3.80x)
  vc1dsp.vc1_v_loop_filter4_bestcase_c:            3.6
  vc1dsp.vc1_v_loop_filter4_bestcase_ssse3:       16.3 ( 0.22x)
  vc1dsp.vc1_v_loop_filter4_worstcase_c:          49.3
  vc1dsp.vc1_v_loop_filter4_worstcase_ssse3:      16.2 ( 3.04x)
  vc1dsp.vc1_v_loop_filter8_bestcase_c:            6.4
  vc1dsp.vc1_v_loop_filter8_bestcase_sse2:        16.3 ( 0.39x)
  vc1dsp.vc1_v_loop_filter8_bestcase_ssse3:       15.8 ( 0.40x)
  vc1dsp.vc1_v_loop_filter8_worstcase_c:          85.7
  vc1dsp.vc1_v_loop_filter8_worstcase_sse2:       16.3 ( 5.24x)
  vc1dsp.vc1_v_loop_filter8_worstcase_ssse3:      15.8 ( 5.42x)
  vc1dsp.vc1_v_loop_filter16_bestcase_c:          12.7
  vc1dsp.vc1_v_loop_filter16_bestcase_sse2:       24.7 ( 0.52x)
  vc1dsp.vc1_v_loop_filter16_bestcase_ssse3:      22.7 ( 0.56x)
  vc1dsp.vc1_v_loop_filter16_worstcase_c:        169.4
  vc1dsp.vc1_v_loop_filter16_worstcase_sse2:      24.1 ( 7.03x)
  vc1dsp.vc1_v_loop_filter16_worstcase_ssse3:     24.6 ( 6.88x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276712

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8f05c446a1e3f16785b3768623050a23c7d7bfb1

Comments

avcodec/x86/vc1dsp_loopfilter: Use psignw when available
The VC1 loop filter uses something equivalent to
if (a0_sign ^ clip_sign) {
    if (clip_sign) d = -d;
} else d = 0;
which can be mapped to psignw.

Old benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:            3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:       32.4 ( 0.09x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:          42.3
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:      32.7 ( 1.29x)
  vc1dsp.vc1_h_loop_filter8_bestcase_c:            6.1
  vc1dsp.vc1_h_loop_filter8_bestcase_sse2:        37.1 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_bestcase_ssse3:       36.4 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_bestcase_sse4:        36.6 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_worstcase_c:          87.2
  vc1dsp.vc1_h_loop_filter8_worstcase_sse2:       36.9 ( 2.36x)
  vc1dsp.vc1_h_loop_filter8_worstcase_ssse3:      35.8 ( 2.35x)
  vc1dsp.vc1_h_loop_filter8_worstcase_sse4:       36.7 ( 2.38x)
  vc1dsp.vc1_h_loop_filter16_bestcase_c:          12.0
  vc1dsp.vc1_h_loop_filter16_bestcase_sse2:       40.0 ( 0.30x)
  vc1dsp.vc1_h_loop_filter16_bestcase_ssse3:      39.8 ( 0.30x)
  vc1dsp.vc1_h_loop_filter16_bestcase_sse4:       44.4 ( 0.27x)
  vc1dsp.vc1_h_loop_filter16_worstcase_c:        166.4
  vc1dsp.vc1_h_loop_filter16_worstcase_sse2:      39.9 ( 4.17x)
  vc1dsp.vc1_h_loop_filter16_worstcase_ssse3:     39.7 ( 4.19x)
  vc1dsp.vc1_h_loop_filter16_worstcase_sse4:      43.8 ( 3.80x)
  vc1dsp.vc1_v_loop_filter4_bestcase_c:            3.6
  vc1dsp.vc1_v_loop_filter4_bestcase_ssse3:       16.3 ( 0.22x)
  vc1dsp.vc1_v_loop_filter4_worstcase_c:          49.3
  vc1dsp.vc1_v_loop_filter4_worstcase_ssse3:      16.2 ( 3.04x)
  vc1dsp.vc1_v_loop_filter8_bestcase_c:            6.4
  vc1dsp.vc1_v_loop_filter8_bestcase_sse2:        16.3 ( 0.39x)
  vc1dsp.vc1_v_loop_filter8_bestcase_ssse3:       15.8 ( 0.40x)
  vc1dsp.vc1_v_loop_filter8_worstcase_c:          85.7
  vc1dsp.vc1_v_loop_filter8_worstcase_sse2:       16.3 ( 5.24x)
  vc1dsp.vc1_v_loop_filter8_worstcase_ssse3:      15.8 ( 5.42x)
  vc1dsp.vc1_v_loop_filter16_bestcase_c:          12.7
  vc1dsp.vc1_v_loop_filter16_bestcase_sse2:       24.7 ( 0.52x)
  vc1dsp.vc1_v_loop_filter16_bestcase_ssse3:      22.7 ( 0.56x)
  vc1dsp.vc1_v_loop_filter16_worstcase_c:        169.4
  vc1dsp.vc1_v_loop_filter16_worstcase_sse2:      24.1 ( 7.03x)
  vc1dsp.vc1_v_loop_filter16_worstcase_ssse3:     24.6 ( 6.88x)

New benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:            3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:       32.0 ( 0.09x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:          43.2
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:      31.8 ( 1.36x)
  vc1dsp.vc1_h_loop_filter8_bestcase_c:            6.0
  vc1dsp.vc1_h_loop_filter8_bestcase_sse2:        36.9 ( 0.16x)
  vc1dsp.vc1_h_loop_filter8_bestcase_ssse3:       35.5 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_bestcase_sse4:        36.2 ( 0.17x)
  vc1dsp.vc1_h_loop_filter8_worstcase_c:          87.3
  vc1dsp.vc1_h_loop_filter8_worstcase_sse2:       37.6 ( 2.33x)
  vc1dsp.vc1_h_loop_filter8_worstcase_ssse3:      35.3 ( 2.40x)
  vc1dsp.vc1_h_loop_filter8_worstcase_sse4:       36.1 ( 2.42x)
  vc1dsp.vc1_h_loop_filter16_bestcase_c:          11.8
  vc1dsp.vc1_h_loop_filter16_bestcase_sse2:       40.1 ( 0.29x)
  vc1dsp.vc1_h_loop_filter16_bestcase_ssse3:      39.2 ( 0.30x)
  vc1dsp.vc1_h_loop_filter16_bestcase_sse4:       42.9 ( 0.28x)
  vc1dsp.vc1_h_loop_filter16_worstcase_c:        166.5
  vc1dsp.vc1_h_loop_filter16_worstcase_sse2:      39.8 ( 4.18x)
  vc1dsp.vc1_h_loop_filter16_worstcase_ssse3:     39.1 ( 4.26x)
  vc1dsp.vc1_h_loop_filter16_worstcase_sse4:      42.9 ( 3.88x)
  vc1dsp.vc1_v_loop_filter4_bestcase_c:            3.6
  vc1dsp.vc1_v_loop_filter4_bestcase_ssse3:       15.7 ( 0.22x)
  vc1dsp.vc1_v_loop_filter4_worstcase_c:          49.1
  vc1dsp.vc1_v_loop_filter4_worstcase_ssse3:      15.8 ( 3.10x)
  vc1dsp.vc1_v_loop_filter8_bestcase_c:            6.4
  vc1dsp.vc1_v_loop_filter8_bestcase_sse2:        16.3 ( 0.39x)
  vc1dsp.vc1_v_loop_filter8_bestcase_ssse3:       15.3 ( 0.42x)
  vc1dsp.vc1_v_loop_filter8_worstcase_c:          85.7
  vc1dsp.vc1_v_loop_filter8_worstcase_sse2:       16.3 ( 5.26x)
  vc1dsp.vc1_v_loop_filter8_worstcase_ssse3:      15.2 ( 5.62x)
  vc1dsp.vc1_v_loop_filter16_bestcase_c:          13.6
  vc1dsp.vc1_v_loop_filter16_bestcase_sse2:       24.8 ( 0.55x)
  vc1dsp.vc1_v_loop_filter16_bestcase_ssse3:      21.3 ( 0.64x)
  vc1dsp.vc1_v_loop_filter16_worstcase_c:        170.0
  vc1dsp.vc1_v_loop_filter16_worstcase_sse2:      23.9 ( 7.11x)
  vc1dsp.vc1_v_loop_filter16_worstcase_ssse3:     23.3 ( 7.28x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276713

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3ea01992d6330a13e56b332d0e99397852ce2e06

Comments

avcodec/x86/vc1dsp_loopfilter: Don't use MMX regs in horiz. loop filter
Using XMM registers in this SSSE3 function leads to fewer shuffles
when transposing the input; it also allows to combine calculating
a1 and a2. Because of this, codesize is the same as before (on Unix64)
although MMX instructions are shorter.

Old benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:           3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:      32.0 ( 0.09x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:         42.9
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:     31.9 ( 1.35x)

New benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:           3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:      29.9 ( 0.10x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:         43.7
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:     29.9 ( 1.46x)

Hint: checkasm's benchmark always uses the same buffer that is
partially updated by the horizontal loop filter function (the
middle two of eight columns are updated using word-sized stores).
They therefore lead to store-to-load-forwarding failure. If
checkasm_alternate were used to alternate between two buffers,
the benchmarks would be as follows:

Old benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:           3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:      16.4 ( 0.18x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:         23.9
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:     16.3 ( 1.47x)

New benchmarks:
  vc1dsp.vc1_h_loop_filter4_bestcase_c:           3.0
  vc1dsp.vc1_h_loop_filter4_bestcase_ssse3:      15.1 ( 0.20x)
  vc1dsp.vc1_h_loop_filter4_worstcase_c:         23.6
  vc1dsp.vc1_h_loop_filter4_worstcase_ssse3:     15.2 ( 1.55x)

Notice that at some callsites, the partially modified buffer
is immediately reloaded again*, so that both scenarios can happen.

*: See the TT_4X4 and TT_4X8 cases at the end of vc1_p_h_loop_filter()
or vc1_b_h_intfi_loop_filter() or the luma field blocks in
vc1_p_h_intfr_loop_filter().

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm

Change #276714

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3db7a55280c7af2a5897c6f9473cc5816ff37631

Comments

avcodec/x86/vc1dsp_loopfilter: Don't use MMX regs
Up until now ff_vc1_v_loop_filter4_ssse3 used MMX registers
despite being an SSSE3 function. Porting it to XMM registers
didn't lead to any change in benchmarks here, but abides
by the ABI (one function less that does not omit emms).
Unfortunately the size of the function increased by 64B
by this.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_loopfilter.asm
tests/checkasm/vc1dsp.c

Change #276715

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 01 Aug 2026 16:50:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	406c5a37aa666d648928a142d367483fe1acdd17

Comments

avcodec/x86/vc1dsp_inv_trans: Port to SSE2
No change in performance here.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vc1dsp_init.c
libavcodec/x86/vc1dsp_inv_trans.asm
tests/checkasm/vc1dsp.c

Change #276721

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 02 Aug 2026 00:44:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1b946a3abd81bd2b50ef6d38003ddb848be7acc7

Comments

avformat/iamf_parse: check that num_sub_mixes and num_audio_elements in Mix Presentations are not zero
As required by the spec in Section 3.7

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/iamf_parse.c

Change #276722

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 02 Aug 2026 00:44:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d2476bfd2b0ab683f24ff92cb2b206549f0a1659

Comments

avformat/iamf_parse: bound the output mix gain duration by the audio elements
Based on a patch by Michael Niedermayer

Fixes: unbounded allocation
Fixes: poc_output_mix_config_dos.iamf
Fixes: 387qWpjUgeKH
Found-by: Duong Tran Thai (CyStack)
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/iamf_parse.c

Change #276723

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 02 Aug 2026 00:47:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d9da090b1d05cd731610462fd91c5f167e698909

Comments

avformat/dashdec: don't stop at the first input EOF
The current logic would stop demuxing the entire manifest once the first Representation
returned EOF, which could result in plenty of packets from other Representations being
lost.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/dashdec.c

Change #276726

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c9d1fd3c0117db40456f2e2f9492a15f36a7534

Comments

tests/checkasm/huffyuvdsp: Return early if there is no function to test
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/huffyuvdsp.c

Change #276727

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dd8730cfe354b84d60b1d70ac36c5fe23425f9dc

Comments

tests/checkasm/huffyuvdsp: Actually test masks
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/huffyuvdsp.c

Change #276728

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b1142ed5be1f48af8b1d8989ab04fb379e24fe59

Comments

tests/checkasm/huffyuvdsp: Actually test width not multiple of 16
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/huffyuvdsp.c

Change #276729

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3928c291b75420cc5fe7e17fb0541f6fceb7314c

Comments

avcodec/x86/huffyuvdsp: Remove unaligned add_int16 version
Both source and dst are always properly aligned (to stride align)
here; also document this fact.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/huffyuvdsp.h
libavcodec/x86/huffyuvdsp.asm

Change #276730

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0d342f01f03530e37e23372467c8d3b9623a19d0

Comments

avcodec/x86/huffyuvdsp_template: Don't push+pop reg unnecessarily
All calling conventions on x64 have enough volatile GPRs.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/huffyuvdsp_template.asm

Change #276731

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d8cb1ef0e52d4b5847796b6c0e2e9a702985daa3

Comments

avcodec/x86/huffyuvdsp: Remove remnants of MMX
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/huffyuvdsp.asm

Change #276732

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:00:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eb40b0b6b7424bb7eee66abfaab8577c5a13cb7b

Comments

tests/checkasm/huffyuvdsp: Add test for add_hfyu_median_pred_int16
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/huffyuvdsp.c

Change #276733

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:12:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0127016c41092e53809b6548cc1487ec5a8c87d0

Comments

avcodec/x86/huffyuvdsp: Add SSE4 add_hfyu_median_pred_int16
Heavily based upon the existing mmxext function, but
it allows to use p{max,min}uw, so also supports 16bpp
pixel formats. It is also faster:

  add_hfyu_median_pred_int16_c:       14368.1
  add_hfyu_median_pred_int16_mmxext:   7384.2 ( 1.94x)
  add_hfyu_median_pred_int16_sse4:     6474.4 ( 2.21x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/huffyuvdsp.asm
libavcodec/x86/huffyuvdsp_init.c

Change #276734

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:19:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c8fb970a9220f79f1e659309daba3a264da78253

Comments

avcodec/x86/huffyuvdsp: Remove ff_add_hfyu_median_prediction_mmxext
Superseded by SSE4. This makes the huffyuv decoders ABI compliant,
e.g. nothing messes up the FPU state at all any more. Therefore
the emms_c() can be removed from huffyuvdec.c.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/huffyuvdec.c
libavcodec/x86/huffyuvdsp.asm
libavcodec/x86/huffyuvdsp_init.c
tests/checkasm/huffyuvdsp.c

Change #276735

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 02:19:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a59498db085e3d635532397128550141ab87408a

Comments

avcodec/huffyuvdsp: Remove pix_fmt parameter from ff_huffyuvdsp_init()
It is unused since the mmx add_hfyu_median_pred_int16 function
has been removed.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/aarch64/huffyuvdsp_init_aarch64.c
libavcodec/huffyuvdec.c
libavcodec/huffyuvdsp.c
libavcodec/huffyuvdsp.h
libavcodec/riscv/huffyuvdsp_init.c
libavcodec/x86/huffyuvdsp_init.c
tests/checkasm/huffyuvdsp.c

Change #276736

Category	ffmpeg
Changed by	Guo Yejun <yejun.guoohnoyoudont@intel.com>
Changed at	Sun 02 Aug 2026 06:38:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	133923a209c4581d31d44f55136976f464934155

Comments

avfilter/dnn: assign correct output layout before use

Changed files

libavfilter/dnn/dnn_backend_tf.c

Change #276737

Category	ffmpeg
Changed by	Guo Yejun <yejun.guoohnoyoudont@intel.com>
Changed at	Sun 02 Aug 2026 09:59:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7e72069f15efbef1e25b25d35c4e0511b43262e

Comments

avfilter/dnn: fix race when querying queue length
Replace av_usleep polling with a condition variable wait.

Changed files

libavfilter/dnn/dnn_backend_common.c
libavfilter/dnn/safe_queue.c
libavfilter/dnn/safe_queue.h

Change #276755

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	898b8aa9220a6c71c61f649f55dbf7407f180c2a

Comments

vulkan_ffv1: survive setup failures on preceding frames
A frame's private data is allocated before start_frame(), and its free
callback runs even if setup failed partway, so partially-initialized
private data has to be tolerated everywhere: inter frames reffed the
previous frame's slice state without checking it exists, end_frame()
waited on a semaphore that was never created, and the free callback
read a feedback buffer that was never allocated or written. Inter
frames after a failed frame now error out until the next keyframe.

Also set the device function pointers before anything can fail, as the
free callback needs them.

Changed files

libavcodec/vulkan_ffv1.c

Change #276756

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b27f276bdd1ec490c67bc9bfeaa8bef113a9f2f4

Comments

vulkan_ffv1: use the extradata slice structure in the low-slice-count warning
slice_count is only set while decoding a frame, so the warning always
printed a slice count of 0 at init time, and fired on streams with
plenty of slices.

Changed files

libavcodec/vulkan_ffv1.c

Change #276757

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2b0c0b2c000b2ecac8c211cddbd6aeee3d0293b1

Comments

hwcontext_vulkan: do not enable host transfers on coinciding decode output
Decode output which doubles as reference frames (COINCIDE mode)
remains an active DPB slot for as long as the decoder keeps
referencing it, and decode submissions bake the image layout into
their barriers at record time. A host-side layout transition can
therefore not be synchronized against them, and host transfers of such
frames were found to corrupt image contents on Intel once frame
threading varied the submission order.

Do not set VK_IMAGE_USAGE_HOST_TRANSFER_BIT_EXT on such frames. This
makes the incompatibility visible through the usage flags, and routes
our own transfer path through the queue.

Changed files

libavutil/hwcontext_vulkan.c

Change #276758

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d42cd604d02c920c8c9bf2a8d8b9962a122044e2

Comments

vulkan_ffv1: always keep the decoder's fltmap descriptor valid
The setup shader statically uses the fltmap binding, but the decoder
only bound a buffer to it for float formats, leaving a dangling
descriptor on everything else, with the write reachable by any
bitstream signalling remap.

Gate remap on a new push-constant flag, so that streams signalling it
without a fltmap buffer error out, and point the descriptor at the
slice feedback buffer when no fltmap exists; the gate guarantees it is
never written through this binding.

Changed files

libavcodec/ffv1_vulkan.h
libavcodec/vulkan/ffv1_common.glsl
libavcodec/vulkan/ffv1_dec_setup.comp.glsl
libavcodec/vulkan_ffv1.c

Change #276759

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2dd3328240cc11a32c1acbae365558e3a1b0232e

Comments

vulkan: give ff_vk_host_map_buffer an explicit import length
The function sized every import from src_data to the end of the source
buffer. For packets which are small views of a much larger allocation,
this pinned and mapped every page of the underlying allocation, at
around 1.6us per page, regardless of what the GPU ever touches.

Take the import length as a VkDeviceSize, clamped to the end of the
source buffer; the clamp doubles as the implementation of
VK_WHOLE_SIZE.

Changed files

libavcodec/proresenc_kostya_vulkan.c
libavcodec/vulkan_apv.c
libavcodec/vulkan_dpx.c
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_prores.c
libavcodec/vulkan_prores_raw.c
libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h

Change #276760

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ac16cfb86e3cdba0c05778ab09309b7d900a136c

Comments

hwcontext_vulkan: only originate host image layout transitions from copyable layouts
Host image layout transitions may only originate from a layout in
pCopySrcLayouts, or from UNDEFINED, discarding the contents. The host
transfer path transitioned from whatever layout the frame was last
left in, which for pool-recycled frames can be a video layout, which
no driver lists as host-copyable.

Uploads overwrite the entire image, so transition from UNDEFINED
there. Downloads have to preserve the contents, so route frames in a
non-host-copyable layout through the GPU path.

Changed files

libavutil/hwcontext_vulkan.c

Change #276761

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sun 02 Aug 2026 15:11:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0da8f2f4eea877d234cbc88fa394ddfa028ffa94

Comments

vulkan_encode_av1: set primary_ref_frame to a reference name, not a slot
primary_ref_frame is an index into the seven reference names, but the
code assigned it the reference's DPB slot. The two coincide only while
the reference sits in slot 0; once it rotates to slot 1,
referenceNameSlotIndices[primary_ref_frame] is -1, which is invalid,
and NVIDIA drivers lose the device. Inherited from vaapi_encode_av1,
where the same confusion is harmless as the raw frame header maps
every reference name to the same slot.

Have set_name_slot() return the name it picked and use that.

Fixes #20540.

Changed files

libavcodec/vulkan_encode_av1.c

Change #276766

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 02 Aug 2026 18:05:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ae0e0ba3c3df840191e7a4c9b76494372ec53202

Comments

avformat/mov: reject a trun sample count the input cannot hold
Fixes: OOM
Fixes: 525088811/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5229499332231168
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg

Changed files

libavformat/mov.c

Change #276767

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 02 Aug 2026 18:06:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a290dec0bfe45f68880dcbf7cddea3a4939d00b2

Comments

avcodec/utils: factor the timecode fields out of ff_alloc_timecode_sei

Changed files

libavcodec/utils.c

Change #276768

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 02 Aug 2026 18:06:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5589dfd9bc0153f9c29832f39bfe1989ebf5f5aa

Comments

avcodec/utils: add ff_alloc_timecode_metadata_av1()

Changed files

libavcodec/internal.h
libavcodec/utils.c

Change #276769

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sun 02 Aug 2026 18:06:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	afd059c340feb7379d33f8102b7dd7d01288fba1

Comments

avcodec/nvenc: write AV1 timecode metadata in AV1 syntax
Fixes: https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/20610

Tested on 5090 by claude

Changed files

libavcodec/nvenc.c

Change #276770

Category	ffmpeg
Changed by	张永鹏 <roc63ohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 18:33:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d09b0c5e1707dc21ecb7df8195d9f896d5c13606

Comments

fftools/ffmpeg_opt: return an error when refusing overwrite
Return EEXIST when output overwrite is refused so ffmpeg exits with a
non-zero status for -n and interactive no responses.

Signed-off-by: 张永鹏 <roc63@outlook.com>

Changed files

fftools/ffmpeg_opt.c
tests/fate/ffmpeg.mak

Change #276772

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 18:36:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b4d11f4046335a1e992d75e4ef9890844f7cf742

Comments

avcodec/x86/lossless_videodsp: Optimize add_gradient_pred
Combinations of shift and add like
pslldq  m3, m2, 1
paddb   m2, m3
are linear in the input register m2. This implies
that instead of applying the same sequence of shifts and adds
to two different registers and adding said registers
one can just add the register first and then apply
said sequence of shifts and adds once.

Old benchmarks:
  add_gradient_pred_c:           2296.2
  add_gradient_pred_ssse3:        570.4 ( 3.95x)
  add_gradient_pred_avx2:         308.8 ( 7.29x)

New benchmarks:
  add_gradient_pred_c:           2315.6
  add_gradient_pred_ssse3:        344.0 ( 6.72x)
  add_gradient_pred_avx2:         205.2 (11.28x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/lossless_videodsp.asm

Change #276773

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 18:36:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6c123e217ba990732be960c99e0571528e6201c6

Comments

avcodec/x86/lossless_videodsp: Hoist calculation out of loop
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/lossless_videodsp.asm

Change #276774

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 02 Aug 2026 18:36:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f0fc529cbe5c918553dd210999b140efd6632a46

Comments

avcodec/x86/lossless_videodsp: Use memory operand where possible
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/lossless_videodsp.asm

Change #276775

Category	ffmpeg
Changed by	ninbura <gabrielohnoyoudont@ninbura.com>
Changed at	Sun 02 Aug 2026 19:10:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c2802e520a08dbf0c939ec8b5653b29dfd886427

Comments

avdevice/avfoundation: add device selection by USB serial and unique ID
A capture device could previously be selected only by index or by name.
Both are unreliable when multiple audio/video devices share a name or
across reboots. AVFoundation reorders the device indices, and the USB
video uniqueID embeds the macOS locationID, which can be reassigned on
reboot or replug and can then resolve to a different physical device.

Add -video_device_id and -audio_device_id, which take a prefixed
identifier: uid:<unique ID> or serial:<USB serial number>.

The USB serial number is the only identifier that stays pegged to a
given physical unit. For video it is resolved to the device's current
locationID via IOKit. For audio it is matched against the uniqueID,
which already embeds it. The unique ID covers devices that have no
serial, such as virtual camera/audio devices.

-list_devices additionally prints each device's uniqueID and USB serial
so the values can be discovered. IOKit is detected in configure and used
only when available.

Signed-off-by: Gabriel Balaich <ffmpeg@ninbura.com>

Changed files

Changelog
configure
doc/indevs.texi
libavdevice/avfoundation.m

Change #276786

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Mon 03 Aug 2026 03:54:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	146e0f7b792033848e549c58ff928645fdeb5991

Comments

avformat/mp3enc: fix underflow of the LAME encoder delay
AV_RL32() is unsigned, so a skip_samples value below 528 + 1 wraps around
instead of clamping to zero and is written out as a delay of 4095 samples.

Changed files

libavformat/mp3enc.c

Change #276787

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 03 Aug 2026 03:57:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	64fe3bee7f169ab034f084e5b1d66859bf2ddbbf

Comments

vulkan: enable VK_KHR_shader_maximal_reconvergence when available

Changed files

libavutil/hwcontext_vulkan.c
libavutil/vulkan_functions.h
libavutil/vulkan_loader.h

Change #276788

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 03 Aug 2026 03:57:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e54f60ed1b9565d0a864595d7ae50d04f53e3536

Comments

vulkan: declare maximal reconvergence in shaders doing subgroup reductions
These shaders reduce across the subgroup after divergent control flow
and implicitly rely on invocations reconverging as written, the
MaximallyReconvergesKHR execution mode turns that assumption into a
guarantee.

Changed files

libavcodec/vulkan/apv_encode_tiles.comp.glsl
libavcodec/vulkan/ffv1_enc_rct_search.comp.glsl
libavcodec/vulkan/prores_ks_estimate_slice.comp.glsl
libavcodec/vulkan/prores_ks_trellis_node.comp.glsl

Change #276789

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 03 Aug 2026 06:18:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	25663e69246ae373efb2358fdf509858045e69c8

Comments

vulkan: only reset execution fences at submission time
The fence was reset when a context began recording, so an error
between recording start and submission (observed in the wild as
intermittent VK_ERROR_MEMORY_MAP_FAILED from vkEndCommandBuffer on
ANV under threaded load) left it permanently unsignalled, deadlocking
the next user of the context.

Reset the fence only once a fully recorded submission is about to be
handed to the queue: abandoned recordings then leave the fence
signalled from the previous submission, and the context self-heals on
reuse. Should the queue submission itself fail, signal the fence with
an empty submission.

Changed files

libavutil/vulkan.c

Change #276798

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	934742548a3c3f34ceaef2116dda2810bacdf625

Comments

tests/checkasm/huffyuvencdsp: Avoid using static variable
This has been done in order to use the same width
for benchmarks of different instruction sets to
make the benchmarks comparable. Yet it has a downside:
Only one width would ever be executed when using --repeat.
Luckily libcheckasm makes it easy to fix this:
At the start of every test function, the internal state
of rnd() is reset, so that it produces the same sequence
of random values. So just removing the static variable works.

This would also make this test trivially parallelizable.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/huffyuvencdsp.c

Change #276799

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8984d7c239095f5fdaf2c55665386fdb0aa42183

Comments

tests/checkasm/sbcdsp: Avoid using static variable
This has been done in order to use the same number of blocks
for benchmarks of different instruction sets to
make the benchmarks comparable. Yet it has a downside:
Only one number would ever be executed when using --repeat.
Luckily libcheckasm makes it easy to fix this:
At the start of every test function, the internal state
of rnd() is reset, so that it produces the same sequence
of random values. So just removing the static variable works.

This would also makes this test trivially parallelizable.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/sbcdsp.c

Change #276800

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	78193e6490b92b9749183beb0defd6c355f452aa

Comments

tests/checkasm/llviddsp: Fix crash with AVX2
llviddsp functions typically operate on whole lines,
i.e. the pointers are aligned to STRIDE_ALIGN. They
can therefore avoid tail handling by just clobbering
the padding (if any).

The test gets a random width via 16 * av_clip(rnd(), 16, 128);
in practice, it is very unlikely for rnd() to return
a value between 16 and 128, so width is typically 16*16 or
16*128. Buffers of this size are allocated later.

When rnd() returns an odd number in the allowed range
(this happens for the seed 4161216273), the buffers used
in the test don't contain the padding that exists in actual
usage, leading to invalid stores and also to segmentation faults
(usage of aligned load instructions on unaligned addresses).

Fix this by adding the necessary alignment to the buffers.

Also don't use a static variable to store width (it is unnecessary,
because since the switch to libcheckasm, rnd() always returns
the same sequence of random numbers for test runs with different
instruction sets) and use a really random width, not something
that is mostly just one of two values and always mod 16.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/llviddsp.c

Change #276801

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4a51f189ea288b2f0692a357fb2ee8fdeb935625

Comments

tests/checkasm/hpeldsp: Avoid using static variables
This has been done in order to use the same height
for each function for benchmarks of different
instruction sets to make the benchmarks comparable.
Yet it has a downside: Only one number would ever be
executed when using --repeat, hiding potential bugs.

Luckily libcheckasm makes it easy to fix this:
At the start of every test function, the internal state
of rnd() is reset, so that it produces the same sequence
of random values. So just initialize the heights every time
(and before actually testing any function).

This also would make this test trivially parallelizable with --repeat.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/hpeldsp.c

Change #276802

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4acf9f27062abbffb8eefd522df0d57355b29167

Comments

tests/checkasm/mpegvideo_unquantize: Avoid using static variables
This has been done in order to use the same parameters
that influence the number of elements to be processed
sets to make benchmarks between different instruction
sets comparable. Yet it has a downside: Only one parameter
combination would ever be executed when using --repeat,
hiding potential bugs.

Luckily libcheckasm makes it easy to fix this:
At the start of every test function, the internal state
of rnd() is reset, so that it produces the same sequence
of random values. So just initialize these parameters
every time (and before actually testing any function).

This also would make this test trivially parallelizable with --repeat.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/mpegvideo_unquantize.c

Change #276803

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 03 Aug 2026 09:20:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a6ed5e29c1b3e29271a3c73fe778a04ddcb5576

Comments

tests/checkasm/checkasm: Remove inappropriate headers
Since the switch to libcheckasm, checkasm no longer uses
FFmpeg's emms_c, timers, AVLFG or string handling functions,
so remove these headers.
This necessitated adding some missing headers (mostly string.h)
to some files.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/aacpsdsp.c
tests/checkasm/aes.c
tests/checkasm/af_afir.c
tests/checkasm/apv_dsp.c
tests/checkasm/cavsdsp.c
tests/checkasm/checkasm.h
tests/checkasm/crc.c
tests/checkasm/diracdsp.c
tests/checkasm/float_dsp.c
tests/checkasm/lls.c
tests/checkasm/mpeg4videodsp.c
tests/checkasm/opusdsp.c
tests/checkasm/png.c
tests/checkasm/qpeldsp.c
tests/checkasm/sbcdsp.c
tests/checkasm/sbrdsp.c
tests/checkasm/vf_nlmeans.c
tests/checkasm/vf_pp7.c
tests/checkasm/vorbisdsp.c

Change #276812

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	08c90c60f7562301858f1bdf06e6bc8abd3ec878

Comments

swscale/ops: fix merge_comp_flags() for SWS_COMP_SWAPPED
This violates the documentation (monoid property).

It's a bit arbitrary whether to consider this an OR-type or AND-type flag,
since mixing swapped and non-swapped components is almost surely a bug, but
keeping it as an OR-type makes sure such cases at least show up in the
result.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops.c

Change #276813

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fc94457bb722159c1a8338b7f8454228210f9646

Comments

swscale/ops: refactor comp flags propagation slightly
Introduce a unified FORWARD() helper macro that can be used for any type of
op, whether it is independent per component or more complex. By initializing
every op to the same IDENTITY state, we can leverage the monoid property to
make this work for naive propagations as well.

As an aside, we also properly zero out the unrelated fields when discarding
a component (i.e. marking it as GARBAGE).

This will make a couple of up-coming refactors a bit easier.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops.c

Change #276814

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6321b5380939a4e676854d67fe41085d5f0c4ad8

Comments

swscale/ops: solve for component dependencies
Needed to determine reverse dependencies of different input planes, for
appropriately splitting op lists involving subsampled planes.

Generates benign diffs that just reflect the new addition, e.g.:

 yuva444p 16x16 -> rgb24 16x16:
   [ u8 ===X] SWS_OP_READ         : 3 elem(s) planar >> 0
     min: {0 0 0 _}, max: {255 255 255 _}
+    inputs: {x y z _}, outputs: {xyz yz xy _}
   [ u8 ===X] SWS_OP_CONVERT      : u8 -> f32
     min: {0 0 0 _}, max: {255 255 255 _}
+    inputs: {x y z _}, outputs: {xyz yz xy _}
   [f32 ...X] SWS_OP_LINEAR       : matrix3+off3 [[85/73 0 1.596027 0 -222.921566] [85/73 -0.391762 -0.812968 0 135.575295] [85/73 2.017232 0 0 -276.835851] [0 0 0 1 0]]
     min: {-222.921566 -171.630839 -276.835851 _}, max: {480.983073 432.493103 534.476153 _}
+    inputs: {xz xyz xy _}, outputs: {x y z _}
   [f32 ...X] SWS_OP_DITHER       : 16x16 matrix + {0 3 2 -1}
     min: {-222.919612 -171.628886 -276.833898 _}, max: {481.981120 433.491150 535.474200 _}
+    inputs: {xz xyz xy _}, outputs: {x y z _}
   [f32 ...X] SWS_OP_MAX          : {0 0 0 _} <= x
     min: {0 0 0 _}, max: {481.981120 433.491150 535.474200 _}
+    inputs: {xz xyz xy _}, outputs: {x y z _}
   [f32 ...X] SWS_OP_MIN          : x <= {255 255 255 _}
     min: {0 0 0 _}, max: {255 255 255 _}
+    inputs: {xz xyz xy _}, outputs: {x y z _}
   [f32 +++X] SWS_OP_CONVERT      : f32 -> u8
     min: {0 0 0 _}, max: {255 255 255 _}
+    inputs: {xz xyz xy _}, outputs: {x y z _}
   [ u8 XXXX] SWS_OP_WRITE        : 3 elem(s) packed >> 0
     ('X' unused, 'z' byteswapped, '=' copied, '$' const, '+' integer, '0' zero)

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops.c
libswscale/ops.h
tests/ref/fate/sws-ops-list

Change #276815

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	327319bb7b34d91a50b479f30055770f7ea603dd

Comments

swscale/ops: remove SwsLinearOp.mask
This was originally introduced to make matching linear ops against
implementations faster. However, since this is now handled on the uops
level, there is no more reason to carry this metadata on the ops level.

Simplifies a lot of places in the code. It will simplify even more, once
the linear optimizations are moved to the uops level.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/aarch64/ops_impl_conv.c
libswscale/format.c
libswscale/ops.h
libswscale/ops_optimizer.c
libswscale/uops.c

Change #276816

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0ff884dabec72f9e8bb548739e2a75ab03d1487b

Comments

swscale/csputils: remove unused struct
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/csputils.h

Change #276817

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d0bf8cff9fd86d43d6eecfa1dba00ae454cf09c9

Comments

swscale/aarch64: reject unknown ops by default
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/aarch64/ops_impl_conv.c

Change #276818

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ea8fb1b71eba487d0d5b5fc394d3760aa7fd7620

Comments

swscale/lut3d: alignment (cosmetic)
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/lut3d.c

Change #276819

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	69f0fa91d47c47c13c5182fc294f650aebacdf98

Comments

swscale/lut3d: pad 3DLUT to eliminate over-read
Instead of clamping in the application function. Trivial simplicity gain
for the SIMD code.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/lut3d.c
libswscale/lut3d.h

Change #276820

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6a56bd03620f7a6cdc934a34f17568e95f25fe19

Comments

swscale/lut3d: use refstruct for 3D LUT allocations
Needed anyways for the upcoming SWS_OP_LUT_3D.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c
libswscale/lut3d.c
libswscale/lut3d.h

Change #276821

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e43517c1d1b1e48399fec7f5a5d2c7158cff40ff

Comments

swscale/lut3d: simplify and hard-code 3DLUT format
The ops-based 3DLUT approach will take care of appropriately normalizing
the input to the expected domain; so the format choice no longer matters here
except for the lut3d_apply() function, which will only be used by the legacy
reference code path. So we can just continue hard-coding the format there.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c
libswscale/lut3d.c
libswscale/lut3d.h

Change #276822

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	96e0967ecac6cd7aaf4e25c5af72de82b9814c8a

Comments

swscale/cms: fix IPT rounding error
av_round16f() was incorrectly scaling to 65534, which would correspond to
a neutral PT offset of 32767 = (1 << 15) - 1, but the code was assuming a
value of (1 << 15). To fix it, and make the PT channel correctly symmetric
around the intended neutral value, we have to map PT = +0.5 to 65536, an
unrepresentable value. This is not an issue because the PT channel values are
strictly inside some subset of [-0.5, 0.5] in practice, for real in-gamut
color values - it's already an envelope that includes quite a bit of safety
margin.

Similarly, av_round16f() is also the wrong tool for the I/RGB channels,
because it incorrectly scaled those to 65534, an off by one of the intended
unorm16 full range peak of 65535. This silently resulted in e.g. RGBA64
true white (65535) not round-tripping through the 3DLUT.

Fix both by splitting this helper into two separate helpers, each of them
fixed to the correct, intended value range.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/cms.c

Change #276823

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6c770f33c8e94ea18000cd81da4f1e0ad0bac7a7

Comments

swscale/ops: add SWS_OP_LUT_3D and supporting code
After extensive testing, prototyping and benchmarking across a range of
systems, I determined that the optimal data layout for the SIMD 3DLUT is
essentially exactly the one we have. 16-bit integers are near optimal for
quality vs compactness, and crucially, x86 lets us load the entire packed
3DLUT entry with a single `vpgatherdq` instruction.

This dwarfs the loss from needing to cast the resulting 16-bit integers
back to f32 and renormalize.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops.c
libswscale/ops.h
libswscale/ops_optimizer.c

Change #276824

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4bdb4d564f677d2f7f7f03b3b9a902a06cd916ee

Comments

swscale/uops: add SWS_UOP_LUT_3D reference implementation
The structure for the tetrahedral interpolation deviates slightly from the
naive formulation in lut3d.c; instead of branching into every separate case,
we sort the weights and offsets using a series of conditional swaps. This
actually performs identically on my end, but results in code that is much
closer to what SIMD will be doing. That should hopefully serve as a better
reference for future SIMD implementors. (Myself included)

I also reordered the dynamic tone-mapping code a bit to better indicate the
sources of live register pressure that will manifest in the real SIMD kernel.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/csputils.h
libswscale/uops.c
libswscale/uops.h
libswscale/uops_backend.c
libswscale/uops_list.h
libswscale/uops_macros.h
libswscale/uops_tmpl.c

Change #276825

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d803957307d3b6ffda8d998d2282f11531477595

Comments

swscale/graph: lift SwsLut3D to SwsGraph top level
Since this depends on the overall picture parameters anyways, updating it
directly from the private setup() function was always a bit hacky. More
importantly, this needed for the ops-based 3DLUT implementation, which is
already using the pass priv pointer to store the compiled function.

Overall, simpler to just lift it to the top level and update it directly as
needed.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c
libswscale/graph.h

Change #276826

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	38196645f5e20b50a5143db6f0d14814fed77116

Comments

swscale/graph: move 3DLUT application to the conversion pass
Instead of generating the 3DLUT and applying it right away, init_passes()
now just generates it and passes it as a parameter to add_convert_pass(),
which will forward it to the underlying implementation.

I pre-emptively moved the function to the legacy section, in anticipation
of the following commit which will make this legacy-only.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c

Change #276827

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	05d8891a20de059f901fbb106f45b7699f81bba0

Comments

swscale/format: add SwsLut3D support to the ops list generator
The value range normalization and input range clamp will normally be
optimized away.

This commit also moves the legacy 3dlut pass to be legacy-exclusive, as the
ops code now uses the new logic. This ordering ensures that the conversion
works on every commit in isolation.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/format.c
libswscale/format.h
libswscale/graph.c
libswscale/op_list_gen_template.c

Change #276828

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3df239e113e5c5d23acbedc5cde4a5d70c5ad620

Comments

swscale/uops_macros: also generate op lists involving 3DLUTs
These may differ from the regular op lists in nontrivial ways, due to e.g.
different optimization steps being taken.

In practice, it seems this just adds the extra LUT_3D uops, but we don't
know that for sure, so better to brute force the list. That said, I do
think we can safely skip the extra backend flags in this case, at least.

Even in the worst case scenario, that would just force a fallback to the
C reference backend.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/op_list_gen_template.c
libswscale/tests/sws_ops.c
libswscale/tests/sws_ops_aarch64.c
libswscale/uops_macros.h
libswscale/uops_macros_gen.c

Change #276829

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 03 Aug 2026 11:32:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	01a25f74cc446a683318bab13dfd98a467082ef7

Comments

checkasm/sw_ops: add SWS_UOP_LUT_3D test
checkasm:
 - CPU: AMD Ryzen 9 9950X3D 16-Core Processor (00B40F40)
 - Timing source: x86 (rdtsc)
 - Bench duration: 100000 µs per function (448667793 cycles)
 - Random seed: 3773883393
Benchmark results:
  name                         cycles (vs ref)
  f32_lut_3d_xyz_dynamic_c:  125193.9
  f32_lut_3d_xyz_static_c:    25792.1
  f32_lut_3d_xyzw_dynamic_c: 123807.1
  f32_lut_3d_xyzw_static_c:   25739.0

This is roughly ~50% faster than the existing code in lut3d.c, from a quick
test.

Sponsored-by: Sovereign Tech Fund
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

tests/checkasm/sw_ops.c

Change #276848

Category	ffmpeg
Changed by	Scott Kidder <scottohnoyoudont@kidder.io>
Changed at	Mon 03 Aug 2026 13:08:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	883e8a6336b2651f7be79a6a9aa5f3cc22937948

Comments

avcodec/libopenh264dec: implement flush callback
Cisco's ISVCDecoder retains DPB and reference state across
avcodec_flush_buffers(), causing the next IDR to be accepted but
subsequent P-frames to fail decode after a seek or loop. Cisco's
API does not expose a non-destructive reset; tear down and rebuild
the decoder via the existing close/init callbacks.

The .flush callback is void and cannot report a re-init failure, so
svc_decode_frame() guards against a NULL decoder left behind by a
failed re-init. svc_decode_init() now also tears the decoder back
down on an Initialize() failure so that path leaves the same clean
NULL state the guard expects and does not leak the ISVCDecoder.

Signed-off-by: Scott Kidder <scott@kidder.io>

Changed files

libavcodec/libopenh264dec.c

Change #276882

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Mon 03 Aug 2026 13:51:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d3ad8a7fee6a647c6362e4a105d949282d50a98f

Comments

avformat/rawutils: reject raw RGB frames that do not fit an AVPacket
Fixes: integer overflow
Fixes: out of array access
Fixes: payload.film
Fixes: czK1F83k3zvT
Found-by: Clouditera Security, Z.ai Security, NSFOCUS

Changed files

libavformat/rawutils.c

Change #276923

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 03 Aug 2026 20:16:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9862dd83b1645b0c6e39b318aefb896138ffb08e

Comments

avcodec/aarch64/vc1dsp_neon: reduce literal pool alignment to 16 bytes
armasm64 warns "A4228: Alignment value exceeds AREA alignment; alignment
not guaranteed" on the .Lcoeffs literal pool. gas-preprocessor.pl emits
the text section as "AREA |.text|, CODE, READONLY, ALIGN=4", i.e. a
16-byte section alignment, so a 32-byte ALIGN inside it cannot be
guaranteed by the assembler.

16 is enough here: the pool holds .quad values that are read with
"ldr d0, .Lcoeffs" literal loads, which only require 8-byte alignment.
It also matches the rest of the aarch64 asm, where the function and
const macros both default to .align 4.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/aarch64/vc1dsp_neon.S

Change #277006

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1d14f9f8c3e4efb32000cb463ebb635ad6f2adb0

Comments

vulkan: decouple execution pool depths from queue and thread counts
Pool depth sets how far the CPU records ahead of the GPU, and has no relation
to the number of queues in a family.
Scaling it by the queue count only multiplied command pools, fences and context state.
Use a small fixed depth everywhere, and one context per thread only for threadsafe hwaccels,
which submit from every frame thread concurrently.

Changed files

libavcodec/vulkan_decode.c
libavfilter/vf_avgblur_vulkan.c
libavfilter/vf_blackdetect_vulkan.c
libavfilter/vf_blend_vulkan.c
libavfilter/vf_bwdif_vulkan.c
libavfilter/vf_chromaber_vulkan.c
libavfilter/vf_flip_vulkan.c
libavfilter/vf_gblur_vulkan.c
libavfilter/vf_interlace_vulkan.c
libavfilter/vf_overlay_vulkan.c
libavfilter/vf_scale_vulkan.c
libavfilter/vf_scdet_vulkan.c
libavfilter/vf_transpose_vulkan.c
libavfilter/vf_xfade_vulkan.c
libavfilter/vsrc_testsrc_vulkan.c
libavutil/hwcontext_vulkan.c
libavutil/vulkan.h

Change #277007

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b6ac28aff6ca8efdf0be3a8a159ae37b7234e408

Comments

vulkan: use fixed-size dependency arrays in FFVkExecContext
Execution contexts track at most a few dozen dependencies so dynamically
growing the arrays was just stupid, and ARR_REALLOC required the field
name to also exist as a local variable.

Changed files

libavcodec/vulkan_ffv1.c
libavutil/vulkan.c
libavutil/vulkan.h

Change #277008

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c5dddc2bbfa6c870092b80cededc1dd714d8e60

Comments

vulkan: release dependencies of completed execution contexts eagerly
Dependencies were only released when a context was about to be reused, so
all buffers and deps got held up and piled up in the queue.

Release them as soon as they're done instead. Cuts RAM usage down by a ton.

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #277009

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	24170896fa14845e2549ef5db47a4cddf660609b

Comments

vulkan: allow tracking AVRefStruct objects as execution dependencies
RefStructs > BufferRefs. We're converting everything.

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #277010

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d4bc34a070c00069e3da81a86fbf22aedb680825

Comments

vulkan: track semaphores/views directly as deps
There's no need for them to live in bufferrefs.

Changed files

libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h

Change #277011

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	318e2e3943cef34a353b17985495719f28ce0bb1

Comments

vulkan: manage buffer pools and host mappings with AVRefStruct
This converts all buffers/buffer pools into refstructs/refstruct pools,
saving allocs and making them easier to work with.

Changed files

libavcodec/apv_encode_vulkan.c
libavcodec/ffv1enc_vulkan.c
libavcodec/proresenc_kostya_vulkan.c
libavcodec/vulkan_apv.c
libavcodec/vulkan_decode.c
libavcodec/vulkan_decode.h
libavcodec/vulkan_dpx.c
libavcodec/vulkan_encode.c
libavcodec/vulkan_encode.h
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_prores.c
libavcodec/vulkan_prores_raw.c
libavfilter/vf_blackdetect_vulkan.c
libavfilter/vf_nlmeans_vulkan.c
libavfilter/vf_scdet_vulkan.c
libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h

Change #277012

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4e141536767ef57f0aaf1f5f33bfed3001d68d7d

Comments

vulkan: manage image views with AVRefStruct
The biggest gain is no longer needing to spin on the CPU on a semaphore
waiting for the image to become freeable. This was awful.

Changed files

libavcodec/vulkan_av1.c
libavcodec/vulkan_decode.c
libavcodec/vulkan_decode.h
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_h264.c
libavcodec/vulkan_hevc.c
libavcodec/vulkan_vp9.c
libavutil/vulkan.c
libavutil/vulkan.h

Change #277013

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	56d60b54196bd0b9e3d187385822acf4f9d3dbc9

Comments

proresenc_kostya_vulkan: gather slices on the GPU
Same as with APV, and FFv1.
1080p on RADV goes from 34 to 48 fps, 4K goes from 20 to 31 fps.

Changed files

libavcodec/proresenc_kostya_vulkan.c
libavcodec/vulkan/prores_ks_encode_slice.comp.glsl

Change #277014

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	73cc45ef0d5b3357d61e19d36624271d0fa3d81e

Comments

vulkan_decode: manage session parameters with AVRefStruct
Ditto.

Changed files

libavcodec/vulkan_av1.c
libavcodec/vulkan_decode.c
libavcodec/vulkan_decode.h
libavcodec/vulkan_h264.c
libavcodec/vulkan_hevc.c

Change #277015

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9548d68e5b54dfa46620d441da4ff44b3b64f2bf

Comments

vulkan: remove ff_vk_exec_add_dep_buf

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #277016

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eed19eefdaa605d02ff0795c9e7e82ca46636f99

Comments

vulkan: use only coherency to flush mapped memory when needed
Host map'd bufs skipped flushes unconditionally, always forcing the coherent flag
regardless of the memory type actually chosen for the import.
All known implementations import host memory as coherent, in which case nothing changes.

Changed files

libavutil/vulkan.c

Change #277017

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	709a72c374dc812c4e363fc68e57cfa28d1158c8

Comments

vulkan_decode: prefer device-local memory for the bitstream buffer
Fixes #23865

Changed files

libavcodec/vulkan_decode.c

Change #277018

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9287eb62b56b2d1e935bcf99912d70e394ec33fb

Comments

vulkan_ffv1: fall back to host memory for the fltmap buffer
A megabyte per slice, systems without resizable BAR can run out of
mappable device memory to place it in.

Changed files

libavcodec/vulkan_ffv1.c

Change #277019

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Wed 05 Aug 2026 09:04:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d295add2225e1ad9ba9d55cb612cce50072dc45d

Comments

vulkan: fix typo when printing the SPIR-V/binary size

Changed files

libavutil/vulkan.c

Change #277058

Category	ffmpeg
Changed by	Steve Lhomme <robux4ohnoyoudont@ycbcr.xyz>
Changed at	Wed 05 Aug 2026 13:06:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	87f17b68a831b3ebe5f10278fc7acd3bebe83ecc

Comments

swscale/x86: use HOSTCPPFLAGS instead of CPPFLAGS for native code
Similar to 976e18fdef8bcd0602b6b94bb157a82a3eb9a1a6.

Changed files

libswscale/x86/Makefile

Change #277070

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Wed 05 Aug 2026 13:49:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f1dbe5252b31f07954d1de42b513bb9f7b3b3970

Comments

avcodec/aarch64: rewrite lfe_fir1 to remove faddp reduction
The original implementation was a regression on Cortex-X925 (0.88x).

               X925 clang    X925 gcc      A725 clang   A725 gcc     A510 clang
c:              891          1638          792          1475         8084
neon_original: 1015 (0.88x)  1016 (1.61x)  747 (1.06x)  747 (1.97x)  6126 (1.31x)
neon_new:       696 (1.28x)   696 (2.35x)  623 (1.27x)  623 (2.37x)  3697 (2.18x)

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>

Changed files

libavcodec/aarch64/dcadsp_neon.S

Change #277086

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 05 Aug 2026 15:06:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8f80421b2da8f2f29bff7de089c3f06eeae32c9c

Comments

swscale/aarch64/ops_static: fix av_assert0 on NDEBUG builds
av_assert0 suppose to run always.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libswscale/aarch64/ops_static.c

Change #277117

Category	ffmpeg
Changed by	Guo Yejun <yejun.guoohnoyoudont@intel.com>
Changed at	Wed 05 Aug 2026 15:11:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e13b2e00e89dc8cdbb05cca7c22ebbbf8d9fd171

Comments

avfilter/dnn: fix object detection of openvino backend
update ctx->ov_option.layout to the explict value which will be
used in function init_model_ov.

also move the code before input_resizable which uses the layout

Changed files

libavfilter/dnn/dnn_backend_openvino.c

Change #277138

Category	ffmpeg
Changed by	iSold Leo <sohnoyoudont@qwqlog.com>
Changed at	Wed 05 Aug 2026 21:34:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ec2a4105e2a5db5941e93326ee89897774a95462

Comments

swscale: avoid overflow in fast bilinear edge handling
The final edge clamp computes the source position using int
multiplication before shifting. With sufficiently wide inputs this
overflows, which may suppress the clamp and leave the last output
pixels interpolated with the padding byte.

Promote the multiplication to int64_t in the C, MMXEXT and VSX
implementations.

Add a regression test covering the rightmost pixel of a wide upscale,
which is wrong before this change on both the C and the MMXEXT path.

Fixes: signed integer overflow: 15 * 255918080 cannot be represented in type 'int'
Fixes: #21591
Signed-off-by: iSold Leo <s@qwqlog.com>

Changed files

libswscale/hscale_fast_bilinear.c
libswscale/ppc/swscale_vsx.c
libswscale/x86/hscale_fast_bilinear_simd.c
tests/fate/filter-video.mak
tests/ref/fate/filter-scale-fast-bilinear-wide-edge

Change #277139

Category	ffmpeg
Changed by	iSold Leo <sohnoyoudont@qwqlog.com>
Changed at	Wed 05 Aug 2026 21:37:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f02c4372ed28b40502a5d377a8a8d3492cafce58

Comments

avfilter/af_pan: check the id of named input channels before use
parse_channel_name() only rejects negative values on the named channel
path, while the numbered "c%d" path also checks against MAX_CHANNELS.
av_channel_from_string() accepts UNSD (512), UNK (768) and AMBI (1024),
and the id was then used directly to index used_in_ch[MAX_CHANNELS] on
the stack and pan->gain[out_ch_id][in_ch_id].

Before this change:

  ffmpeg -f lavfi -i "anullsrc=cl=stereo" -af "pan=stereo|FL=AMBI" -f null -
  af_pan.c:214:17: runtime error: index 1024 out of bounds for type 'int[64]'

The first out of bounds access is the read at the "reference twice"
check, so the symptom depends on what happens to be on the stack:
FL=UNK is silently accepted and produces silence, FL=UNSD reports
"Can not reference in channel 512 twice", and FL=AMBI aborts.

The check has to be at the caller and not in parse_channel_name(), which
is shared with the out channel path: there the named id is converted by
av_channel_layout_index_from_channel() into a layout index, so high ids
are legitimately supported and "pan=AMBI|AMBI=FL" works.

AVERROR_PATCHWELCOME is used because these are valid AVChannel values
that af_pan cannot represent: for named input channels the gain matrix is
indexed by the raw id, and the renumbering loop in config_props() is
bounded by MAX_CHANNELS as well, so such a mapping never worked.

The in_ch_id < 0 half is currently unreachable and only kept for symmetry
with the out channel check above.

Regression since 1f96db959c1235bb7079d354e09914a0a2608f62, which dropped
the "channel_id >= MAX_CHANNELS" part of the check when converting to the
new channel layout API.

Reported-by: karnakarreddi
Fixes: #22963
Signed-off-by: iSold Leo <s@qwqlog.com>

Changed files

libavfilter/af_pan.c

Change #277140

Category	ffmpeg
Changed by	iSold Leo <sohnoyoudont@qwqlog.com>
Changed at	Wed 05 Aug 2026 21:37:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f9ed016d429597810bd3173e131531179e711f1f

Comments

fate/filter-audio: add test for pan named input channel id rejection
"pan=stereo|FL=UNK" resolves to AVChannel id 768, which was used as an
index into a 64 element array before the previous commit.

UNK is used because it was the only one of the three reachable high ids
that exited with 0 before the fix; AMBI (1024) aborted and UNSD (512)
failed with an unrelated message, so a test that only checked for a
non-zero exit status would have passed before the fix as well.

For the same reason the test greps the error message on stderr instead of
only looking at the exit status: a crash also exits non-zero.

Signed-off-by: iSold Leo <s@qwqlog.com>

Changed files

tests/fate/filter-audio.mak

Change #277160

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Thu 06 Aug 2026 00:28:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	85c857a7c3b7ce9d0da041edbb85559e3a8cbd0e

Comments

compat/w32pthreads: add pthread_mutex_trylock

Changed files

compat/w32pthreads.h

Change #277167

Category	ffmpeg
Changed by	Ayoub Nabil <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Thu 06 Aug 2026 02:30:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fc02470c62a71a8fa431f3e424bc8049fec8d90a

Comments

avfilter/f_ebur128: fix top-back channel weighting
ITU-R BS.1770 assigns a weight of 1.0 to height channels, but the
filter included the top-back channels in the 1.41 surround mask.

remove the top-back channels from that mask and add a FATE test for
the resulting loudness measurement.

fixes #23968.

Signed-off-by: Ayoub Nabil <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/f_ebur128.c
tests/fate/filter-video.mak
tests/ref/fate/filter-metadata-ebur128-height

Change #277235

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 11:26:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	95c43d7df7b72e3a4e8dce8c6718cffedb32211d

Comments

avcodec/fmtconvert: Remove int32_to_float_fmul_array8
Unused since aebf07075f4244caf591a3af71e5872fe314e87b.

Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/aarch64/fmtconvert_init.c
libavcodec/aarch64/fmtconvert_neon.S
libavcodec/arm/fmtconvert_init_arm.c
libavcodec/arm/fmtconvert_neon.S
libavcodec/arm/fmtconvert_vfp.S
libavcodec/fmtconvert.c
libavcodec/fmtconvert.h
libavcodec/riscv/fmtconvert_init.c
libavcodec/riscv/fmtconvert_rvv.S
libavcodec/x86/fmtconvert.asm
libavcodec/x86/fmtconvert_init.c
tests/checkasm/fmtconvert.c

Change #277297

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 06 Aug 2026 15:52:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a72b9465cda477b34e35d6d1eda670603a0a9f5b

Comments

configure: make _WIN32_WINNT == 0x0601 the minimum supported version
Every hardware accel implementation and now Vulkan require API present
stating with Windows 7, so it's about time we stop pretending to support
Vista.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

configure

Change #277309

Category	ffmpeg
Changed by	Nicolas Gaullier <nicolas.gaullierohnoyoudont@cji.paris>
Changed at	Thu 06 Aug 2026 17:19:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	73aa056f4934d825af9ef86ffa4c6cca4e9ccd10

Comments

avcodec/dolby_e: Add error recovery when parse_mantissas run out of bits
Mantissas are the last data in the channel subsegment and it appears it
is sometimes missing a very few bits for the parsing to complete.
This must not be confused with data corruption.
In standard conditions with certified products, it has been observed
that the occurence of this issue is pretty steady and about once every 2
hours. The truncation is at about 950 out of the 1024 values (923 is the
minimum I have seen so far).
The current code raises a severe 'Read past end' error and all data is
lost resulting in 20ms(@25fps) of silence for the affected channel.
This patch introduces a tolerance: if 800 out of the 1024 mantissas have
been parsed, a simple warning is raised and the data is preserved.

Signed-off-by: Nicolas Gaullier <nicolas.gaullier@cji.paris>

Changed files

libavcodec/dolby_e.c

Change #277315

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:58:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	91188e6ae0fe6d3ea41f45878232476b20238674

Comments

avfilter/x86/vf_pullup: Avoid addition
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_pullup.asm

Change #277316

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fb15c5dd4f010396f740c13bc9cadb6fbb19e9f4

Comments

avfilter/x86/vf_atadenoise: Reduce number of registers used
Saves a push+pop on Win64 and some REX prefixes everywhere.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277317

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e4f2d7595bae4d004ac1e18b833a310775715b46

Comments

avfilter/x86/vf_atadenoise: Don't load too much data
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277318

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bab7f937a6b246740fb3252827e95d28f5f0c2f6

Comments

avfilter/x86/vf_atadenoise: Combine stores
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277319

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	698f0d22bf983f9b804da31b46b796d5a35c332f

Comments

avfilter/x86/vf_atadenoise: Avoid negating mask
Possible by using pandn.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277320

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5ccfaa872ae46115510fa258c83c14fe528cd222

Comments

avfilter/x86/vf_atadenoise: Avoid shifting unnecessarily
For a word mask register (where all the values are either
0x0 or 0xFFFF) a right logical shift by 15 is equivalent
to negating, so instead of negating the mask and then adding
it to another register one can simply subtract the mask from
the other register.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277321

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 06 Aug 2026 21:59:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fda0cb5cf8b5e34b29e8c1e3e756750e5e2a8a46

Comments

avfilter/x86/vf_atadenoise: Test earlier
ptest has quite a bit of latency, so it should be scheduled earlier.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/vf_atadenoise.asm

Change #277340

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Fri 07 Aug 2026 04:01:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	06deae14bf15e937e639d3c1382c2aecb2453e2a

Comments

avcodec/x86/h264_intrapred_10bit: Replace MMX with SSE2
Replace MMX implementation of pred4x4_dc_10 and pred4x4_horizontal_up_10
with SSE2.

  pred4x4_dc_10_c:                          9.7
  pred4x4_dc_10_mmxext:                     3.7 ( 2.63x)
  pred4x4_dc_10_sse2:                       3.6 ( 2.71x)

  pred4x4_horizontal_up_10_c:              13.7
  pred4x4_horizontal_up_10_mmxext:          5.5 ( 2.52x)
  pred4x4_horizontal_up_10_sse2:            4.6 ( 2.96x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_intrapred_10bit.asm
libavcodec/x86/h264_intrapred_init.c

Change #277341

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 07 Aug 2026 04:33:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c717827955ae7b996923fe6952ad5699b84af4bd

Comments

swresample/rematrix: use unity gain for top-front downmix
ITU-R BS.2127-1 maps U+030 and U-030 to the matching ear-level front channels at unity gain.

the current matrix uses 0.707 when the input also contains front channels. use 1.0 instead.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libswresample/rematrix.c

Change #277342

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 07 Aug 2026 04:33:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5c395992f99feb47860e4cc99a0cea2009457870

Comments

swresample/rematrix: handle top-back channels
top-back channels are currently left unaccounted and can be dropped during downmixing.

when the output retains top-front channels, follow IAMF 1.1.0 and fold top-back into top-front at 0.707. prefer this path over ear-level rear channels to preserve the height layer in x.1.4 to x.1.2 downmixes.

when no matching height output remains, map top-back to back or side channels, then fall back to front or mono outputs. handle top-back center separately and add direct tests for every matrix path.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libswresample/Makefile
libswresample/rematrix.c
libswresample/tests/.gitignore
libswresample/tests/rematrix.c
tests/fate/libswresample.mak

Change #277419

Category	ffmpeg
Changed by	Gabriel Balaich <ffmpegohnoyoudont@ninbura.com>
Changed at	Fri 07 Aug 2026 15:55:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a81b23a85aca34f9942dc1a63a521946a6407d3f

Comments

avdevice/avfoundation: fix build with deployment targets below macOS 12
c2802e520a selected kIOMainPortDefault whenever the SDK is new enough
(__MAC_OS_X_VERSION_MAX_ALLOWED), but that symbol is only available
since macOS 12, so building with a newer SDK against an older
deployment target failed with -Wunguarded-availability-new.

Check the deployment target (__MAC_OS_X_VERSION_MIN_REQUIRED) instead
and keep using the equivalent kIOMasterPortDefault when targeting
older releases.

See: https://lists.ffmpeg.org/archives/list/ffmpeg-devel@ffmpeg.org/thread/S7PGVSGNLIZ2WRYV6VOBAD3RWYUNWA3H/
Reported-by: Helmut K. C. Tessarek <tessarek@evermeet.cx>
Signed-off-by: Gabriel Balaich <ffmpeg@ninbura.com>

Changed files

libavdevice/avfoundation.m

Change #277433

Category	ffmpeg
Changed by	softworkz <softworkzohnoyoudont@hotmail.com>
Changed at	Fri 07 Aug 2026 19:51:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	63d0237033cca9782932b809f960378d34ee4869

Comments

avutil/hwcontext_vulkan: Query export support with the real image flags
Query with the flags the images are actually created with. The
DRM-modifier path is unchanged.

Signed-off-by: softworkz <softworkz@hotmail.com>

Changed files

libavutil/hwcontext_vulkan.c

Change #277446

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Fri 07 Aug 2026 23:18:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	23fccd657c0ac89f7e60f5465cf3f33fb796d29f

Comments

avformat/mp3enc: keep trailing padding spanning several packets
The trailing padding is read from the AV_PKT_DATA_SKIP_SAMPLES side data of
every packet, overwriting the previous value, so only the last packet was
ever accounted for. A single packet holds at most one frame, which caps the
padding that can be written at 1152 + 528 + 1 samples.

LAME regularly reports more than that: gapless/gapless.mp3 carries 1984 and
comes out of a stream copy with 1681, decoding to 303 samples more than the
file it was copied from.

Accumulate instead, and add the decoder delay once the total is known.

Fixes: https://trac.ffmpeg.org/ticket/9755

Changed files

libavformat/mp3enc.c
tests/fate/gapless.mak
tests/ref/fate/gapless-mp3-remux

Change #277447

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 08 Aug 2026 02:33:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f944afd04097178b7e3c0d6c7f4e524a9e8f6063

Comments

avutil/opt: Use correct enum size
AV_OPT_TYPE_PIXEL_FMT and AV_OPT_TYPE_SAMPLE_FMT are documented
to use the corresponding enum type. This matters when using
-fshort-enums.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/opt.c

Change #277512

Category	ffmpeg
Changed by	Jeongkeun Kim <variety0724ohnoyoudont@gmail.com>
Changed at	Sat 08 Aug 2026 23:18:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d920566afea459ea8700b4ac7c888ab78733a981

Comments

tests/checkasm/dcadsp: add test for lfe_x96_fixed

Changed files

tests/checkasm/dcadsp.c

Change #277514

Category	ffmpeg
Changed by	Jeongkeun Kim <variety0724ohnoyoudont@gmail.com>
Changed at	Sat 08 Aug 2026 23:18:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	280fd609164a3137c53d760a52fa78bbaf37d5d0

Comments

avcodec/aarch64: add NEON lfe_x96_fixed for DCA DSP
Implement the fixed-point LFE x96 interpolation filter in AArch64
NEON. Four input samples are processed per iteration. The history
sample is kept in a vector lane, and ext constructs the delayed input.

Benchmarks by Martin Storsjö using checkasm:
               a53     a55    a520     a72     a76    a720
              1.86x   1.88x   1.24x   1.99x   1.62x   2.30x

Author's measurement on Neoverse-N1 (Oracle Cloud, clang 14): 1.29x.

Signed-off-by: Jeongkeun Kim <variety0724@gmail.com>

Changed files

libavcodec/aarch64/dcadsp_init_aarch64.c
libavcodec/aarch64/dcadsp_neon.S

Change #277522

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Sat 08 Aug 2026 23:46:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5e6d663c94936db16fcb381f709524b07373e53d

Comments

avfilter/vf_bwdif: fix line boundary checks for >8 bits content
The mrefs/prefs boundary conditions scaled the y limits by the number of
bytes per sample, even though the ref offsets are already expressed in
elements. For content deeper than 8 bits this clamped perfectly valid
in-bounds line taps.

Found by validating the libplacebo GPU port against this filter, which
matches exactly on 16-bit content everywhere except these lines.

This changes the output for >8 bit content on the affected lines.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavfilter/vf_bwdif.c
tests/ref/fate/filter-bwdif10

Change #277523

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Sat 08 Aug 2026 23:46:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a20737f662fd3f3f5999e873b9e7c90b5efc375

Comments

Revert "lavfi/bwdif: fix heap-buffer-overflow with small height videos"
This reverts commit 795bccdaf57772b1803914dee2f32d52776518e2.

The heap-buffer-overflow was a symptom of the >8 bit boundary
conditions bug fixed in the previous commit. This workaround is no
longer needed.

This restores spatial-only interpolation of the whole frame for the
first and last fields, instead of running the temporal edge filter
against cloned reference frames on the outermost lines. Because of that
first frame references are updated to pre-fix values.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavfilter/vf_bwdif.c
tests/ref/fate/filter-bwdif-mode0
tests/ref/fate/filter-bwdif-mode1
tests/ref/fate/filter-bwdif10

Change #277539

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Sun 09 Aug 2026 04:13:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	09bf8dab5b8f5c9d1c9280af4dec7f84e8c0fe8b

Comments

avfilter/dnn: implement zero-copy CUDA tensor mapping for Torch backend
When a CUDA frame with a supported RGB sw_format is received, map the
GPU device pointer directly into a LibTorch tensor using
torch::from_blob() with CUDA device options, bypassing cudaMemcpy
entirely.

Changes:

 - Add fill_model_input_cuda() that extracts the CUdeviceptr from the
   AVHWFramesContext and wraps it in a PyTorch GPU tensor with a no-op
   deleter (memory owned by FFmpeg AVBuffer ref-counting).
 - Handle GPU memory alignment padding via custom strides derived from
   AVFrame linesize.
 - Permute the mapped NHWC tensor to NCHW (PyTorch convention).
 - Manage CUDA context explicitly with cuCtxPushCurrent/cuCtxPopCurrent
   for thread safety in async filter graphs.
 - Implement zero-copy Device-to-Device output mapping that writes the
   model result directly into the output frame's VRAM, bypassing
   ff_proc_from_dnn_to_frame / sws_scale entirely for CUDA frames.
 - Add torch::cuda::synchronize() to prevent async race conditions
   with downstream encoders (e.g. NVENC reading uninitialized frames).
 - Add format validation in ff_dnn_zero_copy_supported_cuda() to reject
   unsupported sw_formats and batching with CUDA zero-copy early.
 - Register AV_PIX_FMT_CUDA in vf_dnn_processing pixel format list.
 - Gracefully fall back to CPU path when sw_format is not a supported
   RGB variant or when CONFIG_CUDA is not enabled.

Signed-off-by: Raja-89 <imraja729@gmail.com>

Changed files

doc/filters.texi
libavfilter/dnn/dnn_backend_torch.cpp
libavfilter/dnn_filter_common.c
libavfilter/dnn_filter_common.h
libavfilter/vf_dnn_processing.c

Change #277540

Category	ffmpeg
Changed by	Raja-89 <imraja729ohnoyoudont@gmail.com>
Changed at	Sun 09 Aug 2026 04:13:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1e0279143db99d7324b17f9784b3229122269b38

Comments

doc/filters: document async, nireq, and batch_size options for dnn_processing
Add documentation for three previously undocumented common DNN options:
- async: toggle async inference (default enabled)
- nireq: number of concurrent inference requests
- batch_size: frames per inference request

Also add usage examples for the Libtorch backend showing basic
CPU inference and batch inference pipelines.

Signed-off-by: Raja Rathour <rajarathour1829@gmail.com>

Changed files

doc/filters.texi

Change #277547

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:24:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5d53a3f9cdfeea9e40ed218aba9d4e36ff206545

Comments

Squashed 'tests/checkasm/ext/' changes from e13b0bb3ff..e822e429f3
e822e429f3 utils: Silence an MSVC warning about conversion from double to float
abf8fb0261 utils: Add checkasm_randomize_interval() and float equivalent
7aea236f8f utils: Simplify checkasm_randomize_range() formulation
49972c1c72 utils: Fix advertised value range of checkasm_randf()
11c569cbdc utils: Use appropriate PRNG size for checkasm_init()
6fd9bafd11 utils: Add way more PRNG helper functions
d7f10858c8 utils: Use prng() primitive for checkasm_randomize()
781f16f1ac utils: Parallelize PRNG
4681280b65 selftest: Add tests for utils.h functions
02d0b1ee4c Clear CPU state after any checkasm_call()
48821c5d14 Print cpu mask in JSON output
5d500889dc tests/selftest: Test CPU flag masking
dbf5380202 tests: Offset arch-specific CPU flags by 10
09b4c6a8c9 include/checkasm.h: Add CheckasmCpuInfo.mask
dd1ca4ed09 utils: Document PRNG re-seeding at the start of each test
c93961a221 utils: Use substantially more robust PRNG
e46c473f73 Re-order conditional (mostly cosmetic)
16738f028d Seed PRNG before running CheckasmTest.init()
9292f9cf6d Bikeshed summary line a bit more, especially for interrupted runs
0f03612ec9 Suppress non-failure output after first iteration
bfdb230dd9 checkasm: Print statusline to track current test/bench progress
c8996a284b checkasm: Move test iteration variable to global state
46e136ee0e utils: Add self-repainting statusline buffer
800053edca Route all log messages through checkasm_fprintf()
f6791fee59 utils: Redefine checkasm_fprintf() as checkasm_vfprintf()
629a211feb Add optional CheckasmTest.init() and uninit()
354f3d3e8d riscv/callcheck: Avoid out-of-range li immediate under LLVM on rv32
6e1b4cd6ca riscv/cpu: Include <asm/unistd.h> for __NR_riscv_hwprobe
8130043b32 Fix CPU detection for AVX512F
9760810b49 Support YMM copy tests on processors with only AVX
dc8c320313 Support XMM copy tests on processors with only SSE

git-subtree-dir: tests/checkasm/ext
git-subtree-split: e822e429f33e4d02e0815bd497952b3f6deb0a7d

Changed files

ext-build-test.make
include/checkasm/checkasm.h
include/checkasm/test.h
include/checkasm/utils.h
src/checkasm.c
src/internal.h
src/riscv/callcheck.S
src/riscv/cpu.c
src/utils.c
src/x86/cpu.c
tests/meson.build
tests/selftest.c
tests/tests.h
tests/utils.c
tests/x86/tests.c
tests/x86/tests_asm.asm

Change #277548

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:24:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b59a4375acf0176efaf0152117e87c15d1e2aa90

Comments

checkasm: Update from upstream
This update is done because of the newly added init and uninit
callbacks which will be used to fix a segfault when repeating
the tx test; it also improves both the quality and performance
of the PRNG.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

no files

Change #277549

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:28:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	289595a834ecd2ffcc3022d8a445ef3827c4de9c

Comments

tests/checkasm/av_tx: Fix segfault when using --repeat
The TX tests require a context that is initialized
for a specific set of CPU flags. So in order to have
a context for the reference function to use, the last
context is stored (in static storage). This works well
in normal usage, but it does not work with the --repeat
feature that has been added to checkasm by the switch
to libcheckasm: When the C functions ought to be tested
a second time, there is a mismatch between the new
function pointer (which is the C version) and the reference
context (which is initialized to optimized CPU flags;
up until now, there was no stored context when testing
the C versions, but there is with --repeat), leading to segfaults.

Fix this by adding an uninit callback for the TX test,
replacing the atexit handler.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/av_tx.c
tests/checkasm/checkasm.c
tests/checkasm/checkasm.h

Change #277550

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:29:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c4a74c53c3205026a24f360213d181f3502e9c12

Comments

tests/checkasm/crc: Free allocations
The CRC test uses a linked list of containing the CRC tables
of already tested instruction sets; the head of the list
is in static storage and up until now was never freed
(i.e. reported as "still reachable" by Valgrind).
Free it properly by adding an uninit callback.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/checkasm.c
tests/checkasm/checkasm.h
tests/checkasm/crc.c

Change #277551

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:30:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4a2c02a993856576bf2852e2d3482b13f6d9012a

Comments

tests/checkasm/crc: Avoid static variables
This is possible because the state of the PRNG at the beginning
of every test function is the same for every cpu flag (since
the switch to libcheckasm). It also has the advantage that
random sizes are checked when using --repeat.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/crc.c

Change #277552

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 16:31:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a0e01bcee1cdf505bbcaf5fe7319387a35940fd8

Comments

tests/checkasm/sbcdsp: Zero-init SBCDSPContext
ff_sbcdsp_init() checks SBCDSPContext.increment to set
sbc_analyze_8s (which is not tested), leading to Valgrind
errors.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/sbcdsp.c

Change #277557

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:10:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ee89beda21284d0cf5f44cbe0af24f1c9320eb8b

Comments

tests/checkasm: Add ttadsp test
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/Makefile
tests/checkasm/checkasm.c
tests/checkasm/checkasm.h
tests/checkasm/ttadsp.c
tests/fate/checkasm.mak

Change #277558

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:21:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9612684badbfb61d6382545bfbdebef1bb4b4bea

Comments

avcodec/ttadata: Don't use too big arrays
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/ttadata.h

Change #277559

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:22:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4d151c37e0ba0df979b65030e568f1e94aa6bca3

Comments

avcodec/ttadata: Explicitly specify alignment
x86 assembly relies on it.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/ttadata.h

Change #277560

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:22:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a906e31d949e50b104db1b524bb8d66cf05f47d

Comments

avcodec/x86/ttaencdsp: Merge into ttadsp
This allows to avoid duplication of (source) code and constants.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/Makefile
libavcodec/x86/ttadsp.asm
libavcodec/x86/ttaencdsp.asm

Change #277561

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:22:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	de1bc8ea1dcebd00321ad6476db101871022fc2c

Comments

avcodec/ttaencdsp: Use unsigned to avoid UB
The decoder has been made to use unsigned values after
the fuzzer encountered signed integer overflow;
the same can probably happen in the encoder (I don't know),
but it definitely can happen in the checkasm test that
will be added soon. So use unsigned here, too.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/ttaencdsp.c

Change #277562

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:22:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d7ce8c30e0206c94a399de34dfed641b68d6923e

Comments

tests/checkasm/ttadsp: Also test ttaencdsp
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/Makefile
tests/checkasm/checkasm.c
tests/checkasm/checkasm.h
tests/checkasm/ttadsp.c
tests/fate/checkasm.mak

Change #277563

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:23:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a8d3ae2a7183a8243ebaf3d135915dea8c719371

Comments

avcodec/x86/ttadsp: Reduce number of registers used
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/ttadsp.asm

Change #277564

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:23:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ee83c175f817dd413f20d2453e4dc4eabd1e87f5

Comments

avcodec/x86/ttadsp: Optimize updating dl[4..7]
ttaencdsp old:
  filter_process_c:         15.0
  filter_process_ssse3:      9.0 ( 1.66x)
  filter_process_sse4:       6.9 ( 2.18x)

ttaencdsp new:
  filter_process_c:         14.9
  filter_process_ssse3:      8.3 ( 1.79x)
  filter_process_sse4:       6.4 ( 2.32x)

ttadsp old:
  filter_process_c:         14.7
  filter_process_ssse3:      8.7 ( 1.70x)
  filter_process_sse4:       6.6 ( 2.24x)

ttadsp new:
  filter_process_c:         14.6
  filter_process_ssse3:      8.1 ( 1.81x)
  filter_process_sse4:       6.3 ( 2.30x)

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/ttadsp.asm

Change #277565

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:25:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1f276a42dbd693ef58222e2c1499d45691b49089

Comments

avcodec/tta{,enc}dsp: Avoid stack
Pass the input value by value and return the output value
as return value and not by passing the input value by reference.
This improves decoding speed by 2% here.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/tta.c
libavcodec/ttadsp.c
libavcodec/ttadsp.h
libavcodec/ttaenc.c
libavcodec/ttaencdsp.c
libavcodec/ttaencdsp.h
libavcodec/x86/ttadsp.asm
libavcodec/x86/ttadsp_init.c
libavcodec/x86/ttaencdsp_init.c
tests/checkasm/ttadsp.c

Change #277592

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:26:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7635b501b841c8d91179f57cd88c49e270e473e1

Comments

avcodec/x86/diracdsp: Properly sign-extend strides
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm
libavcodec/x86/diracdsp_init.c

Change #277593

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:26:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fc866eaab31e1e05bf377a88a10fc4cff02b4897

Comments

avcodec/diracdsp: Use ptrdiff_t for stride
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/diracdec.c
libavcodec/diracdsp.c
libavcodec/diracdsp.h
libavcodec/qpeldsp.c
libavcodec/x86/diracdsp.asm
libavcodec/x86/diracdsp_init.c
tests/checkasm/diracdsp.c

Change #277594

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:26:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	af87713f8aff14a24e622ded02d7c87cce948a48

Comments

avcodec/diracdec: Fix shadowing
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/diracdec.c

Change #277595

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:26:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0654b34370cbdc1ec4f3768c235a8bf06bef5988

Comments

avcodec/x86/diracdsp: Fix clipping
packusdw saturates to uint16_t, yet CLIPW is designed for signed values.
If any of the saturated unsigned values were >= 2^16*, the macro would
return 0. Anyway, given that packusdw already saturates below,
clipping from both sides again is simply wasteful.

*: I don't know whether this can happen at all.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm

Change #277596

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:27:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a9cf64766434d27078f7887085cab8d35db3baa0

Comments

avcodec/x86/diracdsp: Avoid useless macros, init cpuflags properly
{ADD,PUT}_RECT was only used once since the removal of
mmx functions in d29a9c2aa68fc3eb6d61ff95c698e29316037583.
Furthermore, several of the INIT_XMM calls lacked the
actual cpuflags. Therefore the cpu flag suffix has been
manually added to the function names; furthermore, cpuflags()
as well as the check for supported instructions wouldn't work.
So add the cpu flags to INIT_XMM.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm

Change #277597

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:28:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	51c756382495f15116df6b39f7fceb09758629d2

Comments

avcodec/x86/diracdsp_init: Name function parameters
Also fix the comment of the same function in assembly.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm
libavcodec/x86/diracdsp_init.c

Change #277598

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:28:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	251b43db69db2bb1d935bd4ad1f20b58473bbec7

Comments

avcodec/x86/diracdsp: Avoid push+pop of xmm register
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm

Change #277599

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:28:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7637009f0276c47b4a101cf0ef12458dfea0d9b3

Comments

avcodec/x86/diracdsp_init: Avoid cast
put_signed_rect_clamped uses different types for differnt
8bit and >8bit content. The 8bit SSE2 function used the real
type in its function signature; this does not coincide with
the array of function pointers it gets put into and therefore
also not with how it is called which is UB. Also, casts of
function pointers to void* like it is done here are not
legal ISO-C as function pointers and object pointers need not
be convertible.

So just declare the function to have the required type for
the function pointer array.

Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/diracdsp.asm
libavcodec/x86/diracdsp_init.c

Change #277600

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:29:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7d448c9709ae04734f4e32bd7baf392cbd1c3c16

Comments

avcodec/qpeldsp, qpel_template: Use ptrdiff_t for stride
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/qpel_template.c
libavcodec/qpeldsp.c

Change #277601

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:29:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7289b00894a9d5f0b57deac6d86a228fc8f5607f

Comments

avcodec/dirac_dwt: Use ptrdiff_t for stride
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/dirac_dwt.c
libavcodec/dirac_dwt.h
libavcodec/dirac_dwt_template.c

Change #277602

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 09 Aug 2026 17:30:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	353deaf31e88e2f74b0dcb7d6ba77e68089cf7e3

Comments

avcodec/h264qpel_template: Use ptrdiff_t for stride
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/h264qpel_template.c

Change #277603

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 09 Aug 2026 18:55:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	77998c0dec044f5693188d8d1e2f0e0677d39241

Comments

swresample/rematrix: handle missing channels from 22.2 layout
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libswresample/rematrix.c
libswresample/swresample_internal.h
libswresample/tests/rematrix.c

Change #277604

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 09 Aug 2026 18:55:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	afd6b8451f845508972b43ff29ad837be0228ac1

Comments

libswresample/tests/rematrix: rewrite into a matrix printing tool
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libswresample/rematrix.c
libswresample/tests/rematrix.c
tests/fate/libswresample.mak
tests/ref/fate/swr-rematrix-22.2-5.1
tests/ref/fate/swr-rematrix-22.2-5.1.2
tests/ref/fate/swr-rematrix-22.2-5.1.4
tests/ref/fate/swr-rematrix-22.2-7.1
tests/ref/fate/swr-rematrix-22.2-7.1.2
tests/ref/fate/swr-rematrix-22.2-7.1.4
tests/ref/fate/swr-rematrix-22.2-7.2.3
tests/ref/fate/swr-rematrix-22.2-9.1.4
tests/ref/fate/swr-rematrix-22.2-9.1.6
tests/ref/fate/swr-rematrix-22.2-mono
tests/ref/fate/swr-rematrix-22.2-stereo
tests/ref/fate/swr-rematrix-5.1-mono
tests/ref/fate/swr-rematrix-5.1-stereo
tests/ref/fate/swr-rematrix-5.1.2-5.1
tests/ref/fate/swr-rematrix-5.1.2-7.1
tests/ref/fate/swr-rematrix-5.1.2-mono
tests/ref/fate/swr-rematrix-5.1.2-stereo
tests/ref/fate/swr-rematrix-5.1.4-5.1
tests/ref/fate/swr-rematrix-5.1.4-5.1.2
tests/ref/fate/swr-rematrix-5.1.4-7.1
tests/ref/fate/swr-rematrix-5.1.4-mono
tests/ref/fate/swr-rematrix-5.1.4-stereo
tests/ref/fate/swr-rematrix-7.1-5.1
tests/ref/fate/swr-rematrix-7.1-mono
tests/ref/fate/swr-rematrix-7.1-stereo
tests/ref/fate/swr-rematrix-7.1.2-5.1
tests/ref/fate/swr-rematrix-7.1.2-5.1.2
tests/ref/fate/swr-rematrix-7.1.2-5.1.4
tests/ref/fate/swr-rematrix-7.1.2-7.1
tests/ref/fate/swr-rematrix-7.1.2-mono
tests/ref/fate/swr-rematrix-7.1.2-stereo
tests/ref/fate/swr-rematrix-7.1.4-5.1
tests/ref/fate/swr-rematrix-7.1.4-5.1.2
tests/ref/fate/swr-rematrix-7.1.4-5.1.4
tests/ref/fate/swr-rematrix-7.1.4-7.1
tests/ref/fate/swr-rematrix-7.1.4-7.1.2
tests/ref/fate/swr-rematrix-7.1.4-mono
tests/ref/fate/swr-rematrix-7.1.4-stereo
tests/ref/fate/swr-rematrix-7.2.3-5.1
tests/ref/fate/swr-rematrix-7.2.3-5.1.2
tests/ref/fate/swr-rematrix-7.2.3-5.1.4
tests/ref/fate/swr-rematrix-7.2.3-7.1
tests/ref/fate/swr-rematrix-7.2.3-7.1.2
tests/ref/fate/swr-rematrix-7.2.3-7.1.4
tests/ref/fate/swr-rematrix-7.2.3-mono
tests/ref/fate/swr-rematrix-7.2.3-stereo
tests/ref/fate/swr-rematrix-9.1.4-5.1
tests/ref/fate/swr-rematrix-9.1.4-5.1.2
tests/ref/fate/swr-rematrix-9.1.4-5.1.4
tests/ref/fate/swr-rematrix-9.1.4-7.1
tests/ref/fate/swr-rematrix-9.1.4-7.1.2
tests/ref/fate/swr-rematrix-9.1.4-7.1.4
tests/ref/fate/swr-rematrix-9.1.4-7.2.3
tests/ref/fate/swr-rematrix-9.1.4-mono
tests/ref/fate/swr-rematrix-9.1.4-stereo
tests/ref/fate/swr-rematrix-9.1.6-5.1
tests/ref/fate/swr-rematrix-9.1.6-5.1.2
tests/ref/fate/swr-rematrix-9.1.6-5.1.4
tests/ref/fate/swr-rematrix-9.1.6-7.1
tests/ref/fate/swr-rematrix-9.1.6-7.1.2
tests/ref/fate/swr-rematrix-9.1.6-7.1.4
tests/ref/fate/swr-rematrix-9.1.6-7.2.3
tests/ref/fate/swr-rematrix-9.1.6-9.1.4
tests/ref/fate/swr-rematrix-9.1.6-mono
tests/ref/fate/swr-rematrix-9.1.6-stereo
tests/ref/fate/swr-rematrix-stereo-mono

Change #277605

Category	ffmpeg
Changed by	qaq03101 <1440810736ohnoyoudont@qq.com>
Changed at	Sun 09 Aug 2026 19:24:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9c176644aadb80ff1dfc65dc9d73f132ab81c999

Comments

avformat/scd: reject zero-channel tracks
A zero channel count can cause a division by zero in
scd_read_packet().

Reported-by: sdjasj
Fixes #23181.
Signed-off-by: qaq03101 <1440810736@qq.com>

Changed files

libavformat/scd.c

Change #277606

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 09 Aug 2026 20:01:31
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0f7eec026cb79300a460acb49fb03769abfbce5d

Comments

avcodec/cbs_lcevc: don't return an error if a Process Block is missing padding bits
Samples made following the current (for now) faulty specification may not
include alignment bits at the end of VUI parameters, so instead of erroring
out, warn about it and keep going.
In a writing scenario, like the lcevc_metadata bsf, the output bitstream will
be written with them.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/cbs_lcevc_syntax_template.c

Change #277624

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4af33ec41801b9834fbe6772f9c1e53c48e0cd0d

Comments

swscale/graph: ensure intermediate buffers have extra line padding
This basically guarantees that optimized packed routines have enough
space to write a few bytes past the end of the line.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c

Change #277625

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6b38a4035ab9250d9e04cf211675f87fe279c6c0

Comments

swscale/graph: add plane copy metadata
This can be set opportunistically by passes which are effectively
memcopies. Information about end-to-end plane copies will also propagate
upwards to the SwsGraph, and can be used there by the caller.

The reason we need to solve it this way is deeply tied to the way
SwsGraph is designed; specifically the fact that it can be used with
arbitrary frames and also lives per-field (rather than per-frame),
meaning that we can't just directly mutate the output frame somehow from
the run() call to ref the output planes.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c
libswscale/graph.h

Change #277626

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a167248c4d9caea103ec01d26f45bde246f4ba38

Comments

swscale/ops_memcpy: omit memcpy() if src and dst are identical
This allows already referenced planes to be skipped (and avoids UB in
this case).

Note that this is already safe for the other backends, because reading
from and writing to the same pointer via e.g. AVX operations is also
effectively a no-op.

See-Also: 1e071c8585467539d48cb33f239811309eed46a0
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_memcpy.c

Change #277627

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	49dd7666d1c1f455f627765b2e1c49e2e17caebf

Comments

swscale/ops: use unzwizzled component index for SwsComps.dep_in[]
This is more true to the name; these dependencies track *components*,
not planes - in both directions.

Generates a bunch of benign diffs.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops.c
tests/ref/fate/sws-ops-list

Change #277628

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd20a6f43c84ba2c5c5b428efa91d39a78f0ba36

Comments

swscale/ops: generate plane copy map for passes
Uses the previously added SWS_COMP_COPY to determine which planes are
direct copies of other planes.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/ops_dispatch.c

Change #277629

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d3d74f0ce452918e5c3f3123395c2c7f5d9d6636

Comments

swscale/aarch64/ops: set correct mask for SWS_OP_MIN/MAX
This was already the established convention, aarch64 just didn't follow it.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/aarch64/ops_impl_conv.c

Change #277630

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	770a1ef28de945b894c19c1066692b614f3de1f7

Comments

swscale/ops_optimizer: eliminate unneeded SWS_OP_MIN/MAX components
The current code only checks to see if the entire operation is a no-op,
but doesn't allow removing unneeded components from the operation.

Results in a large number of minor improvements, e.g.

 rgba 16x16 -> ya8 16x16:
   u8_read_packed_xyzw
   u8_to_f32_xyzw
   f32_linear_x_xxx00
   f32_dither_x_0_16x16
-  f32_min_xw
+  f32_min_x
   f32_to_u8_xw
   u8_permute_xyz_y_w
   u8_write_packed_xy

or:

 rgba 16x16 -> yuva444p10le 16x16:
   u8_read_packed_xyzw
   u8_to_f32_xyzw
   f32_linear_xyzw_xxx0x_xxx0x_xxx0x_000x0
   f32_dither_xyzw_0_3_2_5_16x16
-  f32_min_xyzw
+  f32_min_w
   f32_to_u16_xyzw
   u16_write_planar_xyzw

And some major ones, e.g.

 yuva444p 16x16 -> gbrap 16x16:
-  u8_read_planar_xyzw
-  u8_to_f32_xyzw
+  u8_read_planar_x
+  u8_write_planar_x
+ Sub-pass #1:
+  u8_read_planar_xyz
+  u8_to_f32_xyz
   f32_linear_xyz_x0x0x_xxx0x_xx00x
   f32_dither_xyz_0_3_2_16x16
-  f32_max_xyzw
-  f32_min_xyzw
-  f32_to_u8_xyzw
-  u8_write_planar_xyzw
+  f32_max_xyz
+  f32_min_xyz
+  f32_to_u8_xyz
+  u8_write_planar_xyz

Which is now split between two subpasses, one for the no-op alpha
copy and one for the yuv444 -> gbrp conversion. This has been
previously blocked by the SWS_OP_MIN/MAX clamp on the RGB channels
marking the alpha channel as dirty, even though it should be a no-op
on the alpha channel.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/aarch64/ops_entries.c
libswscale/ops_optimizer.c
libswscale/uops_macros.h
tests/ref/fate/sws-ops-list

Change #277631

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:03:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5ca18fdde49d97b7d4c194fab1c9550bb606689d

Comments

swscale/graph: internally ref copied intermediate buffers
If the pass we are allocating output buffers for has a plane copy map,
we can optimize away the internal buffer by setting it to a ref of its
input buffer, if one exists.

In theory, we can also do this if there is not an input buffer, by
directly referencing a sentinel or placeholder SwsPassBuffer
corresponding to the original input image, but this will require a bit
more work so I have decided to hold off on it for now.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/graph.c

Change #277632

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Sun 09 Aug 2026 21:05:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6bbc22dc09c214b2f5334afa30167fa1990eb5df

Comments

swscale: partially ref copied planes
Instead of all-or-nothing, this commit allows ref'ing some planes while
copying others, by referencing the SwsGraph plane copy map.

Notable examples:

yuv444p 1920x1080 -> yuva444p 1920x1080, flags=0x0 dither=1 scaler=0/0
  time=364/10 us (ref=4201/10 us), speedup=11.541x faster
yuv444p16le 1920x1080 -> yuva444p16le 1920x1080, flags=0x0 dither=1 scaler=0/0
  time=1140/10 us (ref=9651/10 us), speedup=8.466x faster
gray 1920x1080 -> yuvj444p 1920x1080, flags=0x0 dither=1 scaler=0/0
  time=894167/10000 us (ref=1942384/10000 us), speedup=2.172x faster
gbrap 1920x1080 -> yuva444p 1920x1080, flags=0x0 dither=1 scaler=0/0
  time=5670793/10000 us (ref=6572410/10000 us), speedup=1.159x faster

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/swscale.c

Change #277708

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:45:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	33d5616fbeb9c0581c6a25a624aae8195335815b

Comments

ffbuild: show install destinations in the INSTALL build output
The INSTALL pretty-printer echoed the rule prerequisites ($^), which
printed duplicate and phony names and never showed the destination.

Convert the install recipes to an INSTALL_FILES helper that receives the
install options, source files and destination as separate arguments. In
quiet mode it prints one "source -> destination" line per installed
file.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

Makefile
doc/Makefile
doc/examples/Makefile
ffbuild/common.mak
ffbuild/library.mak
fftools/Makefile

Change #277720

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:19
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c22a84c655c9bb11c82c232a5bd9b2f994a57fcc

Comments

avfilter/vsrc_amf: drop WINMMAPI from function pointer typedefs
The dllimport attribute is meaningless on a pointer typedef and the
symbols are resolved with dlsym() anyway. Fixes -Wignored-attributes.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavfilter/vsrc_amf.c

Change #277721

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:19
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d3fbc134295b1279d9c0c4a4b45043030c99ba9c

Comments

fftools/ffplay: annotate intentional switch fall-throughs
Both fall-throughs are deliberate: SDL_MOUSEBUTTONDOWN shares the cursor
and seek handling of SDL_MOUSEMOTION, and a resize also needs the refresh
done by SDL_WINDOWEVENT_EXPOSED. Silences -Wimplicit-fallthrough.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

fftools/ffplay.c

Change #277722

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd073ba92facc2fafef001f2cf6950d6d0bb1fc9

Comments

tools/seek_print: mark usage() as av_noreturn
It always calls exit(). Fixes -Wimplicit-fallthrough.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/seek_print.c

Change #277723

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	483ee7c7fac2f7c09edd1d5048d5f20fc8d63bf6

Comments

avcodec/ac3dec: mark the E-AC-3 prototypes as av_unused
Fixes -Wunused-function in checkheaders.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/ac3dec.h

Change #277724

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cb366a4af7a5f0a5333dc517f241ff438db08ca5

Comments

avcodec/lcevcdec: include pixfmt.h for enum AVPixelFormat
Fixes -Wvisibility in checkheaders.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/lcevcdec.h

Change #277725

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ea0ebb6e5cda3103d6a9ff2aa7c39e92601155e4

Comments

avcodec/scpr: mark the shared decode_run helpers as av_unused
Fixes -Wunused-function in checkheaders.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/scpr.h

Change #277726

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 19:52:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	87014aad6b73ca7f30b9fbff73a670418c810dd8

Comments

avutil/crc_internal: mark xnmodp as av_unused
Fixes -Wunused-function in checkheaders.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/crc_internal.h

Change #277727

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 20:22:42
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7432fed62bf0c07a45a82b874197decbb080df66

Comments

avcodec/vaapi_av1: stop setting deprecated VA-API fields
chroma_sample_position is a single bit and cannot hold the 2-bit AV1
value, so CSP_COLOCATED was silently truncated to CSP_UNKNOWN. tg_start
and tg_end are meaningless because the bitstream is submitted with per
tile granularity. libva deprecated all three without a replacement.
Fixes -Wdeprecated-declarations.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/vaapi_av1.c

Change #277728

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 20:29:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	76cfcc1692a39a21f2c651ff965bed260b7cac63

Comments

fftools/ffmpeg_mux_init: don't split forced keyframe spec in place
The spec string is owned by the option list and re-read for every stream
matching the same -force_key_frames, so terminating entries in place
truncated it for later streams.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

fftools/ffmpeg_mux_init.c

Change #277729

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 20:34:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9674156cc6d008a05a4c6c06470f0284a892959d

Comments

avutil/hwcontext_opencl: don't fall through between mapping types
When a mapping type is unavailable the case fell into the next hardware
type's mapping function instead of returning ENOSYS.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_opencl.c

Change #277730

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 10 Aug 2026 20:34:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f25dd4e26f64b84cdd586c16dc522c7f023dafa3

Comments

avutil/hwcontext_qsv: fix fall-through in qsv_fixed_pool_map_to()
The break only applied to the matching case, so every non-matching
surface fell into the next case and reinterpreted the same mfxHDLPair
under another API.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_qsv.c

Change #277748

Category	ffmpeg
Changed by	adorn <adornohnoyoudont@hr.de>
Changed at	Mon 10 Aug 2026 23:59:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7c2853dfc1d328b0dd98b7915e589eb1e8dc15a1

Comments

avformat/demux: restore container color fields individually
Fixes issue #23717

The upstream change 9b709532d5 ("avformat/demux: don't overwrite
container level color information if set") restores color_primaries,
color_trc and color_space from the container as an all-or-nothing
group: as soon as one of the three is signaled at container level,
all three are taken from the container.

MXF files written by some devices only carry the TransferCharacteristic
UL (MXF local tag 0x3210) but no ColorPrimaries (0x3219) or
CodingEquations (0x321a). For an H.264 essence whose VUI correctly
signals bt709 for all three, the grouped logic overwrites
color_primaries and color_space with the container's UNSPECIFIED
values, so ffprobe reports:

    color_space=unknown / color_primaries=unknown / color_transfer=bt709

instead of the bitstream's bt709/bt709/bt709.

Restore each of the three fields only when it is actually set at the
container level. This keeps container priority where a value is
signaled and falls back to the bitstream (decoder/parser) value
otherwise.

Changed files

libavformat/demux.c

Change #277749

Category	ffmpeg
Changed by	Ayoub Nabil <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 00:01:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5ba2a2c841259a04742064da5e79e288f50ef9fa

Comments

avcodec/ac3dec_fixed: fix dither at exponent 24
the fixed decoder discarded fractional coefficient data before the imdct, causing low-level coefficients to be truncated at large exponents. dithered bap 0 coefficients at exponent 24 were consequently quantized with a negative bias.

keep ac-3 coefficients in q2, compensate during windowing, and round the affected dither symmetrically. keep the forced e-ac-3 path in q0.

add a synthetic fate test covering the corrected output.

Changed files

libavcodec/ac3dec.c
libavcodec/ac3dec_fixed.c
tests/fate/ac3.mak
tests/ref/fate/ac3-fixed-dexp24

Change #277750

Category	ffmpeg
Changed by	Aleksander Shveikin <alex.svekohnoyoudont@gmail.com>
Changed at	Tue 11 Aug 2026 00:05:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b9161b93b5c88bd2f6f110bb527336af9a4c0cae

Comments

avcodec/x86/vp8dsp: convert mmx to sse2
This patch converts ff_vp8_idct_dc_add4uv_mmx to ff_vp8_idct_dc_add4uv_sse2.

Here are the checkasm benchmark results:

vp8_idct_dc_add4uv_c:    359.9
vp8_idct_dc_add4uv_mmx:   59.7 ( 6.02x)
vp8_idct_dc_add4uv_sse2:  21.1 (17.06x)

Signed-off-by: Aleksander Shveikin <alex.svek@gmail.com>

Changed files

libavcodec/x86/vp8dsp.asm
libavcodec/x86/vp8dsp_init.c

Change #277752

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 11 Aug 2026 00:44:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b5402d1ecf90343e26679fdfe258de3a490fbb43

Comments

avformat/hls: measure playlist reload interval from load start
RFC 8216 6.3.4 measures reload waits from the time the client began
loading the playlist, not from when the load finished.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c

Change #277753

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Tue 11 Aug 2026 00:44:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ee498f5e82d281851f9468bcbe427ead342894eb

Comments

avformat/hls: refresh stale playlist when seeking
Seeking used only the loaded segment list. Live and EVENT playlists can
gain segments after loading, and sliding-window playlists can also
expire them, so a seek target may lie outside the loaded window even
when it is within the actual presentation. Refresh the target playlist
before resolving the seek when the target may fall outside the window.
Rate-limited to one-half the target duration, the minimum reload
interval of RFC 8216 6.3.4.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/hls.c

Change #277756

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Tue 11 Aug 2026 03:47:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	06a5418aa53d8596edcecc5b7b68c544f9a36977

Comments

tests/checkasm/h264pred: Remove declare_func_emms
This is possible after the last remnants of MMX is gone

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

tests/checkasm/h264pred.c

Change #277758

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 04:35:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	54cab38faba72ac1bc2ccb2defd21fb78d9b5934

Comments

avutil/x86/tx_float: Don't use vextractf128 to write lower lane
Just use vmovaps with an xmm register. This is faster on some systems
and in any case saves codesize (by 688B).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/tx_float.asm

Change #277759

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 04:35:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	29e47b7d9b9b8c0532bb568582f60dff507320a8

Comments

avcodec/x86/dct32: Don't use v{extract,insert}f128 for lower lane
One can just access the lower lane as an ordinary xmm register.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/dct32.asm

Change #277760

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 04:35:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	03dc244a693ce639cebf82f7bae112fb75580919

Comments

avcodec/x86/dct32: Don't use legacy SSE in AVX function
Use VEX encoding instead.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/dct32.asm

Change #277851

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 16:12:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	844e10e1a74929c27dfe0aac1c34e92bb6edbf5a

Comments

avformat/shared: Use correct printf specifier
%zu expects a size_t, yet the argument is of type int64_t;
this is UB if size_t unless size_t is the corresponding unsigned
type to int64_t (i.e. it is UB on 32bit systems).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/shared.c

Change #277861

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 20:20:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d9ece3f721ffd31bb3c1a24484ff05726d66ae6d

Comments

avcodec/cavs: Avoid loads for initializing motion vectors
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/cavs.c
libavcodec/cavs.h
libavcodec/cavsdata.c
libavcodec/cavsdec.c

Change #277866

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 21:16:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e090671b9b521b89940cd5e6706a05b793ec5927

Comments

avcodec/h263dsp: Use ptrdiff_t for stride
Reviewed-by: Rémi Denis-Courmont <remi@remlab.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/h263.c
libavcodec/h263dsp.c
libavcodec/h263dsp.h
libavcodec/mips/h263dsp_mips.h
libavcodec/mips/h263dsp_msa.c
libavcodec/riscv/h263dsp_init.c
libavcodec/x86/h263_loopfilter.asm
libavcodec/x86/h263dsp_init.c
tests/checkasm/h263dsp.c

Change #277867

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 21:17:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a9c0652193ee4872678cdf88e4f322ba7a5cabf5

Comments

avcodec/x86/h263dsp_loopfilter: Avoid shuffle
Also start splatting a bit earlier.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h263_loopfilter.asm

Change #277868

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 11 Aug 2026 21:17:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7b05d16ce71d7525493354f2f725cded9dbc05d2

Comments

avcodec/mpeg4videodsp: Use ptrdiff_t for stride
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/mpeg4videodsp.c
libavcodec/mpeg4videodsp.h
libavcodec/ppc/mpeg4videodsp.c
libavcodec/x86/mpeg4videodsp.c
tests/checkasm/mpeg4videodsp.c

Change #277869

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 23:12:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1c10bcc2e17255dacb717a25ab3db142ce390602

Comments

avformat/librist: honor the caller buffer size in librist_read
librist_read() ignored its size argument and copied the full payload_len,
overflowing a smaller destination (e.g. via the async: wrapper). Clamp the
copy to the caller-provided buffer size.

Fixes: out of array access

Changed files

libavformat/librist.c

Change #277871

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 11 Aug 2026 23:23:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cf39fd2fa8d6ada4c48c0e9effc6d4ea974c7dfd

Comments

avformat/mpegenc: pass buffer size into put_system_header()
Fixes: out of array access
Fixes: many.mkv

Found-by: Joshua Rogers <joshua.rogers@aisle.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/mpegenc.c

Change #277872

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 23:23:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9d786e4b5e9b8482651928574de33772aeee7be1

Comments

avformat/mpegenc: reject stream counts that overflow the system header
put_system_header() writes 12 + 3*N bytes after the pack header into a
fixed 128-byte stack buffer, but is handed a PutBitContext sized past the
real buffer, so its own bounds check never fires; ~35+ streams overflow the
stack. Reject at mux init when the system header would not fit.

Fixes: out of array access
Fixes: many.mkv

Changed files

libavformat/mpegenc.c

Change #277873

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 23:25:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	acf5d7cdc1f9ae8752c23e1ea8d7f355ed780781

Comments

avformat/hevc: reject hvcC NAL arrays that overflow the 16-bit count
numNalus is uint16_t; a crafted hvcC declaring >65535 NAL units of one
type wraps it to 0 and then writes nal[-1]. Reject before the count can
wrap. Reachable by remuxing a crafted file with -c copy.

Fixes: integer overflow
Fixes: out of array access

Changed files

libavformat/hevc.c

Change #277874

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Tue 11 Aug 2026 23:33:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c736c6d10f58c28d1fbb1c7e1b50487015e4dd3e

Comments

doc/developer.texi: More precisse wording for Robustness
Fairy takes this literally

See: #23985

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

doc/developer.texi

Change #277876

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 23:37:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	65b0dab903e5975e036b30ecc58f5935d4f151e0

Comments

avformat/dashdec: reject a negative fragment index
A live manifest whose startNumber decreases across a refresh drives
cur_seq_no negative in move_segments(); get_current_fragment() only checked
the upper bound before indexing fragments[]. Add a lower-bound check and
clamp the negative delta at its source.

Fixes: out of array read

Changed files

libavformat/dashdec.c

Change #277879

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Tue 11 Aug 2026 23:50:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1cdeb3c4e7f1f8566d846b9b451e01c376398818

Comments

avformat/rtpenc_vc2hq: reject data units larger than the RTP payload buffer
send_packet() copied an input-derived unit/fragment size into the fixed
rtp_ctx->buf with no bound, overflowing it for a crafted Dirac unit even at
the default packet size. Reject units that do not fit in max_payload_size.

Fixes: out of array access

Changed files

libavformat/rtpenc_vc2hq.c

Change #277883

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 12 Aug 2026 00:09:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	715d7d85d690ebf9ae4d8f9f0e5ed8bf42ae8a43

Comments

swresample/rematrix: Use AVChannel in av_channel_layout_index_from_chan
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswresample/rematrix.c
tests/ref/fate/swr-rematrix-22.2-7.1.2

Change #277884

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 12 Aug 2026 00:09:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a205ebf3a6a1ea14a7fb1d3fa07ccae5fd297dd

Comments

swresample/rematrix: Avoid repeated calls to av_channel_layout* funcs
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswresample/rematrix.c

Change #277885

Category	ffmpeg
Changed by	Anxious Koisi <ddoaoingohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 01:14:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f92584948489dfcc1ec36612455617a7f7700840

Comments

avformat/gopher: Fix CRLF injection
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/gopher.c

Change #277886

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Wed 12 Aug 2026 01:34:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	983dae9c19f46c87d597598c0fd2f2fcee0ad2f8

Comments

avformat/rtpenc_av1: bound OBU size in the keyframe search loop
The is_keyframe sequence-header search loop advanced buf_ptr/rem_size by
num_lebs + obu_size without bounding obu_size against the remaining data
(unlike the main packetization loop). A crafted obu_size (~0x80000010)
wraps the signed rem_size back positive, so the next iteration
dereferences a pointer past the packet. Consume the LEB bytes first, then
reject an OBU larger than the remaining size. Out-of-bounds read reachable
from a crafted AV1 packet muxed to RTP.

Fixes: out of array read

Changed files

libavformat/rtpenc_av1.c

Change #277887

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 01:34:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d862937c5a9bd6099e46f551846e91a048ed65a1

Comments

avformat/rtpenc_av1: Check num_lebs
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/rtpenc_av1.c

Change #277888

Category	ffmpeg
Changed by	Joshua Rogers <MegaManSecohnoyoudont@users.noreply.github.com>
Changed at	Wed 12 Aug 2026 01:34:41
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b4c199c5906ff53368926c2a5839881f41957e7f

Comments

avformat/rtpenc_av1: do not narrow the OBU size to (long)
(long)obu_size sign-flips values in 0x80000000..0xfffffffd on ILP32/LLP64
targets (32-bit long, e.g. 64-bit Windows), bypassing the size check.
Compare as uint32_t; frame_size is non-negative at this point.

No PoC as the used setup was 64bit specific

Changed files

libavformat/rtpenc_av1.c

Change #277889

Category	ffmpeg
Changed by	Umar Pathan <helloohnoyoudont@umar.ac>
Changed at	Wed 12 Aug 2026 02:23:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cfaaf0062c085433ccb500689f137849794aeb63

Comments

avformat/webp_anim_dec: use ffio_read_size for ICCP and EXIF chunks
Fixes: use of uninitialized memory

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changed files

libavformat/webp_anim_dec.c

Change #277891

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 03:05:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a681001244b697e33fb94557dfd3f924250edb8c

Comments

avfilter/afir: bound the crossfades by the samples of the input frame
Fixes: ffmpeg -i tail.wav -i ir.wav -filter_complex "[0][1]afir=dry=1:wet=1:enable='lte(t,0.5)'" -f null -

Fixes: out of array access
Fixes: uQcNui1HxUqe
Regression since: 1835f884b7ae475d2079d044003966981823502e
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

libavfilter/afir_template.c

Change #277892

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 03:50:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4d896a22dd440dc886fd4c0d89992a55e886390b

Comments

avfilter/spectrumsynth: negotiate one pixel format for both inputs
Fixes: out of array access
Fixes: ndSS0e433gq
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

libavfilter/vaf_spectrumsynth.c

Change #277893

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 04:02:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c42ac5c5668fef93f84960a1d8bf0605bdd812ea

Comments

avformat/mpegts: keep the PES payload within max_packet_size
Fixes: out of array access
Fixes: eYvYXUJZbPJ
Regression since: bca30570d28bbaa07badadabf55ec3589201a82f
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

libavformat/mpegts.c

Change #277894

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 04:02:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	37721094ab4bac1215168b5b8a7defa1bc2d6581

Comments

avformat/mpegts: reject a max_packet_size below one TS payload
Fixes:  ffmpeg -max_packet_size 100 -i fate-suite/mpegts/h264small.ts -f null -

Fixes: out of array access
Fixes: eYvYXUJZbPJ
Regression since: bca30570d28bbaa07badadabf55ec3589201a82f
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

doc/demuxers.texi
libavformat/mpegts.c

Change #277895

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 04:34:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	73130e77dc2e866386088e8e4edf7a051fd2949a

Comments

avformat/mlv: end the LJ92 packet where its data ends
Fixes: read of uninitialized memory
Fixes: PJaSVrRTt6Kf
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

libavformat/mlvdec.c

Change #277896

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 04:50:23
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	98e45d60746e25fc4a5529bd6ffb8ae7f28f6ce8

Comments

avcodec/lcldec: clear what the multithread chunks leave undecoded
Fixes: read of uninitialized memory
Fixes: zTrrhikJHTRR
Found-by: Eunsoo Kim, Microsoft FORGE Labs

Changed files

libavcodec/lcldec.c

Change #277976

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 14:39:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e5ecfe8970899efa6ac0447fde4cb7c36ec23b75

Comments

avformat/id3v2: make legacy COMM descriptor-as-key behavior opt-in
A COMM frame carries both a language and a short content descriptor, so its
only unambiguous metadata key is "comment-<descriptor>-<lang>", while the
extra export of the bare descriptor as a key silently overrides a real tag
whenever the descriptor matches a known tag name such as "album".

Restrict that legacy export to the container-level ID3v2 tag of callers that
ask for it through the new AVFMT_FLAG_LEGACY_ID3V2_COMM_KEYS flag (-fflags
+legacy_id3v2_comm_keys), which is deprecated along with
FF_API_OLD_ID3V2_COMMENT and only exists to give those callers a transition
path.

Changed files

doc/APIchanges
libavformat/aacdec.c
libavformat/avformat.h
libavformat/demux.c
libavformat/hls.c
libavformat/id3v2.c
libavformat/id3v2.h
libavformat/mpegts.c
libavformat/options_table.h
libavformat/version.h
tests/fate/id3v2.mak
tests/ref/fate/id3v2-comm-dashed-descriptor
tests/ref/fate/id3v2-comm-descriptor
tests/ref/fate/id3v2-comm-descriptor-no-lang
tests/ref/fate/id3v2-comm-foo-lang-as-descriptor
tests/ref/fate/id3v2-comm-invalid-lang
tests/ref/fate/id3v2-comm-lang-as-descriptor
tests/ref/fate/id3v2-comm-long-descriptor
tests/ref/fate/id3v2-comm-multi-invalid-lang
tests/ref/fate/id3v2-comm-sort
tests/ref/fate/id3v2-comm-sort-
tests/ref/fate/id3v2-comm-sort-eng
tests/ref/fate/id3v2-reenc-delete-metadata-keep
tests/ref/fate/id3v2-reenc-delete-metadata-keep-format
tests/ref/fate/id3v2-reenc-remux-keep
tests/ref/fate/id3v2-wma-comm
tests/ref/fate/id3v2-wma-comm-asf_o
tests/ref/fate/id3v2-wma-comm-legacy-keys

Change #277977

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Wed 12 Aug 2026 14:57:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	82d03c6d8e1a16182b3db873eada6b968996437c

Comments

tests/checkasm: Give the ttaencdsp function a separate name
Otherwise, if running both ttadsp and ttaencdsp tests in the same
checkasm run, we get failures as the two functions don't match.

This fixes running the "checkasm" binary without any "--test"
parameter.

Changed files

tests/checkasm/ttadsp.c

Change #277996

Category	ffmpeg
Changed by	Brad Smith <bradohnoyoudont@comstyle.com>
Changed at	Wed 12 Aug 2026 17:19:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9432f85830f2c0a453c6caa9f4353cef208eab8a

Comments

avcodec/mips/cabac: Fix building on OpenBSD/mips64
src/libavcodec/mips/cabac.h:160:77: error: instruction requires a CPU feature not currently enabled
  160 |         PTR_SUBU   "%[tmp1],         %[c_low],        %[tmp1]             \n\t"
      |                                                                             ^
<inline asm>:13:2: note: instantiated into assembly here
   13 |         movz       $15,          $13,          $24
      |         ^

OpenBSD targets MIPS III ISA.

Changed files

libavcodec/mips/cabac.h

Change #277997

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	11536e78d22bfca6051d80d126e755b4bd45987a

Comments

checkasm/sw_ops: don't return a value from void functions
Fixes: e65dab2b59
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tests/checkasm/sw_ops.c

Change #277998

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	16ef68f90091f5195efe435aa9037c0a0c0f38c4

Comments

avutil/hwcontext_d3d12va: fix unused label warning with UWP
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_d3d12va.c

Change #277999

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e255b21831f927da0ca4fcfc96b3fb883beb70dd

Comments

tools/ismindex: fix const qualifier discard warning
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/ismindex.c

Change #278000

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d832b7ed05f047729d11571d6d9b6c83509920b3

Comments

avutil/timer: fix ignored read() result
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/timer.h

Change #278001

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	51699b70cd75f4c00bffa4b03cd93b1d50787fd5

Comments

checkasm/sbcdsp: rename CONST macro to avoid clash with windows.h
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tests/checkasm/sbcdsp.c

Change #278002

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5656e9e5ab6a58e0a9c30510040b4724e6ff04ab

Comments

avcodec/tests/x86/dct: fix ff_prores_idct_put_10 declaration mismatch
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/tests/x86/dct.c

Change #278003

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	40bb61c0d440af15567b65692f1bc28e54c9bb47

Comments

avutil/log: use the ANSI color table in ansi_fputs() on Windows
ansi_fputs() is reachable on Windows when stderr is not a console
or with forced color. Use real ANSI color table in this cases, otherwise
wrong values are used from Windows codes.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/log.c

Change #278004

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f73d12f6ee9d9f72fe21a6842eb3b8961c46d7af

Comments

avformat/rtmppkt: add fall-through annotations
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/rtmppkt.c

Change #278005

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3f596e2035786bd407cc50c60edefa33a3fce3c5

Comments

avformat/tls_securetransport: add fall-through annotation
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/tls_securetransport.c

Change #278006

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	40ed144d3adb4a792c7d531c64c77e299dcb3518

Comments

avcodec/riscv/vvc: fix implicit fall-through into no-op default cases
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/riscv/vvc/dsp_init.c

Change #278007

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5583534fc2022fc021e44544b40c2be737dde3b7

Comments

swscale/cms: remove no-op const from return type
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libswscale/cms.c

Change #278008

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9d9b347de7a37153bfb39de05ec4261e1d6d1ada

Comments

fftools,tools: drop stray semicolons after macro-defined functions
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

fftools/ffmpeg_filter.c
fftools/ffmpeg_opt.c
tools/crypto_bench.c

Change #278009

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e7b9c5fe779ad0497881d466d0c3692c619300e3

Comments

avformat/wvdec: convert flags to defines
Don't use values that would extend emum data size.

Fixes: warning: enumerator value which exceeds the range of 'int' is a
       C23 extension (2147483648 is too large)

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavformat/wvdec.c

Change #278010

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	debd8528afe797d0061ce40df913f30bf2eb7b0b

Comments

avcodec/wavpack: keep flags as unsigned value
This is bitfield with flags, keep it as unsigned.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/wavpack.c

Change #278011

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d6e5dcbb47198a961bf0dffdbd96b1015b418bb6

Comments

avutil/hwcontext_vulkan: suppress -Wtype-limits warning
When the optional instance extensions list is empty the unsigned loop
bound folds to 0, making the comparison always false. No functional
change.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_vulkan.c

Change #278012

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 12 Aug 2026 17:21:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	799cf2bc1658939432d3cd910c2e763541155e82

Comments

tools/sidxindex: fix const qualifier discard warning
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tools/sidxindex.c

Change #278021

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 20:05:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	645529cbd427b25cbdbe4b33b9a77509ff9399e0

Comments

avcodec/mips/hevcdsp: declare the biweight functions with the offset they take

Changed files

libavcodec/mips/hevcdsp_mips.h

Change #278022

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 20:05:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0a6145a344c11156f09b67b92654fcafd0b9e42c

Comments

avcodec/huffyuv: seed the median predictor endian independently

Changed files

libavcodec/huffyuv.h
libavcodec/huffyuvdec.c
libavcodec/huffyuvenc.c

Change #278023

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 12 Aug 2026 20:05:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4be657ffd84f6dcaf622a8ce04418a6b71bab2b3

Comments

avcodec/ppc/vp8dsp: store the 16 pixel wide rows at any alignment

Changed files

libavcodec/ppc/vp8dsp_altivec.c
libavutil/ppc/util_altivec.h

Change #278024

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Wed 12 Aug 2026 20:05:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	445cba7f4c9161a0c237d70cb986d7892abbc58c

Comments

forgejo/workflows: add cross fate tests for ppc64/mips64/riscv64
And a full-dependency linux-aarch64 one.

Changed files

.forgejo/workflows/test.yml

Change #278042

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 13 Aug 2026 00:29:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b397eba2f0d3d86daf1098d0f27daffccc74fea5

Comments

avcodec/h264dsp: Use ptrdiff_t for stride in idct functions
Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/aarch64/h264dsp_init_aarch64.c
libavcodec/aarch64/h264idct_neon.S
libavcodec/arm/h264dsp_init_arm.c
libavcodec/h264_mb.c
libavcodec/h264_mb_template.c
libavcodec/h264addpx_template.c
libavcodec/h264dsp.h
libavcodec/h264idct.h
libavcodec/h264idct_template.c
libavcodec/loongarch/h264dsp_lasx.c
libavcodec/loongarch/h264dsp_loongarch.h
libavcodec/loongarch/h264idct.S
libavcodec/loongarch/h264idct_loongarch.c
libavcodec/mips/h264dsp_mips.h
libavcodec/mips/h264dsp_mmi.c
libavcodec/mips/h264idct_msa.c
libavcodec/ppc/h264dsp.c
libavcodec/riscv/h264dsp_init.c
libavcodec/x86/h264_idct.asm
libavcodec/x86/h264_idct_10bit.asm
libavcodec/x86/h264dsp_init.c
tests/checkasm/h264dsp.c

Change #278078

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 13 Aug 2026 13:17:30
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a53f3f861388da37ac5c02397a7eee11b71c3fad

Comments

avcodec/svq1enc: Workaround GCC bug 102513
encode_blocks() in svq1enc.c triggers a bug in several versions
of GCC: GCC tries to create eight clones of recursive functions
to inline some parameters when it deems this worth it.
Yet encode_blocks() only has a recursion depth of six and
the last two clones contained out-of-bounds array accesses,
triggering -Warray-bounds warnings.

This has already caused problems in the past, see
894191e7e10520109db983032d1cd8d45c85af6d which disabled
the creation of clones for GCC < 12; GCC 12 itself got better
at not creating unnecessary clones and therefore was not targetted
by this.

Yet since GCC 16 (commit 3fd5a1e76bbf1bc031934a9d96f284a22a5f307f),
it again fails at discarding unneeded clones and emits warnings
again.

This patch fixes this by simply telling the compiler the proper
range via an av_assume(); this also allows to remove the old
workaround for GCC 11.

Furthermore, GCC 12-15 still produced insane output: Six clones
for encode_block() and a general function without inlined parameters
that is not referenced by anything. The av_assume() eliminates
this function, too.

Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/svq1enc.c

Change #278079

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 13 Aug 2026 13:17:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f49ed6235e13df957787dd062a1b270fbfa6642e

Comments

avcodec/svq1enc: Fix shadowing
Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/svq1enc.c

Change #278099

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:00:00
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bd68a6130495fe17bf33e0a9ddc053baae18c81f

Comments

avutil/opt: allow passing a fake object to av_opt_set()
The result will be a dry run of the process, where no value is set but all
the checks are made.
Useful for, as the AV_OPT_SEARCH_FAKE_OBJ flag states, run the relevant
process without having to allocate the corresponding object.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavutil/opt.c
libavutil/opt.h

Change #278100

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:03:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3776aadf4b12b82703eb01b3f4a1c0a2dd14e46d

Comments

avcodec/encode: add a function to gracefully reconfigure an encoder
In some cases there are compression and resource usage benefits from changing
certain encoder values at runtime over of closing, freeing, allocating and
initializing a whole new encoder. The most obvious case is rate control,
quality settings, and so on. As such, this commit introduces a new function
that attempts to reconfigure an opened encoder with a dictionary of options.

Private options that can be changed are flagged with the runtime AVOption flag.
Global options that can be changed, knowing that each encoder may support some
but not others and as such can't be flagged in the global AVOptions struct,
are signaled the same way through a new field in the FFCodecDefault struct.

An encoder may provide a callback function that will be called for the
reconfiguration process. Simple encoders that can work with only the AVOptions
being reset and don't require any extra steps may just let the default function
be called instead.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/APIchanges
libavcodec/avcodec.h
libavcodec/codec.h
libavcodec/codec_internal.h
libavcodec/encode.c
libavcodec/encode.h
libavcodec/version.h

Change #278101

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c13bf75fd320e8476a866993587f1b948c680eb3

Comments

avcodec/libx264: refactor encoder configuration functions
Will be useful in the next commit.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libx264.c

Change #278102

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	791183015837f7415da03ef564afe5985b9b3c3f

Comments

avcodec/libx264: add support for encoder reconfiguration
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libx264.c

Change #278103

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5fd1b8b1f63f34984ef485159473c2be101c046a

Comments

avcodec/libaomenc: refactor encoder configuration functions
Will be useful in the next commit.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libaomenc.c

Change #278104

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4f3b01a32d544bee608dfffd3693c735f3a2d062

Comments

avcodec/libaomenc: add support for encoder reconfiguration
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libaomenc.c

Change #278105

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a89f799bd8c9b52b5ba727aff5f9d02fd85d113

Comments

fftools/ffmpeg_filter: add AVOptions to OutputFilter
Preparation work for the following commits.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_filter.c

Change #278106

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f241d7bbed6cc0d558008515728521867931b899

Comments

fftools/ffmpeg_enc: split off parts of enc_open() into a separate function
Preparation work for the following commits.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_enc.c

Change #278107

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b14162c9d43262e35ed9462f8534ec5236300819

Comments

fftools/ffmpeg_enc: split off encoder flushing in encoder_thread() into a separate function
Preparation work for the following commits.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg_enc.c

Change #278108

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7d905abec6ab565ac3afbaa7840ed278a330eab2

Comments

fftools/ffmpeg_enc: store a few more AVCodecContext fields in Encoder
Preparation work for the following commits.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

fftools/ffmpeg.h
fftools/ffmpeg_enc.c
fftools/ffmpeg_mux_init.c

Change #278109

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 13 Aug 2026 16:04:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ca821e458aabe2fa211d9e94eac38cd69fe2ea09

Comments

fftools/ffmpeg: support reinitializing the encoder
With this, an encoding process can change certain values that until now were
settable during init only.
A graceful reconfiguration will be attempted first, which depends on the
capabilities of the underlying encoder. If that fails, a full encoder restart
is performed instead.

The reconfiguration parameters are passed as a list of target timestamp and
key=value pairs to be applied at that point during encoding. Future
improvements may also include targetting encoded frame number, and extending
the command feature to also support passing encoder runtime arguments and not
just filter runtime change arguments.

Example:

ffmpeg -i INPUT -reinit_opts:v:0 "pts=100000|crf=18,pts=200000|video_size=1280x720:crf=20" -crf 30 -c:v:0 libx264 OUTPUT

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/ffmpeg.texi
fftools/ffmpeg.c
fftools/ffmpeg.h
fftools/ffmpeg_enc.c
fftools/ffmpeg_filter.c
fftools/ffmpeg_mux_init.c
fftools/ffmpeg_opt.c

Change #278117

Category	ffmpeg
Changed by	yuanhecai <yuanhecaiohnoyoudont@loongson.cn>
Changed at	Thu 13 Aug 2026 21:25:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8b4fad11acfc958dfde29fb0799d3ca1818bbbf7

Comments

swscale/loongarch: fix buffer underflow in yuv2plane1_8_lsx/lasx
The old tail paths backed the source and destination pointers up by a
full vector and then advanced them by the remainder. When no full vector
had been processed, this placed dest before its allocation, followed by
a full-vector store. Process tails forward with LSX and byte stores and
use equivalent packed arithmetic so exactly dstW bytes are written.

Fixes: #23558
Reported-by: sebastinas

Changed files

libswscale/loongarch/output.S

Change #278119

Category	ffmpeg
Changed by	Haochen Chen <haochencohnoyoudont@google.com>
Changed at	Thu 13 Aug 2026 22:46:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	87f6fa9d41e4a31faeabd355a9c2365c6bb9465f

Comments

avcodec/bmp: fix misaligned read in 16-bit decoding
Fixes a misaligned read in bmp_decode_frame when decoding 16-bit BMPs.
The code was casting a potentially unaligned uint8_t pointer to uint16_t* and dereferencing it.
This triggers a UBSAN trap (SIGILL / UD1) on environments configured to trap on undefined behavior.
Replaced with AV_RL16 to perform safe unaligned reads.

Signed-off-by: Haochen Chen <haochenc@google.com>

Changed files

libavcodec/bmp.c

Change #278121

Category	ffmpeg
Changed by	Daniel Verkamp <danielohnoyoudont@drv.nu>
Changed at	Fri 14 Aug 2026 00:05:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1d7b14f61d66fdf18f15204c613df9d65396c319

Comments

avformat/wavenc: Omit JUNK for non-seekable -rf64 auto
The `-rf64 auto` option requires seeking in order to go back and convert
the reserved header space into a ds64 chunk when switching from RIFF to
RF64. If the output is not seekable, then it isn't possible to make use
of the space reserved with the JUNK chunk, so it is more reasonable to
just produce a regular RIFF with no JUNK chunk in this case, equivalent
to the `-rf64 never` option.

Signed-off-by: Daniel Verkamp <daniel@drv.nu>

Changed files

libavformat/wavenc.c

Change #278137

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 14 Aug 2026 10:41:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dab32dcc2129b80df7190cdfc1063b3940c711af

Comments

avutil/slicethread: add return value to slicethread callbacks
Several places in the code currently manually roll the return value
saving. This small signature change allows slicethread.c to collect
return values directly. Unlike the existing solutions which mostly
rely on collecting the return values in an array, we can just use an
atomic CAS to set the error iff no error has been set already.

We need a slight bit of extra complication to avoid breaking the ABI,
which requires keeping around the old names as deprecated aliases that
just set up the appropriate wrappers. However, fortunately, the way I
implemented the deprecated signatures allows callers to freely mix
avpriv_slicethread_execute() and avpriv_slicethread_execute2(), even
on contexts created with one or the other.

It's worth pointing out that this commit just changes the signatures,
but doesn't yet update the various to use the return value.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavcodec/pthread_slice.c
libavfilter/pthread.c
libavutil/slicethread.c
libavutil/slicethread.h
libswscale/cms.c
libswscale/graph.c
libswscale/swscale.c
libswscale/swscale_internal.h
libswscale/utils.c

Change #278138

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 14 Aug 2026 10:41:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	37083b3105acebcdfbd8b9e2046e971db5d108c2

Comments

swscale/utils: add ff_sws_thread_exec() helper
Simplified API that wraps slicethread create/execute/free, with
fallback for systems without threading support.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/swscale_internal.h
libswscale/utils.c

Change #278139

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 14 Aug 2026 10:41:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	183c14e20e8049b34f140bd0f5b4a7e3b24433e8

Comments

swscale: simplify slice_ctx error propagation
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/swscale.c
libswscale/swscale_internal.h
libswscale/utils.c

Change #278140

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 14 Aug 2026 10:41:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0c7b5b2b5cdf34ec3805984e44cfc520771873ab

Comments

swscale/ops: use slice threading for ff_sws_enum_op_lists()
Enabled by the existing SwsContext.threads option.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/op_list_gen_template.c

Change #278141

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Fri 14 Aug 2026 10:41:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2f0848c9bb49dd08b259489d774c0da98558d891

Comments

swscale/uops: use threading to speed up macros generation
26.74s -> 4.18s on my laptop

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libswscale/uops_macros_gen.c

Change #278162

Category	ffmpeg
Changed by	Araz Iusubov <Primeadviceohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 13:55:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0d5b77341fa30728bb51b26619a4933e053329b8

Comments

avfilter/vf_vqe_amf: remove unsupported opencl engine type

Changed files

libavfilter/vf_vqe_amf.c

Change #278166

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 14:02:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a0078c78262f9ad31e2fbaecc4e82b42ce636ac4

Comments

avcodec/h264_direct: move check before loading things

Changed files

libavcodec/h264_direct.c

Change #278167

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 14:02:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ead437865214972624a047840c8e9b6660a2ec85

Comments

avcodec/h264_direct: await both fields of a colocated field pair
When a frame coded picture uses a complementary field pair as its
colocated picture, ff_h264_direct_ref_list_init() selects the field whose
POC is closest to the current one and stores it in col_parity. The direct
mode setup then remaps mb_xy to that parity and reads the colocated
mb_type, motion_val and ref_index from it.

await_reference_mb_row() waits on the field given by "ref_field_picture &&
ref_field", which is 1 for any such reference, since the pair is
referenced as a frame and ref->reference is 3. Only the bottom field is
therefore awaited, while the data is just as likely to be read from the
top one. With frame threading the two fields are decoded by different
threads, so the colocated data can be read while the field it belongs to
is still being decoded, yielding stale motion vectors and a corrupt
frame. await_references() already handles the same case correctly for
motion compensation.

Fixes sporadic failures of fate-h264-conformance-cabac_mot_picaff0_full,
where display frame 14.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/h264_direct.c

Change #278173

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 14:44:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c48230eb86ff02246f6a14fa1475a0d9398363b4

Comments

avcodec/pthread_frame: sync decoded side data to all worker threads
Previously only the first thread's context and the user-facing one had
it, so frames decoded by the remaining worker threads were missing
global side data whenever the per-frame state could not provide it, e.g.
extradata-derived HDR metadata after avcodec_flush_buffers().

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/pthread_frame.c

Change #278186

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 15:31:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	763c3356f115b5186e6d5ad102717fc972ff0399

Comments

avcodec/codec: rename encoder reconf codec cap
It was commited only a day ago, and the new name is more accurate, so this change should be safe.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/APIchanges
fftools/ffmpeg_enc.c
libavcodec/avcodec.h
libavcodec/codec.h
libavcodec/encode.c
libavcodec/libaomenc.c
libavcodec/libx264.c
libavcodec/version.h

Change #278188

Category	ffmpeg
Changed by	Nicolas Gaullier <nicolas.gaullierohnoyoudont@cji.paris>
Changed at	Fri 14 Aug 2026 15:39:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6092f060f431f587b3d619a36d9b9d5f963d8b57

Comments

avcodec/h264_parser: align field_order default with the decoder
Use case: interlaced AVC Intra is typically missing the required information
for proper tff detection.
Currently, the decoder defaults to tff, but the parser defaults to progressive.

./ffmpeg -f lavfi -i testsrc=1920x1080 -codec libx264 -pix_fmt yuv422p10le \
    -x264-params avcintra-class=100:tff=1 -frames 1 tff.h264
./ffprobe tff.h264 -of flat -show_entries stream=field_order:frame=interlaced_frame,top_field_first

Before:
frames.frame.0.interlaced_frame=1
frames.frame.0.top_field_first=1
streams.stream.0.field_order="progressive"

After:
frames.frame.0.interlaced_frame=1
frames.frame.0.top_field_first=1
streams.stream.0.field_order="tt"

Issue since field_order setting in 3f1a7ceb2c604deff.

Signed-off-by: Nicolas Gaullier <nicolas.gaullier@cji.paris>

Changed files

libavcodec/h264_parser.c

Change #278206

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 18:26:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c9a585da5665433a1be14679683d7ff13fc6960d

Comments

lavf: Add myself as maintainer for the id3v2 files

Changed files

.forgejo/CODEOWNERS
MAINTAINERS

Change #278226

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 18:28:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0f3e1bf4fe189e82d79125e5ce8d5f8746bc2c8e

Comments

avcodec/tests/avcodec: Test reconf callback, flags
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/tests/avcodec.c

Change #278227

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 18:28:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	be3757f6e93e07c1d9322438a1d0ce7422427c35

Comments

avcodec/codec_internal: Put bsfs and reconf in a union
Reduces sizeof(FFCodec)

Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/codec_internal.h
libavcodec/tests/avcodec.c

Change #278228

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 18:28:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d10125319d6ed26a57006ce8a28c43b9a8b56b23

Comments

avcodec/encode: Remove pointless opt.h inclusion
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/encode.h

Change #278229

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 14 Aug 2026 21:08:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	85d2cfd52e679cb4e0e240456f74e1955cf91d12

Comments

libswresample/tests/rematrix: uninit channel layouts
release parsed channel layouts before exiting.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libswresample/tests/rematrix.c

Change #278230

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 14 Aug 2026 21:08:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d02bad0c903baf67891dbe6b87c6577d9a27c79f

Comments

swresample/rematrix: accept unused channels
custom layouts containing AV_CHAN_UNUSED were rejected by
swr_build_matrix2().

accept unused channels in input and output layouts. preserve their
physical positions, clear input columns, and leave output rows silent.

add fate coverage for both cases. this fixes #24094.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libswresample/rematrix.c
libswresample/swresample.h
libswresample/tests/rematrix.c
tests/fate/libswresample.mak
tests/ref/fate/swr-rematrix-unused-input
tests/ref/fate/swr-rematrix-unused-output

Change #278231

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Fri 14 Aug 2026 21:11:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c39bc13fdb492c3fec1bbdaf031bc41e250eedcf

Comments

swresample/tests/rematrix: print rows in layout order
Otherwise, custom layouts will be printed as if they were in native order.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libswresample/tests/rematrix.c
tests/ref/fate/swr-rematrix-unused-output

Change #278239

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	07f632c7faa48d63c28e044a7dd1a7646f8fd095

Comments

avcodec/x86/{h264_chromamc,rv40dsp}: Don't use PAVGB macro
It always just expands to pavgb as there are no pure mmx
functions here.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_chromamc.asm
libavcodec/x86/rv40dsp.asm

Change #278240

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a6b2f0a911d3c4c6544901c41efa737f07c5d863

Comments

avutil/x86/x86util: Remove PAVGB macro
It is no longer used anywhere and it would make no sense
to ever use it again.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278241

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	df547d9267303101a0232cd8d6f82b9f92ded877

Comments

avutil/x86/x86util: Remove unused PSWAPD
Unused since ec735579814b6b73e17da601ff011a918c49e40f.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278242

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	046e2bcc2f9772c738dd60b0aef07e4b8e57b546

Comments

avutil/x86/x86util: Remove ABSB{,2} macros
Seem to have always been unused.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278243

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	99d378fd0d0d198c1ae8075405a6d5266895a238

Comments

avutil/x86/x86util: Remove ABSD2 macro
Always unused, new code can just use SSSE3 pabsd directly.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278244

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7452eccc083a847e9703d03ac0407190d77fb248

Comments

swscale/x86/rgb_2_rgb: Remove useless macro
x86inc's VEX emulation can handle this just fine.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswscale/x86/rgb_2_rgb.asm

Change #278245

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	be1c5c065a4b96f5e5ecd009ab895e0cf6d16c3f

Comments

av*/x86/yadif,lossless-videodsp: Don't use LSHIFT, RSHIFT macros
There is no mmx code in these files.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/lossless_videodsp.asm
libavfilter/x86/vf_yadif.asm
libavfilter/x86/yadif-10.asm
libavfilter/x86/yadif-16.asm

Change #278246

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9a62712427b75ac56dfd9069f13c037b13143d15

Comments

avutil/x86/x86util: Remove unused LSHIFT, RSHIFT macros
Useless in the absence of mmx.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278247

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d2495911cce6b2cd0867ceaf617d87f6ab645f4a

Comments

avcodec/x86/ac3dsp: Avoid using PMINUB macro
Forgotten in fd98594a8831ce037a495b6d7e090bd8f81e83a1.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/ac3dsp.asm

Change #278248

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	145ec527df728bb877e59832b3e2da79e74f868b

Comments

avutil/x86/x86util: Remove unused PMINUB macro
Useless in the absence of mmx.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278249

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3956f2ffe3f239c24bfe29a0463512d96bca2b3d

Comments

av*/x86/resample,h264_intrapred: Avoid using PSHUFLW
It is useless in the absence of mmx.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_intrapred.asm
libswresample/x86/resample.asm

Change #278250

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7c1326b16c5cf4f89d9d64b838f7d9dfd53cafd

Comments

avutil/x86/x86util: Remove PSHUFLW macro
Useless in the absence of mmx(ext).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavutil/x86/x86util.asm

Change #278251

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	736550dc9bf4abdc13b6a1dfdc5992d7b183ab42

Comments

avutil/x86/x86util: Move broken PSIGNW macro to its only user
The non-SSSE3 version does not do what psignw does:
If the second operand is zero, the first is unchanged.
In any case, the end result is '(%1 and not %2)-(%2 and not %1)'.
So it only works in case %2 is -1, which is how vp9itxfm.asm
actually uses it. So move it there and rename it to NEGATE.

See also f2fd167835b6f039a593e46ab3a84e1b9a453660 for an instance
when this macro actually caused problems.

(Would pxor %2,%2 psubw %2,%1 SWAP %1,%2 not actually be better?)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9itxfm.asm
libavutil/x86/x86util.asm

Change #278252

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Fri 14 Aug 2026 21:21:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	950a96bd258b8251a516ed7bffda127fee299945

Comments

avcodec/x86/vp9itxfm: Avoid load
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp9itxfm.asm

Change #278255

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Sat 15 Aug 2026 02:43:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cb98f5c3dfa1ee64c8ede81346b012a5c7562c99

Comments

avcodec/x86/hevc/add_res: Replace MMX with SSE2
Convert hevc_add_residual_4_10 to use SSE2 and remove declare_func_emms
in test code.

  hevc_add_res_4x4_10_c:          21.7
  hevc_add_res_4x4_10_mmxext:      9.4 ( 2.31x)
  hevc_add_res_4x4_10_sse2:        9.2 ( 2.35x)

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/hevc/add_res.asm
libavcodec/x86/hevc/dsp.h
libavcodec/x86/hevc/dsp_init.c
tests/checkasm/hevc_add_res.c

Change #278256

Category	ffmpeg
Changed by	Marcos Ashton Iglesias <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Sat 15 Aug 2026 03:08:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bc4cbbb85f16ff6eadd6548bafa0d6a1da0c0d06

Comments

avutil/hdr_dynamic_metadata: fix from_t35 NULL return and version doc
av_dynamic_hdr_plus_from_t35() returned AVERROR(ENOMEM) when passed a
NULL AVDynamicHDRPlus. A NULL output struct is invalid API usage by the
caller, not an allocation failure, and nothing is allocated on that
path. av_dynamic_hdr_smpte2094_app5_from_t35() already returns
AVERROR(EINVAL) for the same condition; make the two parsers agree.

The documentation for AVDynamicHDRPlus.application_version said the
value shall be set to 0, which contradicts both sides of the
implementation: av_dynamic_hdr_plus_to_t35() ignores the field and
always writes 1, the value CTA-861-H specifies for the HDR10+ payload,
and av_dynamic_hdr_plus_from_t35() sets the field from the parsed
payload rather than validating it against 0. No FFmpeg code assigns
this field, so document the contract the implementation provides.

Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

libavutil/hdr_dynamic_metadata.c
libavutil/hdr_dynamic_metadata.h

Change #278262

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 10:14:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ef76da316d7a90bac3c8db92389aa59e9c864cfe

Comments

avformat/whip: Avoid calling strlen multiple times
Also avoid zeroing the buffer unnecessarily, ff_data_to_hex()
also initialized it (including the terminating \0).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/whip.c

Change #278263

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 10:14:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	44095007d00f908dbdeaadaccc50d7bdeff36a37

Comments

avformat/whip: Avoid allocation
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/whip.c

Change #278269

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:02:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a072b4961ac36fae1d4a62432d154f6616add68

Comments

vulkan_encode: do not carry source picture access flags into the barrier
The source picture is written on other queues, whose access flags are
meaningless on the encode queue: ALL_COMMANDS expands to only the
stages the queue supports, none of which support e.g. TRANSFER_WRITE,
making the barrier invalid (VUID 03915). The semaphore wait already
made those writes available; use no source access, as decoding does.

Changed files

libavcodec/vulkan_encode.c

Change #278270

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:02:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	db481362ed37dd44e85faf464c7f9da884644258

Comments

vulkan_video: map the plain H264 baseline profile to a Vulkan profile
The encoder maps the SPS profile_idc, which has the constraint set
flags stripped, giving session parameters an invalid profile enum for
constrained baseline.

Changed files

libavcodec/vulkan_video.c

Change #278271

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:02:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	173eb83c152186dabbc136764b7de1d3003b4c77

Comments

vulkan_encode_h264: signal High profile syntax only where the profile permits it
The capability flags only mean the implementation supports the syntax:
baseline and main profiles prohibit the PPS extension and scaling
matrices, and setting them on drivers advertising support made CBS
reject the headers, as transform_8x8_mode_flag must match its inferred
value.

Fixes #21022.

Changed files

libavcodec/vulkan_encode_h264.c

Change #278272

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:02:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c4b4a9c0ea46adaef2f046e1bbe888e4a5f188d4

Comments

vulkan_encode_h264: disable B-frames for baseline profiles
Baseline only permits I and P slices, yet streams declaring
constraint_set1_flag contained B-slices.

Changed files

libavcodec/vulkan_encode_h264.c

Change #278273

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:02:17
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e2c335ab5c81bad21a641979feaaafae8c894d16

Comments

vulkan_encode_h264: fix the high444p profile value
It was set to High 10, which no Vulkan profile enum maps to, breaking
both the option and >8bpc profile probing.

Changed files

libavcodec/vulkan_encode_h264.c

Change #278295

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:34:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8fc0de2e5b88dbad1d8ec1be6474729fc0542b26

Comments

vulkan_h264: move the session parameter structures off the stack
Nearly 160KiB, which overflows musl's default 128KiB thread stack on
its own.

Fixes #22342.

Changed files

libavcodec/vulkan_h264.c

Change #278296

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Sat 15 Aug 2026 11:34:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b16b5f2a01f3c4f8c9a7769d7a35e8b193946d3e

Comments

vulkan_hevc: move the inline session parameter structures off the stack
123KiB, dominated by the 1024-entry HRD parameters array in
HEVCHeaderVPS; together with anything below it, this overflows musl's
default 128KiB thread stack. Reuse the persistent header buffer, which
the non-inline path never uses at the same time. This also lifts the
artificial vps_num_hrd_parameters limit the fixed-size array imposed.

Fixes #22342.

Changed files

libavcodec/vulkan_hevc.c

Change #278301

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Sat 15 Aug 2026 15:00:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	db632174910e88a08f16ef668c236de9c077a297

Comments

fate/mov: skip cover-art re-encode test without mjpeg decoder and scale filter

Changed files

tests/fate/mov.mak

Change #278305

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Sat 15 Aug 2026 15:10:36
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6fa1295f4f9b3704847728f4f389ed968cf216c5

Comments

tests/fate/filter-video: add aresample dependency to ebur128-height test
Fixes: fc02470c62a71a8fa431f3e424bc8049fec8d90a

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

tests/fate/filter-video.mak

Change #278314

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 18:51:18
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	df86c5ef3a5524b92c4083473bf816f7df8a1df4

Comments

swresample/tests/swresample: Avoid av_assert2, remove ASSERT_LEVEL override
Said #define ASSERT_LEVEL 2 was ineffective because it followed
the inclusion of avassert.h. It only leads to warnings when
ASSERT_LEVEL is defined to something other than 2 (it is not
defined by default).

Also avoid using av_assert2() in a test tool: These are supposed to
be picky, so using always-enabled av_assert0 is appropriate.
This was probably the intention of the ASSERT_LEVEL override
all along.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libswresample/tests/swresample.c

Change #278329

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 22:40:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a2b7272b4a76089ceb69b5792e2766283e473d36

Comments

avcodec/mips/hevc_mc_biw_msa: Fix offsets parameter
Broken in baad19d0b5840f637e95ba4b87d73e58637ec702.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/mips/hevc_mc_biw_msa.c

Change #278330

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 22:40:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	124e8393e5b80ab294170d02e8e559c20a4a8832

Comments

avcodec/mips/mpegvideo_{mmi,msa}: Use correct scantable
This only worked because raster_end derived from
the WMV1/2 intra scantable (i.e. derived from
ff_wmv1_scantable[1]) is <= raster_end derived
from the WMV1/2 inter scantable (i.e. derived
from ff_wmv1_scantable[0]). But it broke the
checkasm test.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/mips/mpegvideo_mmi.c
libavcodec/mips/mpegvideo_msa.c

Change #278331

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 22:40:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d5b8439bdd968f3bea5f973ecb4d51b4325ee364

Comments

avcodec/mips/hevcpred_msa: Fix compilation
Broken in ba56a300a94bdf5520ac1324a8e7fbaeea430904,
ba56a300a94bdf5520ac1324a8e7fbaeea430904 and again
in 38b8ae4112684c3415ab2f7d717605759d764d27.

This means that no has compiled git master on mips
since 2024-06-04. And no one complained.

Hint: One still needs to pass -flax-vector-conversions
here in order to compile this file.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/mips/hevcpred_msa.c

Change #278332

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sat 15 Aug 2026 22:40:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0056dd32fd94e739e14bb3c463c68ebe806dfd1d

Comments

avcodec/mips/vc1dsp_msa: Fix mspel functions
The VC-1 mc prediction for combined horizontal and vertical offset
involves first a vertical 4-tap filter, followed by shifting,
followed by a horizontal 4-tap filter, followed by shifting by 7.
Both passes involve negative coefficients and therefore the
intermediate values for the >> 7 can exceed the range of
0..255*128 a bit in both directions. Therefore one can not simply
use either a logical or an arithmetic right shift. But one can use
them after having added a certain bias to ensure that one is in
the proper range. This is what this commit does.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/mips/vc1dsp_msa.c

Change #278372

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 16 Aug 2026 18:35:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3a8e4fc8062c1065594e7914875da4cea113b295

Comments

avcodec/h264qpel: Deduplicate >8 bit fpel functions
Saves 18760B of .text here.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/h264qpel.c
libavcodec/h264qpel_template.c
libavcodec/snow.c

Change #278378

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 16 Aug 2026 19:20:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	74705b3970cbc277e9366d82ee32d7c3a0a9e3b3

Comments

tests/checkasm/Makefile: Don't build empty files
Empty translation units are actually invalid C that compilers
happen to accept (but warn about when using -pedantic).
In any case, they are unnecessary, so don't build them.

Reviewed-by: Kacper Michajłow <kasper93@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/Makefile

Change #278396

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Mon 17 Aug 2026 01:10:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a48e31e501b950babcf7a47498ecf9e4e16cd7de

Comments

avcodec/x86/pngdsp.asm: replace MMX with SSE2
No more use of mm registers and emms. Performance identical.

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/pngdsp.asm

Change #278402

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	888430127108501f6408df14b0d4c4e04fb9afe4

Comments

vulkan: enable maintenance9 when supported
Enables VK_KHR_maintenance9 when available. The feature of interest
is that queue family ownership transfers become optional for resources
created with exclusive sharing mode, with their contents preserved
when accessed by another queue family.

For optimal-tiling images, this is only guaranteed between queue
families flagged in VkQueueFamilyOwnershipTransferPropertiesKHR, so
query it for each queue family alongside the other properties.

Changed files

libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h
libavutil/vulkan_functions.h
libavutil/vulkan_loader.h

Change #278403

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d0d7d813178fec964716aa2c48f5896ce5c26284

Comments

vulkan: de-phase queue rotation and pin video pools to a single queue
All pools rotated through a family's queues in lockstep from index 0,
colliding on the same queue at the same time. Give each pool a phase
from a per-family, per-library counter: video pools are pinned to the
queue it selects, as sessions order their execution anyway, while all
other pools rotate starting from it. On e.g. Intel, a transcode now
runs decode and encode on separate engines of the shared video family.

The counters are seeded once with a random value, so that separate
processes, and separate copies of this code within one process,
de-phase statistically on drivers with a fixed queue-to-engine mapping.

Also assigns ffv1enc's per-context index once, at init.

Changed files

libavcodec/ffv1enc_vulkan.c
libavutil/vulkan.c

Change #278404

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1c83c513f3cd3705f9001edcad57dd1b8fceb6a

Comments

vulkan: carry the execution context's busy state in a timeline semaphore
An execution abandoned on error between start and submission left the
ownership mutex locked forever, deadlocking the context's next reuse.

Replace the submission fence with a timeline semaphore:
ff_vk_exec_start() claims the context by incrementing the awaited
value, the submission signals it, and the mutex only briefly guards
the dependency lists. Abandonment becomes ff_vk_exec_discard(), called
exactly once on a started, unsubmitted context: it frees the
dependencies, drops the recording, and un-claims by signalling from
the host - which a fence cannot do. ff_vk_exec_submit() cleans up
after its own failures; error paths which leaked the claim now
discard, and ff_vk_exec_start() failures are checked everywhere.

Changed files

libavcodec/apv_encode_vulkan.c
libavcodec/ffv1enc_vulkan.c
libavcodec/proresenc_kostya_vulkan.c
libavcodec/vulkan_apv.c
libavcodec/vulkan_decode.c
libavcodec/vulkan_dpx.c
libavcodec/vulkan_encode.c
libavcodec/vulkan_ffv1.c
libavcodec/vulkan_prores.c
libavcodec/vulkan_prores_raw.c
libavfilter/vf_blackdetect_vulkan.c
libavfilter/vf_nlmeans_vulkan.c
libavfilter/vf_scdet_vulkan.c
libavfilter/vulkan_filter.c
libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h
libavutil/vulkan_functions.h

Change #278405

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	62ad3dd732ac03681dcaaaecce8d4c076d144ec7

Comments

vulkan: add an internal image allocation helper
Creates single standalone images for internal use: exclusive sharing
with no queue family ownership transfers, and dedicated memory when
the driver prefers it.

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #278406

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ccb51e6bf062d7138650209bcce47e789af7a1ed

Comments

hw_base_encode: allow allocating reconstruction frames via a callback
APIs which do not use a frames context for reconstruction images can
provide them through it instead of recon_frames_ref.

Changed files

libavcodec/hw_base_encode.c
libavcodec/hw_base_encode.h

Change #278407

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b3ee8c8920b13160faf66a1bccaa2691477e7c6a

Comments

vulkan_video: allocate DPB images internally, outside of any frame machinery
DISTINCT-mode references and encoder reconstruction images are never
user-visible and only ever accessed by a single queue, yet were
allocated through a full AVHWFramesContext: CONCURRENT-shared images
(disabling compression on several drivers), plus an unused frames
context with its own execution pools, per codec instance.

Pool them internally as plain {image, memory, view, layout} entries:
one video-stage memory barrier per submission orders all prior work,
so they need no semaphores or dependencies, views are created once,
and freeing a DISTINCT-mode picture no longer waits on the host.
COINCIDE-mode outputs remain ordinary frames.

Changed files

libavcodec/vulkan_decode.c
libavcodec/vulkan_decode.h
libavcodec/vulkan_encode.c
libavcodec/vulkan_video.c
libavcodec/vulkan_video.h

Change #278408

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b7e8611b2a6a035ca79afd2818962518126265da

Comments

vulkan: enable unified image layouts when supported
Enables VK_KHR_unified_image_layouts when available, and queries its
feature bits into FFVulkanContext. The extension guarantees that using
the GENERAL image layout comes with no performance penalty, with the
unifiedImageLayoutsVideo feature extending this to video usage.

Changed files

libavutil/hwcontext_vulkan.c
libavutil/vulkan.c
libavutil/vulkan.h
libavutil/vulkan_functions.h
libavutil/vulkan_loader.h

Change #278409

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d06e3815342882ee88492a734a84d1bd532db221

Comments

vulkan_decode: fix operator precedence in the COINCIDE-only sanity check
== binds tighter than &, so the mask expression evaluated to
flags & ((A|B) == A), which is always zero, and the check for drivers
signalling COINCIDE without separate reference images never fired.

Changed files

libavcodec/vulkan_decode.c

Change #278410

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 17 Aug 2026 07:57:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b02fa6b7cdbbcf1ca14ad15b25c6fffb55bb1b8d

Comments

vulkan_decode: prefer DISTINCT-mode references when a driver offers both
When both DPB modes are supported, the choice is up to the application.
COINCIDE was picked, but it forces DPB usage onto every user-visible
output image, costing compression on several drivers, and keeps
references alive as full frames. With references pooled internally as
queue-exclusive, single-usage images, DISTINCT is the better default.

Changed files

libavcodec/vulkan_decode.c

Change #278419

Category	ffmpeg
Changed by	Tymur Boiko <tboikoohnoyoudont@nvidia.com>
Changed at	Mon 17 Aug 2026 08:48:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	426841da9d9167781007e4238583a7c0da5f04c4

Comments

avcodec/vulkan_encode: speed up hevc_vulkan with encode optimizations
Without ALLOW_ENCODE_PARAMETER_OPTIMIZATIONS and with min CU forced to
8x8, hevc_vulkan keeps a much heavier encode config at common quality
levels than h264_vulkan (16x16 macroblocks, no min-CU SPS field). Set
the session flag and use min CU 16x16 when CTB >= 32; also fix
max_transform_hierarchy_depth_inter assignment.

When applying GetEncoded SPS feedback, sync overridden fields together
(extent/conf_win, min/diff CU and TB sizes, transform hierarchy). Copying
only log2_diff left CtbLog2SizeY=7 and undecodable HEVC after some
implementations rewrote min CU.

Reporter cmd (1080p testsrc2, quality 1, 1000 frames):
  before: hevc ~200–317 fps vs higher h264
  after:  hevc 289/325/358 fps, h264 288/317/359 fps (parity)

Changed files

libavcodec/vulkan_encode.c
libavcodec/vulkan_encode_h265.c

Change #278481

Category	ffmpeg
Changed by	Zhao Zhili <zhilizhaoohnoyoudont@tencent.com>
Changed at	Mon 17 Aug 2026 15:52:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8a159afa9855cdd77897c10507fe96c18f3c6485

Comments

avutil/tests/hwcontext_cuda: stop leaking frames contexts
av_hwframe_get_buffer() requires an empty frame and attaches the
frames context reference itself. The test pre-set
frame->hw_frames_ctx, so the plain assignment in
av_hwframe_get_buffer() overwrote the field without unref-ing the
pre-set reference.

Changed files

libavutil/tests/hwcontext_cuda.c

Change #278482

Category	ffmpeg
Changed by	Dmitrii Gershenkop <bortengineerohnoyoudont@protonmail.com>
Changed at	Mon 17 Aug 2026 16:09:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	96f93ece11f47eb5b7677618caeb858e275ae3c4

Comments

avutil/hwcontext_amf: Fix double init for derived devices.
Right after device_derive, FFmpeg also calls device_init. And in device_init (amf_device_init), the fact that AMF might be already initialized with, for example, Vulkan, is ignored and gets initialized again with DX11.

Changed files

libavutil/hwcontext_amf.c
libavutil/hwcontext_amf.h

Change #278489

Category	ffmpeg
Changed by	Hawthorn <rv.help23ohnoyoudont@gmail.com>
Changed at	Mon 17 Aug 2026 16:41:27
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	745680c2d2fce7a87b4fe861d2b1fc2534349ae7

Comments

avformat/mov: bound iloc/iinf item counts by the box size
mov_read_iloc() and mov_read_iinf() read a 16- or 32-bit item/entry
count from the file and pass it straight to av_realloc_array() with no
limit, so a small box can drive a multi-gigabyte allocation before the
read loop hits EOF. Reject counts larger than the remaining box size,
matching the existing sgpd bound.

Signed-off-by: Hawthorn <rv.help23@gmail.com>

Changed files

libavformat/mov.c

Change #278490

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 17 Aug 2026 17:23:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cc50be6d0dd0b1fd518b5f12d44a8f163ddafdf1

Comments

avutil/hwcontext_vulkan: probe host image transfer support before enabling
Drivers may expose the host image transfer format feature, yet reject
the final image parameter combination, or provide no memory type from
which such an image can be allocated. Notably video decode/encode
images are often not possible to allocate from host visible memory
type. Validate the final image parameters, including DRM format
modifiers, and keep the usage flag only when it is actually usable.

Fixes: #21704
Co-authored-by: llyyr <llyyr.public@gmail.com>
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_vulkan.c
libavutil/vulkan_functions.h

Change #278491

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Mon 17 Aug 2026 17:23:43
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4ec050211a285e710a2bd3b539d3434ebed3f71a

Comments

avutil/hwcontext_vulkan: drop NVIDIA/MoltenVK host transfer blocklist
Now that we correctly test if images are valid for host transfer, we
can drop the hardcoded blocklist. The MoltenVK issue has been fixed
upstream [1]. The NVIDIA issue has no upstream tracking reference, and
current drivers show no problems with host image copy in libplacebo,
which uses it in the same way.

[1] https://github.com/KhronosGroup/MoltenVK/issues/2618

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/hwcontext_vulkan.c

Change #278501

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 17 Aug 2026 18:43:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	43284b61a39a7d35633ebbc637db50ff4e97c57e

Comments

Revert "fftools/ffmpeg_sched: allow throttling decoder outputs"
This reverts commit 03dfac563018e6e8b81e331ebae0732d8edfe754.
Said commit led to numerous regressions, namely:
a) It can lead to wrong output interleavement (issue #23731
and (presumably) #24162).
b) It can lead to streams stopping prematurely (issue #24008
and (presumably) #23988 and #24135).
c) It can lead to extreme memory usage even when the input files
are fine, especially mkvmerge files (see issue #24168).

Reviewed-by: Niklas Haas <ffmpeg@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

fftools/ffmpeg_sched.c

Change #278502

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Mon 17 Aug 2026 19:49:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8b003c660fa5f29789acb85827efd31524ee4fd3

Comments

avcodec/libdav1d: fix leak of frames on failure
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libdav1d.c

Change #278519

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Mon 17 Aug 2026 21:06:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1d860a00de25191666b16cdff0517b2bccb6f6f

Comments

avformat/tests/movenc: Convert to API test tool
This is possible because it only uses the public API;
and beneficial when using shared builds (as not all
the internals of lavf are linked statically in the binary).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/Makefile
libavformat/tests/.gitignore
libavformat/tests/movenc.c
tests/api/Makefile
tests/api/api-movenc-test.c
tests/fate/api.mak
tests/fate/libavformat.mak
tests/ref/fate/api-movenc
tests/ref/fate/movenc

Change #278521

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Tue 18 Aug 2026 00:21:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e4be207d8293f30b1d97bcd00f7cc68f97658231

Comments

Add missing .gitignore.

Changed files

libavformat/tests/.gitignore
libswresample/tests/.gitignore
libswscale/.gitignore

Change #278528

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Tue 18 Aug 2026 01:41:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3dddcc3282366b340fdf5de7a6cfd6b1ea153272

Comments

avformat/hashenc: extend output of streamhash muxer
Add the same features as the framehash muxer.

This makes it much more useful as it prints important information that some
tests may need where framehash is not an option given it prints too many lines
(e.g., samples with tons of frames where timestamps or packet flags are
irrelevant, and just a single hash is enough).

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/hashenc.c
tests/fate/dca.mak
tests/ref/fate/dca-xll
tests/ref/fate/dca-xll-coded
tests/ref/fate/dca-xll_51_24_48_768-dmix_2_matrix
tests/ref/fate/filter-channelsplit

Change #278532

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 18 Aug 2026 06:22:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aac66834c230c5a6f0f448dcdc7af879ea4b8da5

Comments

ffv1enc_vulkan: use unaligned bitstream writes in the Golomb shader
The issue is that golomb code writes start after the slice header,
which means they're not guaranteed to be aligned.

Changed files

libavcodec/vulkan/ffv1_enc.comp.glsl

Change #278533

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 18 Aug 2026 06:22:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	92f292807e9194b32175c416fc2ec8727797c1d1

Comments

vulkan: properly align the GET_BITS_SMEM cache refills
The issue is that init_get_bits is not guaranteed to be called with
an aligned address, so LOAD64 tried to align the offset for SMEM
loading, but only aligned to a 4-byte alignment, whilst FILL_SMEM
requires 16 byte alignment.
Fix it by just loading with FILL_SMEM from an aligned position,
then throwing away the padding data at the start.

Changed files

libavcodec/vulkan/common.glsl

Change #278534

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 18 Aug 2026 06:58:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	208a950c275d34a1eca1fe5b34e687a52da78787

Comments

vulkan/ffv1: stop decoding the extra RGB bit on 4.8+ streams
4.8 dropped the extra coded bit on the primary and alpha planes of
RCT slices, but the integer path of get_slice_bits() hardcoded the
old widths, garbling every software-encoded v4 RGB stream. The
Vulkan encoder writes 4.3 and was unaffected.

Changed files

libavcodec/vulkan/ffv1_common.glsl

Change #278535

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 18 Aug 2026 06:58:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7ebeb2bec722c1d6686d2742980c075a72c1db8

Comments

vulkan/ffv1: decode the extra Bayer gm bit on pre-4.8 streams
Mirror of the RGB fix: the hardcoded gm width matched 4.8+ only,
garbling valid 4.3..4.7 Bayer streams. FFmpeg's encoders always
stamp Bayer with 4.10, so only external streams were affected.

Changed files

libavcodec/vulkan/ffv1_dec.comp.glsl

Change #278539

Category	ffmpeg
Changed by	jiangjie <jiangjieohnoyoudont@agora.io>
Changed at	Tue 18 Aug 2026 09:01:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	89153eb701d372f54a5d7d29de5067abc09e11d3

Comments

avformat/http: free cookie parameters on parse failure
Fixes issue #24172

Changed files

libavformat/http.c

Change #278602

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 18 Aug 2026 11:51:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	515a1a4eb082f7fd5ff8e054158470e5b504fc3d

Comments

tests/checkasm/vp8dsp: Don't use declare_func_emms unnecessarily
Possible since b9161b93b5c88bd2f6f110bb527336af9a4c0cae.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/checkasm/vp8dsp.c

Change #278603

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 18 Aug 2026 11:51:33
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4ca56d0f90d0a80ff3e1b1dcff78013e63c26287

Comments

avcodec/x86/vp8dsp: Port ff_vp8_luma_dc_wht_sse to xmm regs
Some parts of it benefit from wider registers:

  vp8_luma_dc_wht_c:        22.7
  vp8_luma_dc_wht_sse:      14.7 ( 1.55x)
  vp8_luma_dc_wht_sse2:     13.2 ( 1.73x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/vp8dsp.asm
libavcodec/x86/vp8dsp_init.c
tests/checkasm/vp8dsp.c

Change #278616

Category	ffmpeg
Changed by	jiangjie <jiangjieohnoyoudont@agora.io>
Changed at	Tue 18 Aug 2026 13:06:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ade01193a1a7f77de0f466067e9961871c8e0e8a

Comments

avformat/dashenc: free http options on avio_open_dyn_buf failure
Fixes #24171

Changed files

libavformat/dashenc.c

Change #278633

Category	ffmpeg
Changed by	jiangjie <jiangjieohnoyoudont@agora.io>
Changed at	Tue 18 Aug 2026 15:55:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	21bbd98e7b05542e9be5661a8532d2513af8c22c

Comments

avfilter/blend: clip expression output to pixel depth
Clip integer expression results in the double domain before storing them. This keeps high-bit-depth samples within their declared range and gives NaN and infinities defined behavior while preserving float formats.

Fixes issue #24121.

Changed files

libavfilter/blend.h
libavfilter/vf_blend.c
tests/fate/filter-video.mak
tests/ref/fate/filter-blend-expr-clipping

Change #278638

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 18 Aug 2026 19:14:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	712a8f9e84038a3e10015de78fb513f650bb7aec

Comments

lavu/vulkan: properly set the required subgroup size for loaded shaders
This was forgotten/left out somehow.

Changed files

libavutil/vulkan.c

Change #278650

Category	ffmpeg
Changed by	unfunnyatearug <yalisib2222ohnoyoudont@gmail.com>
Changed at	Tue 18 Aug 2026 20:56:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	29f513af7833499ddff3f2443609356eacfdd665

Comments

avformat/fsb: reject truncated thp coefficient extradata
the fsb demuxer used `avio_read(..., 32)` for each channel but ignored the return value.
`ff_alloc_extradata()` only clears the padding area. if the input file is truncated, the remaining extradata can therefore contain uninitialized memory.
`nb_channels` is a 16-bit value and was only limited by the `int_max / 32` calculation. this allows a truncated file to cause a leak of up to ~2 MB of uninitialized memory.
the affected extradata is used by the `adpcm_thp` decoder and can also be copied to the output during remuxing.
the fix replaces the unchecked reads with `ffio_read_size()`, which reports an error when the requested amount of data cannot be read. both the thp v3 and v4 paths are fixed.

Signed-off-by: unfunnyatearug <yalisib2222@gmail.com>

Changed files

libavformat/fsb.c

Change #278651

Category	ffmpeg
Changed by	unfunnyatearug <yalisib2222ohnoyoudont@gmail.com>
Changed at	Tue 18 Aug 2026 20:56:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4b47405e1c251405bfa5196e79aa395c59ad2323

Comments

avformat/genh: check thp extradata read and allocation
the genh demuxer had the same short-read issue at one location. the read is now performed with `ffio_read_size()` so truncated input is rejected instead of leaving uninitialized data in the extradata.
`ff_alloc_extradata()` also had its return value unchecked. if allocation failed, the code could dereference a null pointer. the return value is now checked.
the affected path is capped at two channels, so the potential information leak is relatively small, but it is the same underlying bug class.

Signed-off-by: unfunnyatearug <yalisib2222@gmail.com>

Changed files

libavformat/genh.c

Change #278660

Category	ffmpeg
Changed by	Vignesh Venkat <vigneshvohnoyoudont@google.com>
Changed at	Wed 19 Aug 2026 00:26:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d998016f921cd7a909cd64296c39a981722b1be2

Comments

avformat/mov: Support rndr track reference type
This is similar to `cdsc` but intended for rendering metadata.
From a demuxer's point of view both of them are essentially the
same.

Added in the draft spec here: https://www.mpeg.org/standards/MPEG-4/12/

Link to the newly added fate sample: https://drive.google.com/file/d/1qfRALYT6ZCcijKEAkrWWjIDq_E2dm3Ix/view?usp=drive_link

Signed-off-by: Vignesh Venkat <vigneshv@google.com>

Changed files

libavformat/mov.c
tests/fate/mov.mak
tests/ref/fate/mov-t35-cdsc-track
tests/ref/fate/mov-t35-rndr-track

Change #278661

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Wed 19 Aug 2026 03:54:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	564f92cce23ae95399476617b8a1dc357f002a47

Comments

avcodec/ac3dec_fixed: preserve overlap across dependent substreams
ac-3 cores and dependent e-ac-3 substreams may use different
coefficient scales while owning separate delay slots.

the global reset added in 5ba2a2c841 cleared both sets of slots on
every substream switch, corrupting the mdct overlap of valid 7.1
streams.

track the scale per substream and clear only its own delay slots when
it changes. add a fate test using the existing the_great_wall sample.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavcodec/ac3dec.c
libavcodec/ac3dec.h
tests/fate/ac3.mak

Change #278680

Category	ffmpeg
Changed by	Dmitrii Gershenkop <bortengineerohnoyoudont@protonmail.com>
Changed at	Wed 19 Aug 2026 10:52:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cb2370e546bb4c51b2a0f914f06d4698ae3ab1aa

Comments

avfilter/vf_{vpp,sr,vqe}_amf: Enforce memory type

Changed files

libavfilter/vf_sr_amf.c
libavfilter/vf_vpp_amf.c
libavfilter/vf_vqe_amf.c
libavutil/hwcontext_amf.c

Change #278704

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 19 Aug 2026 14:25:40
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3bdd895832244780c250713e49135615ac4de003

Comments

avcodec/lcevcdec: use a counter as timestamp instead of frame pts
If the source is an elementary stream, PTS may be unset and thus
liblcevc-dec will not be able to match nor order input raw pictures and
enhancement payloads.

Given lcevc_process() is called from the main thread in presentation order,
a simple counter will suffice.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/lcevcdec.c
libavcodec/lcevcdec.h

Change #278721

Category	ffmpeg
Changed by	Ayoub Nabil <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Wed 19 Aug 2026 17:29:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2312ae4d10e10e315d12797b78155dbc0bc56cea

Comments

avcodec/ac3dec: avoid coefficient truncation in float decoder
The float decoder dequantized spectral coefficients into an int32_t
buffer using mantissa >> exponent. This discarded the fractional part
before conversion to float and substantially raised the dither noise
floor for low-level coefficients.

Store decoded coefficients as INTFLOAT and apply 2^-exponent directly
in the floating-point path. Keep the fixed-point decoder unchanged, and
perform coupling, rematrixing and final scaling in the native type.

This also removes the now-unused fmtconvert dependency. Add a synthetic
FATE test that measures the corrected dither level. Existing PCM
references remain within one sample except fate-ac3-2.0, whose measured
maximum difference is two samples.

Fixes #23684.

Changed files

configure
libavcodec/ac3dec.c
libavcodec/ac3dec.h
tests/fate/ac3.mak

Change #278722

Category	ffmpeg
Changed by	KO Myung-Hun <komh78ohnoyoudont@gmail.com>
Changed at	Wed 19 Aug 2026 18:20:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1f26f9d9e1d145e53bdd62dccfa3900c06efb032

Comments

configure: enable os2threads with atomics support

Changed files

configure

Change #278723

Category	ffmpeg
Changed by	KO Myung-Hun <komh78ohnoyoudont@gmail.com>
Changed at	Wed 19 Aug 2026 18:20:15
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e1e325235ee2f9f81b39d47ac2f9fe529257589e

Comments

compat/os2threads.h: add pthread_mutex_trylock

Changed files

compat/os2threads.h

Change #278737

Category	ffmpeg
Changed by	Hao Chen <haochencohnoyoudont@google.com>
Changed at	Wed 19 Aug 2026 22:06:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d9ed77cb52ccd5b50f5416c4fa43e875433a1deb

Comments

avfilter/aarch64/vf_bwdif_init: Fix stride bound check
Fix stride condition when bounding w.

Signed-off-by: Hao Chen <haochenc@google.com>

Changed files

libavfilter/aarch64/vf_bwdif_init_aarch64.c

Change #278743

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Wed 19 Aug 2026 22:25:07
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b6e90cc92c6182b07d0ebd6d853b5e39f922fa3c

Comments

avcodec/fmtconvert: Remove now unused subsystem
Unused since 2312ae4d10e10e315d12797b78155dbc0bc56cea.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

configure
doc/mips.txt
libavcodec/Makefile
libavcodec/aarch64/Makefile
libavcodec/aarch64/fmtconvert_init.c
libavcodec/aarch64/fmtconvert_neon.S
libavcodec/arm/Makefile
libavcodec/arm/fmtconvert_init_arm.c
libavcodec/arm/fmtconvert_neon.S
libavcodec/arm/fmtconvert_vfp.S
libavcodec/fmtconvert.c
libavcodec/fmtconvert.h
libavcodec/mips/Makefile
libavcodec/mips/fmtconvert_mips.c
libavcodec/ppc/Makefile
libavcodec/ppc/fmtconvert_altivec.c
libavcodec/riscv/Makefile
libavcodec/riscv/fmtconvert_init.c
libavcodec/riscv/fmtconvert_rvv.S
libavcodec/x86/Makefile
libavcodec/x86/fmtconvert.asm
libavcodec/x86/fmtconvert_init.c
tests/checkasm/Makefile
tests/checkasm/checkasm.c
tests/checkasm/checkasm.h
tests/checkasm/fmtconvert.c
tests/fate/checkasm.mak

Change #278747

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 20 Aug 2026 00:12:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f91beaf9296b902b8d2fff712ba018192ad2e5b3

Comments

fate/tests/ac3: add missing eac3 fixed tests
This ensures the fixed decoder has the same coverage as the float one.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/ac3.mak

Change #278749

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Thu 20 Aug 2026 00:35:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	96d82d90c3f5b74afda99fc51acaf7c89f39bdf2

Comments

avcodec/cuviddec: honor application-selected software output
The CUARRAY output option made the decoder override a software format
selected by get_format() with CUDA. This broke ordinary CUVID decoding
to software consumers.

Default output_format to auto and only override get_format() when the
user explicitly requests cuda/cuarray or enables zero_copy. Keep the
internal frames context in the corresponding hardware format and reject
opaque/non-opaque transitions during decoder reinitialization.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavcodec/cuviddec.c

Change #278755

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Thu 20 Aug 2026 08:28:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a168907d36e7fbbb782b06d527c20ddefc1fccd

Comments

avcodec/x86/vp9mc.asm: Replace MMX with SSE2
Use SSE2 for (put|avg)[48]; identical performance.

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/vp9dsp_init.c
libavcodec/x86/vp9dsp_init_16bpp.c
libavcodec/x86/vp9mc.asm
tests/checkasm/vp9dsp.c

Change #278756

Category	ffmpeg
Changed by	Kostazol <kostazolohnoyoudont@users.noreply.github.com>
Changed at	Thu 20 Aug 2026 09:19:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7d77562d2a18adca969431e5c870870eb8b5edb9

Comments

avformat/hlsenc: provide next packet timing before fMP4 fragment flush
Pass the next HLS boundary packet timing to the child MOV muxer before flushing an fMP4 fragment. The MOV helper only shrinks an overlapping accumulated fragment duration; normal boundaries remain unchanged and the boundary packet itself is not modified.

Changed files

libavformat/hlsenc.c
libavformat/movenc.c
libavformat/movenc.h

Change #278777

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Thu 20 Aug 2026 14:09:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	50a9e34e1bc4bcf54331af39d4a53d8cffde3c22

Comments

avfilter/vf_scale_cuda: don't try to execute disabled filter passes

Changed files

libavfilter/vf_scale_cuda.c

Change #278778

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Thu 20 Aug 2026 14:09:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3d7ea1dac4007b4897ad00991268bbdb51816743

Comments

avfilter/vf_scale_cuda: skip allocating intermediate texture if not needed

Changed files

libavfilter/vf_scale_cuda.c

Change #278779

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Thu 20 Aug 2026 14:09:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dda8b8253cb020d0a09942e6507505ab608128b3

Comments

avfilter/vf_scale_cuda: extract pass determination into its own function

Changed files

libavfilter/vf_scale_cuda.c

Change #278780

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Thu 20 Aug 2026 14:09:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4566eee58816d9a927b460f73c773d18e6060274

Comments

avfilter/vf_scale_cuda: always have valid passes setup

Changed files

libavfilter/vf_scale_cuda.c

Change #278781

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Thu 20 Aug 2026 14:09:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a621ef4291148739e6a301ac5e7aee839c5d0933

Comments

avfilter/vf_scale_cuda: consider chroma subsampling for scaling decisions

Changed files

libavfilter/vf_scale_cuda.c

Change #278788

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Thu 20 Aug 2026 14:52:04
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f101fce22d64db10f500242e23e43a251fe14414

Comments

configure: fix msvc _WIN32_WINNT check
Fixes a regression introduced in a72b9465cda477b34e35d6d1eda670603a0a9f5b.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

configure

Change #278833

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 21 Aug 2026 05:35:19
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	87182e2abc88969ed57998a7c11518199feaa008

Comments

avcodec/audiotoolbox: use 64-bit channel masks
the channel mask expressions used an int literal, making shifts for
channel ids 31 and above undefined before conversion to uint64_t.

use a 64-bit literal in the encoder mapping and decoder layout mask.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavcodec/audiotoolboxdec.c
libavcodec/audiotoolboxenc.c

Change #278834

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Fri 21 Aug 2026 05:38:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	20b70144a6834dbf2984a26d90a4476068fc1f5c

Comments

tests/fate/gapless: complete the mp3 test dependencies
The gapless() helper decodes through the wav muxer into the md5 protocol,
neither of which FRAMECRC requires, and ffprobe_demux reads the sample
through the file protocol, against a reference that covers the sample
format the mp3 decoder picked.

Changed files

tests/fate/gapless.mak

Change #278835

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Fri 21 Aug 2026 05:38:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	bfcf9fcb3735a62648a968caad55d6b37c9998a2

Comments

avformat: factor out iTunSMPB parsing
Apple encoders store gapless playback information in this tag, in the
metadata of mp4 files and, as an ID3v2 comment, of mp3 files, so move the
parsing to a helper both demuxers can share.

The sample counts are widened to the 64 bits the tag can write, scanned
unsigned as it writes them, and rejected past what can be a sample count,
so that callers can add and rescale the three without overflowing.

Changed files

libavformat/demux.h
libavformat/demux_utils.c
libavformat/mov.c

Change #278836

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Fri 21 Aug 2026 05:38:24
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1d17fdd28ba9e53c2257e782fd509fbf4848254

Comments

avformat/mp3dec: parse iTunes gapless info
Apple encoders do not write the LAME extension of the Xing tag, so the
files they produce get no gapless handling at all and no exact duration
either, the sample used here carrying no VBR tag whatsoever.

Contrary to the LAME tag, the iTunSMPB priming count already includes the
decoder delay, so apply it as is, the way the mov demuxer already does,
and validate it against the three counts adding up to what the frames
decode to.

Changed files

libavformat/mp3dec.c
tests/fate/gapless.mak
tests/ref/fate/gapless-mp3-itunes
tests/ref/fate/gapless-mp3-itunes-side-data
tests/ref/fate/id3v2-reenc-delete-metadata
tests/ref/fate/id3v2-reenc-delete-metadata-keep
tests/ref/fate/id3v2-reenc-delete-metadata-keep-format
tests/ref/fate/id3v2-reenc-delete-metadata-keep-stream
tests/ref/fate/id3v2-reenc-delete-metadata-map-metadata
tests/ref/fate/id3v2-reenc-remux-keep

Change #278850

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Fri 21 Aug 2026 10:25:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	806f739fe553678496ac903917f4154803309ea8

Comments

avcodec/d3d12va_encode_av1: add AV1 B-frame (compound prediction) support
This commit adds AV1 B-frame support.
    - Enable B-frame scheduling for the D3D12VA AV1 encoder by setting the
      ref_l0/ref_l1 capability.
    - Stash hidden (non-independent) anchor frames coded data and prepend
      it to the next visible frame packet (show_existing_frame handling).
    - Add FF_HW_FLAG_TIMESTAMP_NO_DELAY to avoid non-monotonic DTS warnings
      at higher B-frame reorder depths.
    - Set 2 B-frames as default(align with D3D12VA H.264 and HEVC encoder)

Example usage:
    - default (2 B-frames will be applied):
      ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4 -c:v av1_d3d12va output.mp4
    - 3 B-frames:
      ffmpeg -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4 -c:v av1_d3d12va -bf 3 output.mp4

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavcodec/d3d12va_encode.c
libavcodec/d3d12va_encode.h
libavcodec/d3d12va_encode_av1.c
libavcodec/hw_base_encode.h

Change #278851

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Fri 21 Aug 2026 10:25:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5f69124aaaed9f7e01dad8d849577826e7a5b20c

Comments

avcodec/vaapi_encode: remove duplicate FLAG_TIMESTAMP_NO_DELAY
Use the shared FF_HW_FLAG_TIMESTAMP_NO_DELAY flag from
hw_base_encode.h instead of VAAPI's own private definition with the
same value and semantics.

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavcodec/vaapi_encode.c
libavcodec/vaapi_encode.h
libavcodec/vaapi_encode_av1.c

Change #278857

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Fri 21 Aug 2026 11:45:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ce1fedc039a27e9e330659abb6e3eb2cddbbabb7

Comments

configure: filter -experimental:* from armasm flags
This fixes build with `--extra-cflags=-experimental:c11atomics`.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

configure

Change #278862

Category	ffmpeg
Changed by	Shiyou Yin <yinshiyou-hfohnoyoudont@loongson.cn>
Changed at	Fri 21 Aug 2026 12:39:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	16a3484985efd7510592ee9872d1a874e27fe36c

Comments

swscale/loongarch: Enable range_convert optimization for LA
The base implementation has been updated to runtime coeff and offset.
Benchmark results:
  name                             ticks (vs ref)
  chrRangeFromJpeg8_1920_c:        237.9
  chrRangeFromJpeg8_1920_lsx:       39.3 ( 5.91x)
  chrRangeFromJpeg8_1920_lasx:      19.6 (11.95x)
  chrRangeFromJpeg16_1920_c:       240.3
  chrRangeToJpeg8_1920_c:          444.3
  chrRangeToJpeg8_1920_lsx:         63.7 ( 6.36x)
  chrRangeToJpeg8_1920_lasx:        28.9 (15.31x)
  chrRangeToJpeg16_1920_c:         502.9
  lumRangeFromJpeg8_1920_c:        140.3
  lumRangeFromJpeg8_1920_lsx:       29.0 ( 4.83x)
  lumRangeFromJpeg8_1920_lasx:      15.8 ( 8.39x)
  lumRangeFromJpeg16_1920_c:        71.7
  lumRangeToJpeg8_1920_c:          303.4
  lumRangeToJpeg8_1920_lsx:         29.2 (10.34x)
  lumRangeToJpeg8_1920_lasx:        14.8 (20.47x)
  lumRangeToJpeg16_1920_c:         294.1

Changed files

libswscale/loongarch/swscale.S
libswscale/loongarch/swscale_init_loongarch.c
libswscale/loongarch/swscale_loongarch.h

Change #278863

Category	ffmpeg
Changed by	Shiyou Yin <yinshiyou-hfohnoyoudont@loongson.cn>
Changed at	Fri 21 Aug 2026 12:39:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	40000d0f5bd6f2ddd162db1131f8a832ea04d64f

Comments

swscale/loongarch: enable planar GBRP/GBRAP luma input optimization
Benchmark results:
  name                          ticks (vs ref)
  planar_gbrp_to_y_512_c:        78.8
  planar_gbrp_to_y_512_lsx:      19.9 ( 3.74x)
  planar_gbrp_to_y_512_lasx:     10.4 ( 7.46x)
  planar_gbrap_to_y_512_c:       79.4
  planar_gbrap_to_y_512_lsx:     20.5 ( 3.61x)
  planar_gbrap_to_y_512_lasx:    15.5 ( 5.12x)

Changed files

libswscale/loongarch/input_lasx.c
libswscale/loongarch/input_lsx.c

Change #278868

Category	ffmpeg
Changed by	Hao Chen <haochencohnoyoudont@google.com>
Changed at	Fri 21 Aug 2026 14:27:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	88ae625e698d7a1c0b7dcea8d915915acda3f054

Comments

libswresample/aarch64: Fix sign extension in audio conversion
The channel count passed as a 32-bit register (w3) is later used
as a 64-bit register (x3) for pointer offsets without proper sign
extension, resulting in misbehavior. This commit adds an explicit
sxtw instruction to ensure the channel count is properly sign-extended.

Additionally, standardizes the use of the 32-bit w2 and w12 registers
in the core length logic across functions to prevent sign extension
issues with lengths without relying on explicit 64-bit casting, matching
expected input bounds tracking.

Signed-off-by: Hao Chen <haochenc@google.com>

Changed files

libswresample/aarch64/audio_convert_neon.S

Change #278872

Category	ffmpeg
Changed by	Tymur Boiko <tboikoohnoyoudont@nvidia.com>
Changed at	Fri 21 Aug 2026 16:06:57
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1a6a186b808a5bf65a408c700846d208227d6549

Comments

avcodec/vulkan_decode: set H.264 pictureLayout from SPS frame_mbs_only_flag
VkVideoDecodeH264ProfileInfoKHR::pictureLayout must match the sequence.
ITU-T H.264 frame_mbs_only_flag==1 means frames only; ==0 allows field
pictures and MBAFF. Vulkan PROGRESSIVE sessions are progressive-only;
non-PROGRESSIVE enables interlaced field DPB (h264_decode / videocoding).

pictureLayout values are mutually exclusive video profiles (vk.xml).
Map with !frame_mbs_only_flag, then validate with
vkGetPhysicalDeviceVideoCapabilitiesKHR:

- frame_mbs_only_flag == 1, or no SPS -> PROGRESSIVE (one query)
- frame_mbs_only_flag == 0 -> prefer INTERLACED_INTERLEAVED_LINES only

Do not use avctx->field_order: MBAFF may still report progressive.

Signed-off-by: Tymur Boiko <tboiko@nvidia.com>

Changed files

libavcodec/vulkan_decode.c

Change #278880

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Fri 21 Aug 2026 20:11:48
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd4fc473be3ca20704615d8ea110effe7fe11e13

Comments

avfilter/af_afir: clamp initial ir selection
the ir option may exceed nbirs because its static range is independent
of the configured number of impulse responses. init() only clamps
prev_selir, leaving selir able to index ctx->inputs out of range.

with the default nbirs=1, afir=ir=1 reads ctx->inputs[2] and crashes.
clamp selir like process_command() and add a fate test.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_afir.c
tests/fate/filter-audio.mak

Change #278888

Category	ffmpeg
Changed by	Leo Izen <leo.izenohnoyoudont@gmail.com>
Changed at	Fri 21 Aug 2026 20:13:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eb0bfa852e7b9c524960300607ba2c4617060a9b

Comments

avutil/opt: reformat set_string_binary
Fix the code formatting in libavutil/opt.c's static function
set_string_binary after bd68a6130495fe17bf33e0a9ddc053baae18c81f.

No functional changes.

Signed-off-by: Leo Izen <leo.izen@gmail.com>

Changed files

libavutil/opt.c

Change #278924

Category	ffmpeg
Changed by	WyattBlue <wyattblueohnoyoudont@auto-editor.com>
Changed at	Sat 22 Aug 2026 19:41:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	11b68cce09900f8a4a25ceba42055094af00bf37

Comments

avfilter/af_whisper: Add 'lock', 'eval' lang modes
Add 'lock' and 'eval' languages modes. With autodetection, whisper
reruns language detection on every queued chunk, an extra encoder pass
that can also drift to a different language mid-file.

Add two explicit values for the language option: 'eval' keeps the
per-chunk reevaluation, 'lock' caches the language detected by the
first chunk that produces transcribed text and reuses it for the
following chunks. Requiring actual text means a silent, musical or
[BLANK_AUDIO]-only opening chunk leaves detection to a later one.

'auto' remains an alias for 'eval' so existing behavior is unchanged.

Changed files

doc/filters.texi
libavfilter/af_whisper.c

Change #278925

Category	ffmpeg
Changed by	J. Brandt Buckley <brandtohnoyoudont@runlevel1.com>
Changed at	Sat 22 Aug 2026 19:48:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ec8411e8548de7270f6e799cdc94f45b7a7a92d3

Comments

avcodec/vc1_block: don't drop the last macroblock when 1 bit remains
The P- and B-picture per-MB guards added in 2a8785966c bail out of the loop
to error concealment when get_bits_left() <= 1, but P- and B-picture MBs can
legitimately cost exactly 1 bit in certain cases. This patch relaxes the guards
to accommodate those cases.

## Explanation

With the SKIPMB bitplane in Raw mode a skipped macroblock is a single SKIPMBBIT
(SMPTE ST 421:2013, 7.1.3.7, p.109) and then "may only contain the HYBRIDPRED
syntax element" (8.3.4.4, p.140), which is only read when predictors A and C
are in-bounds and the predicted vector differs from A or C by more than 32
quarter-pels (8.3.5.3.5, Figure 55, pp.149-150). That condition isn't met in a
static region so such a MB costs just 1 bit. A direct+skipped B-picture MB
likewise reads nothing beyond its raw flag (Table 30, p.86).

In Simple/Main profile a picture is byte-aligned with no stop bit, so if a
picture ends with a 1-bit skipped macroblock and its last coded bit lands on a
byte boundary, the final MB is reached with exactly one bit left. The guard
then conceals a valid macroblock, sets FF_DECODE_ERROR_CONCEALMENT_ACTIVE
on an undamaged frame, and the damage propagates to every frame up to the next
keyframe. (Advanced Profile BDUs end in a trailing '1' bit per Annex E.2/E.3
p.437, so they always have at least two bit left and are therefore unaffected.)

## Reproducing

Reproducible with an existing sample (prior to this patch):

    $ ffmpeg -i https://samples.ffmpeg.org/V-codecs/WMV9/wmv-samples/sample-wmv9-simple-SP@ML.wmv -map 0:v -f null - 2>&1 | grep -A1 concealing
    [wmv3 @ 0x122605430] concealing 75 DC, 75 AC, 75 MV errors in P frame
    [vist#0:0/wmv3 @ 0x122604080] [dec:wmv3 @ 0x122604df0] corrupt decoded frame
    [wmv3 @ 0x122605430] concealing 50 DC, 50 AC, 50 MV errors in P frame
    [vist#0:0/wmv3 @ 0x122604080] [dec:wmv3 @ 0x122604df0] corrupt decoded frame
    [wmv3 @ 0x122605430] concealing 100 DC, 100 AC, 100 MV errors in P frame
    [vist#0:0/wmv3 @ 0x122604080] [dec:wmv3 @ 0x122604df0] corrupt decoded frame

## References

- SMPTE ST 421:2013 spec: https://pub.smpte.org/doc/st421/20131008-pub/

Signed-off-by: J. Brandt Buckley <brandt@runlevel1.com>

Changed files

libavcodec/vc1_block.c

Change #278926

Category	ffmpeg
Changed by	Michael Partridge <michael.partridgeohnoyoudont@petrosys.com.au>
Changed at	Sat 22 Aug 2026 20:10:11
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	45bc2518bed20b1bab4e71afb13feba0e4aeb205

Comments

configure: support z.lib and zlib.lib for zlib on MSVC-style linkers
Probe the zlib pkg-config link flags to detect whether the Windows import library is named z.lib or zlib.lib, and use that when rewriting -lz for MSVC-style linkers.

Also make the non-pkg-config zlib fallback try both names explicitly so configure works with either z.lib or zlib.lib.

Changed files

configure

Change #278944

Category	ffmpeg
Changed by	Brian Leake <phydohnoyoudont@phyd.org>
Changed at	Sun 23 Aug 2026 00:58:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f10fdd631026056517edbad79ee0a136ef71c5df

Comments

avformat/movenc: fix dec3 chan_loc derivation for E-AC-3 dependent substreams
chan_loc was derived from the bitstream's chanmap with a plain shift, but
the two fields use different bit orders, so every E-AC-3 stream carrying
a dependent substream (7.1 and above) was muxed with the wrong channel
layout signalled in the container.

chanmap is a 16-bit field read MSB-first, so flag index i sits at bit
(15 - i); this matches how ff_eac3_custom_channel_map_locations is
indexed in ac3_parser.c. Per ETSI TS 102 366 Annex F, chan_loc bits 0-7
carry flag indices 5-12, i.e. chanmap bit (10 - j). The mapping is not
contiguous at the top: index 13 (Lts/Rts) has no chan_loc bit and must be
skipped, and chan_loc bit 8 carries LFE2, which is index 14 and so
chanmap bit 1. The 0x1f mask was also 5 bits where 9 are needed, so some
flags were dropped rather than merely misplaced.

Effect on the layouts that use a dependent substream:

    layout   chanmap   before   after
    7.1      0x1A00    0x010    0x002
    5.1.2    0x0010    0x000    0x040
    5.1.4    0x0210    0x010    0x042
    LFE2     0x0002    0x000    0x100

For 7.1 the field claimed Lsd/Rsd (surround direct), a position no 7.1
speaker layout has, instead of Lrs/Rrs. This is invisible to FFmpeg,
whose decoder reads the layout from the bitstream, but demuxers that
trust the container act on it: Apple's AudioToolbox reported the affected
files as "7.1 (L C R Ls Rs LFE Lsd Rsd)" and collapsed them to a stereo
downmix, losing the centre channel, all four surrounds and the LFE.

Verified with samples.ffmpeg.org/A-codecs/AC3/eac3/7_pt_1.eac3, whose
dependent substream declares chanmap 0x1A00. After this change the muxed
dec3 carries 0x002, matching GPAC MP4Box for the same input, and
AudioToolbox reports "7.1 (L C R Ls Rs LFE Rls Rrs)" with all eight
channels intact.

Signed-off-by: Brian Leake <phyd@phyd.org>

Changed files

libavformat/movenc.c

Change #278945

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 01:25:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b79d4c4c0a160fc46988e98505af6039a53ad53e

Comments

tests/fate/lavf-container: add an EAC3 in MP4 test
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/lavf-container.mak
tests/ref/lavf-fate/eac3.mp4

Change #278967

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 13:46:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	89f58356b31c63cea8604ac19a2365c6ae02c0b9

Comments

avcodec/tests/encinfo: Convert into API test
It doesn't use any internals of lavc. Converting it to an API
test tool reduces disk usage for shared builds.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/Makefile
libavcodec/tests/.gitignore
libavcodec/tests/encinfo.c
tests/api/Makefile
tests/api/api-enc-info-test.c
tests/fate/wav.mak

Change #278968

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 13:46:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	732a8644b87f1d5a7f9409fcc184702d59239183

Comments

tests/fate/wav: Deduplicate encinfo rules
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/fate/wav.mak
tests/ref/fate/wav-enc-adpcm-ima-wav-encinfo
tests/ref/fate/wav-enc-adpcm-ms-encinfo
tests/ref/fate/wav-enc-adpcm_ima_wav-encinfo
tests/ref/fate/wav-enc-adpcm_ms-encinfo

Change #278969

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 13:46:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dea13cca4f6ff3c281ba848be5ce0e288245394a

Comments

tests/api/Makefile: Don't undef main
Should be unnecessary since fde3bb16f90ac456709c7305b9e230e8327d8625.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/api/Makefile

Change #278974

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 17:10:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	68834f79cf07ccba2b1e9b88cebc9ef66a342e9c

Comments

avcodec/x86/h264_deblock: Write lines in original order
I.e. not lines 9-16, then lines 1-8. No changes in benchmarks
here; it is mainly done to reduce differences with the 32bit code.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #278975

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 17:10:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a960e8dba9efefd23411afeca82503edbb75291d

Comments

avcodec/x86/h264_deblock: Combine 32,64 code for deblock_h_luma_intra_8
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #278976

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 17:10:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2fd5efa112ee66393cffbe34552ad65f82936732

Comments

avcodec/x86/h264_deblock: Avoid mmx in deblock_h_luma_intra_8
Old benchmarks:
  h_loop_filter_luma_intra_8bpp_c:        28.0
  h_loop_filter_luma_intra_8bpp_sse2:     74.5 ( 0.38x)
  h_loop_filter_luma_intra_8bpp_avx:      74.6 ( 0.38x)

New benchmarks:
  h_loop_filter_luma_intra_8bpp_c:        26.3
  h_loop_filter_luma_intra_8bpp_sse2:     57.4 ( 0.46x)
  h_loop_filter_luma_intra_8bpp_avx:      57.6 ( 0.46x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #278977

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 17:10:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9f911c1ead41cb68f1e798b1e6322974b911f67e

Comments

avcodec/x86/h264_deblock: Remove obsolete macro parameters
Forgotten in b1140d3c982e0720d2dd638f2d7ae0ca6552ac13.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #278978

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Sun 23 Aug 2026 17:10:58
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9419d4c3a5a92563720115beb8831a6f906da96b

Comments

avcodec/x86/h264_deblock: Don't use mmx in deblock_h_luma_intra_8
Old benchmarks:
  h_loop_filter_luma_intra_8bpp_c:        26.3
  h_loop_filter_luma_intra_8bpp_sse2:     57.4 ( 0.46x)
  h_loop_filter_luma_intra_8bpp_avx:      57.6 ( 0.46x)

New benchmarks:
  h_loop_filter_luma_intra_8bpp_c:        26.5
  h_loop_filter_luma_intra_8bpp_sse2:     52.2 ( 0.51x)
  h_loop_filter_luma_intra_8bpp_avx:      52.3 ( 0.51x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm
tests/checkasm/h264dsp.c

Change #278979

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:28:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0e45eebe49a88079022d9cabee2b4eced755b003

Comments

avutil/channel_layout: remove back suffix from some layouts
The idea was for the suffix to refer to the fact they feature Top Back channels
instead of Top Side, but this distinction is unnecessary given there's no
confusion between back and side like with base level channels (Ls/Rs, Lss/Rss,
Lrs/Rsr), nor known specs where Top Side channels are used without Top Back
also being used.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

doc/APIchanges
libavutil/channel_layout.c
libavutil/channel_layout.h
libavutil/version.h

Change #278980

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:28:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	261df1d136c5621d63700d044d40cfb30568966a

Comments

avutil/channel_layout: remove AVMatrixEncoding from channel_mask_c doxy group
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavutil/channel_layout.h

Change #278981

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:28:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dd38c57676a25daadc44109555cf1bfb27caab97

Comments

avformat/iamf: use the new channel layout macros
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/iamf.c

Change #278982

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:35:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ed923a7a897763471c65b507cea2ef661e1e1646

Comments

avcodec/aacenc: add support for writing PCE layouts with height channels
Fixes ticket #24218

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/aacenc.c
libavcodec/aacenc.h

Change #278983

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:35:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f35acb72ac567821d52da6d53a3e06a3c97855f3

Comments

avcodec/aacenc: add support for ambisonic layouts
AAC can't signal such layouts, so this is merely a hack to allow such
streams to be passed to the encoder instead of having to define layouts
using arbitrary speaker positions for the sake of having one with the
required amount of channels.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/aacenc.c

Change #278985

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:35:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	10c3380e478cfa36b7d94110f511cb5929900b3c

Comments

tests/audiogen: support up to 16 channels
Needed by the next commit.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/audiogen.c

Change #278986

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 23 Aug 2026 18:35:29
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1019f8f036602a8464185baa4857654337eeca14

Comments

tests/fate/aac: add tests for PCE layouts with height channels
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/fate/aac.mak
tests/ref/fate/aac-encode-5_1_2
tests/ref/fate/aac-encode-5_1_4
tests/ref/fate/aac-encode-7_1_2
tests/ref/fate/aac-encode-7_1_4
tests/ref/fate/aac-encode-7_2_3
tests/ref/fate/aac-encode-9_1_4
tests/ref/fate/aac-encode-9_1_6

Change #279166

Category	ffmpeg
Changed by	iSold Leo <sohnoyoudont@qwqlog.com>
Changed at	Mon 24 Aug 2026 22:09:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ca99230f7176c589b8390b34f2d1c130b9e1881d

Comments

avformat/mp3enc: buffer ID3v2 for non-seekable outputs
ID3v2 writes the tag size by seeking back to its header. This fails after output has been flushed to a non-seekable AVIO context, leaving a zero-sized tag. Attached pictures expose this reliably; sufficiently large metadata is affected too.

Build the tag in a dynamic buffer for non-seekable MP3 outputs and copy it after finalization. Keep the direct path for seekable outputs.

Fixes: #23917
Signed-off-by: iSold Leo <s@qwqlog.com>

Changed files

libavformat/aiffenc.c
libavformat/id3v2.h
libavformat/id3v2enc.c
libavformat/mp3enc.c
tests/fate/cover-art.mak
tests/fate/id3v2.mak

Change #279181

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Tue 25 Aug 2026 03:02:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	007cd1fd439927b212c44dbd6ca16bf65ea9ddde

Comments

avcodec/libfdk-aacenc: replace use of deprecated channel layout
Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libfdk-aacenc.c

Change #279334

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Tue 25 Aug 2026 11:25:49
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3d5534d27151cd2d7f0acba080a1eabec7c79a40

Comments

checkasm: Fix building for macos/aarch64
Commit 74705b3970cbc277e9366d82ee32d7c3a0a9e3b3 changed to only
compile some of the checkasm source files when they are expected
to be non-empty, but the condition for when perf/macos_kperf.c
is needed is wrong, breaking building checkasm in default macos
aarch64 configurations.

Checkasm decides whether to enable the macos kperf codepaths on
its own, irrespective of ffmpeg's configure's --enable-macos-kperf
(which defaults to disabled) - see the end of checkasm's
src/perf_internal.h for the condition:

    #if defined(__APPLE__) && ARCH_AARCH64
      /* This only depends on <dlfcn.h>, and isn't needed on other architectures. */
      #define HAVE_MACOS_KPERF 1
    #else
      #define HAVE_MACOS_KPERF 0
    #endif

To approximate this, build macos_kperf.c for any aarch64 target.
For non-macOS cases, perf_internal.h defines inline functions in
most aarch64 build configurations, so we don't end up with an
empty translation unit in any case.

Changed files

tests/checkasm/Makefile

Change #279351

Category	ffmpeg
Changed by	zhorzhetta1404-ux <zhorzhetta1404ohnoyoudont@gmail.com>
Changed at	Tue 25 Aug 2026 16:21:47
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d7d119bbe26c7e5c27a8b50fc8a25da43564c48a

Comments

avformat/oggdec: check for integer overflow in buf_realloc
In buf_realloc, os->bufsize is unsigned int and the code doubles it:
    nb = av_realloc(os->buf, 2*os->bufsize + AV_INPUT_BUFFER_PADDING_SIZE);

When os->bufsize exceeds UINT_MAX/2, the multiplication overflows
to a small value, av_realloc returns a tiny buffer, and the subsequent
page write in ogg_read_page goes out of bounds.

Reject the reallocation before the overflow can occur. The largest
safe value is (UINT_MAX - AV_INPUT_BUFFER_PADDING_SIZE) / 2.

Signed-off-by: age5000 <age5000@proton.me>

Changed files

libavformat/oggdec.c

Change #279354

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Tue 25 Aug 2026 18:06:34
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	09219cc6832671211c1963da0e27181d7bea62e5

Comments

swscale: print the effective scaler with SWS_PRINT_INFO
use the resolved luma scaler for the log. keep the SWS_BICUBLIN description only when chroma is actually bilinear.

add coverage for scaler_sub overrides and the missing FATE dependency. this fixes the incorrect scaler output reported in #23975.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libswscale/utils.c
tests/fate/filter-video.mak

Change #279355

Category	ffmpeg
Changed by	Zsolt Vadasz <zsolt_vadaszohnoyoudont@protonmail.com>
Changed at	Tue 25 Aug 2026 18:10:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a1050d48b0df3f7c3fa1c631ec8e82528b3f7c85

Comments

avcodec/j2kenc: Compute division only once, preshift bit
Pre-shift n to the highest bit in put_bits(), and compute a division
outside of a hot loop in encode_tile().

Changed files

libavcodec/j2kenc.c

Change #279394

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 26 Aug 2026 02:41:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	46f327624835a97a9c838b103b4e75b47a2d34f0

Comments

avcodec/h264_slice: clear the ER picture when starting a second field
h264_frame_start() clears h->er.cur_pic, and decode_nal_units() only sets
it again for pictures that are not field coded, right before running error
concealment. A second field goes through neither, so h->er.cur_pic keeps
pointing at whatever frame coded picture the context decoded earlier.

With frame threading the two fields of a pair are decoded by different
contexts, so whether this triggers depends on what the context that picks
up the second field decoded before, which is why the number of bogus
warnings varies between runs.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/h264_slice.c

Change #279395

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 26 Aug 2026 02:41:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	27b7fa0c107cf6f8f8d3fac541a96ee14940b202

Comments

avcodec/h264_direct: do not read the other field's rows of colocated mb_type
To determine whether the colocated picture is field coded, the temporal
and spatial direct setup peek at
sl->ref_list[1][0].parent->mb_type[mb_xy] before mb_xy has been remapped
to the colocated field's parity. That row belongs to the current field's
parity, while await_reference_mb_row() above waited on the parity of the
referenced field, so whenever the two differ the load is unsynchronised
and with frame threading can observe macroblocks of a field that is
still being decoded.

Every macroblock of a field picture has MB_TYPE_INTERLACED set, so
checking field_picture first gives the same result in all correctly
synchronized cases without touching the mb_type array.

This also fixes few TSAN reported warnings.

Fixes: https://trac.ffmpeg.org/ticket/10891
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavcodec/h264_direct.c

Change #279438

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	255276f780c3deeda1730acf42e4254a68e81833

Comments

avfilter/vf_scale_cuda: select separable passes per plane
Plan luma and chroma passes from their sampling geometry. Allocate
intermediate storage only for planes needing horizontal scaling and
inherit untouched input textures.

Use virtual chroma extents to handle odd sizes and subsampling changes.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/vf_scale_cuda.c

Change #279439

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	41045c5c6ab8d37dbb6ff971ad3377ccd5eeeb83

Comments

avfilter/vf_scale_cuda: encapsulate generic filter resources
Group filter weights, the pass plan, and intermediate textures in a
replaceable filter set. Route generic kernel selection through the
active set and load both output directions so it owns all
geometry-dependent state.

Keep the requested use_filters value separate from the resolved mode.
Support a one-tap generic copy when no plane needs scaling.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/vf_scale_cuda.c

Change #279440

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a2b7a8210e3fbffbbb959752f832484b57864325

Comments

avfilter/vf_scale_cuda: regenerate filters for cropped frames
Generic filter coefficients depend on visible input dimensions, which
can change through per-frame crop metadata. Rebuild the filter set when
that geometry changes and synchronize before freeing active CUDA
resources.

Match av_frame_apply_cropping() for subsampled planes by flooring crop
origins while rounding visible extents up. This prevents odd left or top
crops from shifting chroma by one sample.

Clear applied cropping and size-dependent side data from the output.
Use visible dimensions when updating the sample aspect ratio.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/vf_scale_cuda.c

Change #279441

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c4887d9de309d14e6ead84a730a6b7e7e3996434

Comments

avfilter/vf_scale_cuda: quantize generic output to source depth
Clamp generic-filter output to the logical source depth before restoring
its storage alignment. This prevents filter overshoot from setting
padding bits in formats such as P010 and P012.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/vf_scale_cuda.c
libavfilter/vf_scale_cuda.cu
libavfilter/vf_scale_cuda.h

Change #279442

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eb4c99d49c455bad5e3747135b3df95fcacaa09f

Comments

avfilter/vf_scale_cuda: use floating-point Lanczos intermediates
Store horizontal Lanczos results in floating-point textures so two-axis
filtering quantizes only after the final pass.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

libavfilter/vf_scale_cuda.c
libavfilter/vf_scale_cuda.cu

Change #279443

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	50665df43b81dbdc6bc73eb645f8289bebc511c4

Comments

avfilter/vf_scale_cuda: make the Lanczos radius configurable
Pass param to libswscale when generating generic Lanczos filters,
allowing radii from 1 to 10. The fixed-function path remains unchanged.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

doc/filters.texi
libavfilter/vf_scale_cuda.c

Change #279444

Category	ffmpeg
Changed by	Diego de Souza <ddesouzaohnoyoudont@nvidia.com>
Changed at	Wed 26 Aug 2026 11:52:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a8c7afa7d778cefec08f781ef5bcc09c640161fb

Comments

avfilter/vf_scale_cuda: use generic filters for Lanczos by default
Use the generic weight-LUT path for Lanczos in automatic mode to provide
the full filter window and downscaling anti-aliasing. Keep the legacy
four-tap path available with use_filters=0.

Signed-off-by: Diego de Souza <ddesouza@nvidia.com>

Changed files

doc/filters.texi
libavfilter/vf_scale_cuda.c

Change #279513

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 26 Aug 2026 17:28:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	66a8d8cff4daabf32fc98ceec0c1fcf330fa88ac

Comments

avcodec/libfdk-aacdec: add support for top back and top side channels
While PCE can signal these, libfdk-aac supports only up to 8 channels, so any sample
that could contain these will most likely have more channels than can be decoded anyway.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libfdk-aacdec.c

Change #279514

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Wed 26 Aug 2026 17:28:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3654a14795fb67ef5bdc0702183fedf74fb356ba

Comments

avcodec/libfdk-aacdec: add support for bottom front channels
Same as the previous commit, actual samples with less than eight channels including
these is very unlikely.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavcodec/libfdk-aacdec.c

Change #279515

Category	ffmpeg
Changed by	Reda DRISSI <reda.drissiohnoyoudont@yahoo.fr>
Changed at	Wed 26 Aug 2026 17:34:51
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	56c3d963ec42cb7e02cd3d96ccc1ed1e5cb0e0c7

Comments

avformat/movenc: add Apple APAC codec tag for mp4 muxing
The MP4 muxer rejects streams with AV_CODEC_ID_APPLE_APAC because the
codec_mp4_tags[] table has no entry for it, while the MOV muxer already
accepts the codec. Add the apac sample entry tag so APAC streams can be
copied into MP4 with -c:a copy.

Signed-off-by: Reda DRISSI <reda.drissi@yahoo.fr>

Changed files

libavformat/movenc.c

Change #279529

Category	ffmpeg
Changed by	Kacper Michajłow <kasper93ohnoyoudont@gmail.com>
Changed at	Wed 26 Aug 2026 22:13:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fccb887903fa799146ecf7500c3339f9c9b27e6e

Comments

avutil/mem: check attribute support for allocation attributes
Clang reports itself as gcc 4.2, so av_alloc_size was never applied for
it, and neither attribute was applied for clang with MSVC target, which
does not define __GNUC__ at all. Check for the attributes directly.

Signed-off-by: Kacper Michajłow <kasper93@gmail.com>

Changed files

libavutil/mem.h

Change #279546

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:11:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c40a8a235bb8d4d599916dae98c01b18f73e53a1

Comments

avformat/shared: Don't log incomplete lines
Incomplete lines (without \n) may be torn apart by av_log()s
from other threads. So log them once. Also use ff_data_to_hex()
for the data->hex conversion.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/shared.c

Change #279547

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:11:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e9ec302b3137aec22ecfd45167f371ac99c3a814

Comments

avformat/shared: Use ff_data_to_hex() where appropriate
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/shared.c

Change #279548

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:11:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5279cf09758d32a7944a96f5796cda42cd9a840b

Comments

avformat/ape: Don't log incomplete lines
Incomplete lines (without \n) may be torn apart by av_log()s
from other threads. So log them once. Also use ff_data_to_hex()
for the data->hex conversion.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/ape.c

Change #279549

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:11:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0a6f7590274a89895c074779f606acd62b614058

Comments

avformat/utils: Make ff_data_to_hex() usable with sizes > INT_MAX/2
It uses an int size field, so it should support the whole range.
Notice that no current caller could ever have been affected by this.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/utils.c

Change #279550

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:11:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5fb7f9c5deee9fb4b580b20efe150be300bed6f8

Comments

avformat/smoothstreamingenc: Use ff_data_to_hex() where advantageous
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavformat/smoothstreamingenc.c

Change #279551

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f7500c6d715c4ce7700b0e17c7d386ceb41b11d4

Comments

avcodec/x86/h264_deblock: Avoid zeroing register unnecessarily
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #279552

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	22b3de0eccf42174ea00ebc30c659bf0c555611b

Comments

avcodec/x86/h264_deblock: Avoid unnecessary stores/loads
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #279553

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f6691293a50cbfe4edfaf3eeeb73b79fc6c16ccf

Comments

avcodec/x86/h264_deblock: Write lines in original order
I.e. not lines 9-16, then lines 1-8. No changes in benchmarks
here; it is mainly done to reduce differences with the 32bit code.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #279554

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a51f4456c84c923983873beb81dd88ded7c71226

Comments

avcodec/x86/h264_deblock: Remove redundant rederiving of stride3
It was stored in a nonvolatile register.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #279555

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f8ca7c8f15375cfee460a28484cf518bbce75ed9

Comments

avcodec/x86/h264_deblock: Combine 32,64 code for deblock_h_luma_8
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm

Change #279556

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Thu 27 Aug 2026 00:51:10
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a6f573a1dbfc6861a96f3a453805398f86f0e7a9

Comments

avcodec/x86/h264_deblock: Avoid mmx register in deblock_h_luma_8
Old benchmarks:
  h_loop_filter_luma_8bpp_c:        41.1
  h_loop_filter_luma_8bpp_sse2:     60.6 ( 0.68x)
  h_loop_filter_luma_8bpp_avx:      60.4 ( 0.68x)

New benchmarks:
  h_loop_filter_luma_8bpp_c:        42.0
  h_loop_filter_luma_8bpp_sse2:     48.1 ( 0.87x)
  h_loop_filter_luma_8bpp_avx:      48.3 ( 0.87x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavcodec/x86/h264_deblock.asm
tests/checkasm/h264dsp.c

Change #279595

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 12:54:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2a61dbb99f45fdecd69ddc9ace6f3996cc023cbb

Comments

tests/fate/libavutil: add FATE test for pixdesc
Test av_get_pix_fmt / av_get_pix_fmt_name round-trip across all
pixel formats, av_color_{range,primaries,transfer,space}_name,
av_chroma_location_name and av_alpha_mode_name round-trips
through their from_name() counterparts using a name-equivalence
check (two enum values may legitimately share the same name
string), av_chroma_location_enum_to_pos and pos_to_enum
round-trip across all valid chroma locations, and iterate every
pixel format printing av_get_padded_bits_per_pixel and
av_pix_fmt_count_planes.

Coverage for libavutil/pixdesc.c: 83.63% -> 90.03%

Numbers measured against a full FATE run. Remaining uncovered
lines are av_get_pix_fmt_string, the av_get_pix_fmt_loss /
av_find_best_pix_fmt_of_2 public wrappers and a few loss
branches, the palette branches of av_read_image_line and
av_write_image_line, and the *_ext name lookups in
av_color_{primaries,transfer}_from_name.

Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

.forgejo/CODEOWNERS
libavutil/tests/pixdesc.c
tests/fate/libavutil.mak
tests/ref/fate/pixdesc

Change #279597

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 12:54:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b799c7772b49961cbd3eb1975ffdb636ad69e15c

Comments

tests/fate/libavutil: add FATE test for downmix_info
Test av_downmix_info_update_side_data() including first-call
allocation with zero-initialized defaults, field write/read-back,
second-call reuse (same pointer, preserved values), and the OOM
path via av_max_alloc.

Coverage for libavutil/downmix_info.c: 100.00% -> 100.00%

On a full FATE run the file is already covered by the AC-3
decoder tests (fate-ac3-4.0 and friends), which require
FATE_SAMPLES. This test exercises the API directly and keeps the
file covered on builds that run without samples.

Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/downmix_info.c
tests/fate/libavutil.mak
tests/ref/fate/downmix_info

Change #279599

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 12:54:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	58c47ca583f23a5ab1329f5e026f8f2c33775464

Comments

tests/fate/libavutil: add FATE test for hdr_dynamic_metadata
Test av_dynamic_hdr_plus_alloc, av_dynamic_hdr_plus_create_side_data,
av_dynamic_hdr_plus_to_t35 in all three documented modes (size
query, caller-allocated, existing caller-owned buffer) including
the AVERROR_BUFFER_TOO_SMALL path, and av_dynamic_hdr_plus_from_t35
as a round-trip inverse.

Two HDR10+ fixtures are used. The first is a conforming
ST 2094-40:2020 ApplicationVersion 1 payload: one processing
window, the nine distribution maxrgb percentiles required by
section 8.5.4, and neither actual-peak-luminance item nor
ColorSaturationWeight, both excluded for that version by section
9.4. The second is marked in the source as deliberately
non-conforming and exists only to reach the optional
serializer/parser branches: a second processing window with its
geometry, both actual-peak-luminance matrices, per-window tone
mapping and color saturation. Every field either fixture
populates is decoded and printed, so a serializer or parser that
drops or corrupts a value changes the reference rather than
passing a flag-only check.

Also test that av_dynamic_hdr_plus_to_t35 ignores
application_version and always writes the CTA-861-H value, the
NULL and oversized error paths, and
av_dynamic_hdr_smpte2094_app5_alloc,
av_dynamic_hdr_smpte2094_app5_create_side_data and the App5
to_t35 / from_t35 round trip over four fixtures covering both
early-out paths (no adaptive tone map, reference-white tone
mapping) and both sides of the component mixing and gain curve
selectors: per-image parameters with component mixing type 3 and
explicit control point slopes, and common parameters shared from
alternate 0 with a mixing type other than 3 and the pchip slope.
The App5 to_t35 validation failures and the OOM paths via
av_max_alloc are covered as well.

Coverage for libavutil/hdr_dynamic_metadata.c: 31.02% -> 95.53%

The baseline comes from the HDR10+ sample tests on a full FATE
run (fate-hevc-hdr10-plus-metadata, fate-png-mdcv and the
Matroska/WebM HDR10+ remux tests).

Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/hdr_dynamic_metadata.c
tests/fate/libavutil.mak
tests/ref/fate/hdr_dynamic_metadata

Change #279600

Category	ffmpeg
Changed by	marcos ashton <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 12:54:38
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	17c109e8d94d19976dc2f5db9c3577b3f831df43

Comments

tests/fate/libavutil: add FATE test for side_data
Test av_frame_side_data_desc and av_frame_side_data_name on
valid and invalid types, av_frame_side_data_get and its _c
variant for present and absent types,
av_frame_side_data_remove (including MULTI types with
duplicates), av_frame_side_data_remove_by_props filtering by
AV_SIDE_DATA_PROP_*, av_frame_side_data_clone with and without
AV_FRAME_SIDE_DATA_FLAG_REPLACE (including the AVERROR(EEXIST)
and AVERROR(EINVAL) paths) and with AV_FRAME_SIDE_DATA_FLAG_UNIQUE,
av_frame_side_data_add in move mode, with
AV_FRAME_SIDE_DATA_FLAG_NEW_REF and with
AV_FRAME_SIDE_DATA_FLAG_REPLACE over an existing entry, and
av_frame_side_data_free.

Coverage for libavutil/side_data.c: 72.14% -> 86.43%

Numbers measured against a full FATE run. Remaining uncovered
lines are internal allocation-failure error paths requiring
injected failures mid-operation.

Signed-off-by: marcos ashton <marcosashiglesias@gmail.com>

Changed files

.forgejo/CODEOWNERS
libavutil/Makefile
libavutil/tests/.gitignore
libavutil/tests/side_data.c
tests/fate/libavutil.mak
tests/ref/fate/side_data

Change #279621

Category	ffmpeg
Changed by	Marcos Ashton Iglesias <marcosashiglesiasohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 13:45:46
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	278eeeb16e67f24d46851e3355bef5a51322c3bd

Comments

avcodec/x86/hevc: add SSSE3 pred_planar
Implements planar intra prediction for 8 bit. x86 has no HEVC intra
prediction at all; aarch64 NEON already covers planar, DC and angular.

Every term except (size-1-x)*left[y] is linear in y, so for a group of
eight columns they collapse into one running accumulator whose increment
left[size]-top[x] is loop invariant. Running columns outermost keeps that
accumulator and its delta in registers, so a row costs one pmullw and two
paddw with no reload of top[]. 16x16 and 32x32 run two groups together,
sharing the left[y] broadcast and writing a row with one 16 byte store.
Words suffice throughout: each coefficient pair sums to size, so the
largest value is 2*32*255 + 32 = 16352. SSSE3 rather than SSE2 since
pshufb does the broadcasts and the byte to word widening.

checkasm --bench on a Core Ultra 7 155H:
hevc_pred_planar_4x4_8_c:            18.4
hevc_pred_planar_4x4_8_ssse3:         5.5 ( 3.35x)
hevc_pred_planar_8x8_8_c:            83.1
hevc_pred_planar_8x8_8_ssse3:        16.3 ( 5.10x)
hevc_pred_planar_16x16_8_c:         224.8
hevc_pred_planar_16x16_8_ssse3:      47.3 ( 4.75x)
hevc_pred_planar_32x32_8_c:         629.5
hevc_pred_planar_32x32_8_ssse3:     196.5 ( 3.20x)

Part of #23022.

Changed files

libavcodec/hevc/pred.c
libavcodec/hevc/pred.h
libavcodec/x86/hevc/Makefile
libavcodec/x86/hevc/pred.asm
libavcodec/x86/hevc/pred_init.c

Change #279623

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 15:43:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d3d2e7f1f691b8ac06062e37b826dc673a7a864e

Comments

avformat/http: split the response status line parser out of process_line
Makes the parser reachable from a test program.  No functional change, beyond
checking the allocation of the exported version string.

Changed files

libavformat/http.c
libavformat/http.h

Change #279624

Category	ffmpeg
Changed by	Romain Beauxis <romain.beauxisohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 15:43:01
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	72dc02e52ff544c18d09f4da1cccb96883143ae2

Comments

tests/http: test response status line parsing
Covers well formed status lines, a missing or non numeric status code, lines
that do not start with an HTTP version, and status lines whose version or code
does not respect the grammar of RFC 9112.

The reference output records the current behavior: every one of those malformed
lines is accepted, most of them as a 2xx success.

Changed files

libavformat/Makefile
libavformat/tests/.gitignore
libavformat/tests/http.c
tests/fate/libavformat.mak
tests/ref/fate/http-status-line

Change #279625

Category	ffmpeg
Changed by	Théo Valette <315040668+theovaletteohnoyoudont@users.noreply.github.com>
Changed at	Thu 27 Aug 2026 16:08:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	78efc17cc5b6cfa62bea5bca39a4c4b5dea731d7

Comments

avformat/movenc: preserve timecode flags when copying tmcd tracks
The MOV demuxer retains tmcd timecode flags in codec extradata, but copied
tracks never initialize MOVTrack.timecode_flags. mov_write_tmcd_tag()
therefore writes zero when reconstructing the sample description, causing
unchanged frame counters to be interpreted as non-drop-frame.

Initialize the complete flags field from sample-description extradata for
copied tmcd tracks. fate-copy-trac236 covers three drop-frame tmcd tracks.

Fixes issue #20841.

Reported-by: unsword01 <unsword01@noreply.code.ffmpeg.org>
Signed-off-by: Théo Valette <315040668+theovalette@users.noreply.github.com>

Changed files

libavformat/movenc.c
tests/ref/fate/copy-trac236

Change #279626

Category	ffmpeg
Changed by	Ackanir <git.ackanirohnoyoudont@gmail.com>
Changed at	Thu 27 Aug 2026 16:44:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9f35e220ffbba21c88356eb2bbfa3679ede93793

Comments

avformat/movenc: fix crash when flushing a fragment with no data
mov_flush_fragment() calls ffio_reset_dyn_buf() on mov->mdat_buf when
writing the initial moov. That buffer is allocated lazily on the first
packet, so it is still NULL when the trailer is written before anything
was muxed, and ffio_reset_dyn_buf() dereferences it.

627da1111c9d replaced ffio_free_dyn_buf(), which tolerates NULL, with
ffio_reset_dyn_buf(), which does not, turning this case into a crash.

This affects every fragmented mode that defers the initial moov to the
first flush.

Skip the block entirely when the buffer was never opened. Nothing
references the empty mdat it would otherwise write, as no samples were
muxed.

Reproduced with:
ffmpeg -f lavfi -i color=s=64x64 -frames:v 0 -movflags +frag_keyframe out.mp4

Fixes: 627da1111c9d ("libavformat/movenc: Uses dynamic buffers for fragmented chunks")
Signed-off-by: Ackanir <git.ackanir@gmail.com>

Changed files

libavformat/movenc.c

Change #279650

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Thu 27 Aug 2026 23:07:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3a9920690274f032e6e1fe05624e50378e0bc285

Comments

avfoundation: Consolidate the HAVE_IOKIT ifdefs
This makes it easier to tweak the condition for when these codepaths
are enabled.

Changed files

libavdevice/avfoundation.m

Change #279651

Category	ffmpeg
Changed by	Martin Storsjö <martinohnoyoudont@martin.st>
Changed at	Thu 27 Aug 2026 23:07:13
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0bffa4a84dbf0d05a29e580ea7f7a4b84e122f4d

Comments

avfoundation: Fix compilation for iOS
This fixes building for iOS after commit
c2802e520a08dbf0c939ec8b5653b29dfd886427.

On iOS, the fallback define kIOMasterPortDefault is never available.
The newer define kIOMainPortDefault is available since iOS 15.

However, if building with older iOS SDKs (iOS 15 SDK or older) then
IOKit is entirely missing. This caused compilation to succeed if
building with such older versions of the iOS SDK.

If building with a newer iOS SDK/Xcode but targeting an older version
of iOS, then IOKit is available, but compilation would fail as
kIOMasterPortDefault wasn't available.

Thus, this fixes compilation for iOS with iOS 16 SDKs and newer,
which was previously broken.

Changed files

libavdevice/avfoundation.m

Change #279661

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Fri 28 Aug 2026 05:17:50
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3acec0a1af2dda0a0838689b8b8649e7deb080a0

Comments

avformat/iamf_writer: don't discard skip samples signaled through side data for Opus
If it's set, then it should be used.
Fixes issue #24285.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

libavformat/iamf_writer.c
tests/ref/fate/iamf-5_1-copy
tests/ref/fate/iamf-ambisonic_1-projection-copy

Change #279720

Category	ffmpeg
Changed by	Sylvestre Ledru <sylvestreohnoyoudont@debian.org>
Changed at	Fri 28 Aug 2026 15:01:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d29ff884221042af219e53cc339a62356b061816

Comments

vulkan_av1: fix out-of-bounds write of tile_sizes
The tile count limit was only checked once on entry, but tileCount is
incremented by ff_vk_decode_add_slice() inside the loop, so a stream
carrying more than MAX_TILES tiles writes past the end of the
MAX_TILES-element tile_sizes array.

Check the limit on each iteration, and reject at MAX_TILES rather than
above it, since tileCount is used as the write index.

Changed files

libavcodec/vulkan_av1.c

Change #279721

Category	ffmpeg
Changed by	Sylvestre Ledru <sylvestreohnoyoudont@debian.org>
Changed at	Fri 28 Aug 2026 15:01:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d411d9e752b9100f8d88f6f2b2bd20900148955c

Comments

vulkan_decode: use the pointee size when growing slice_off
sizeof(slice_off) is the size of the uint32_t pointer, not of an entry.
Harmless in practice as it over-allocates on 64-bit, but wrong, and
under-allocates on targets where a pointer is smaller than uint32_t.

Changed files

libavcodec/vulkan_decode.c

Change #279765

Category	ffmpeg
Changed by	Wu Jianhua <toqsxwohnoyoudont@outlook.com>
Changed at	Fri 28 Aug 2026 20:50:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ecd6fcb5ab0e6531f7d0000975bb6542e152f9c8

Comments

avcodec/tiffenc: handle all TIFF types on big-endian
Signed-off-by: Wu Jianhua <toqsxw@outlook.com>

Changed files

libavcodec/tiffenc.c

Change #279768

Category	ffmpeg
Changed by	Wu Jianhua <toqsxwohnoyoudont@outlook.com>
Changed at	Fri 28 Aug 2026 20:50:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	235b81cb9a393e67137cb16d8b55c1bbae0d6514

Comments

avcodec/tiffenc: support writing ICC profiles
Write ICC profiles as TIFF UNDEFINED data and align all out-of-line IFD values to a word boundary. Account for non-empty ICC side data while allocating the packet and calculate the image size in checked 64-bit arithmetic.

Signed-off-by: Wu Jianhua <toqsxw@outlook.com>

Changed files

libavcodec/tiffenc.c
tests/ref/lavf/tiff

Change #279769

Category	ffmpeg
Changed by	Wu Jianhua <toqsxwohnoyoudont@outlook.com>
Changed at	Fri 28 Aug 2026 20:50:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	836ff94338bd078f3c3dfde7dfbe6e5c608ae71f

Comments

avcodec/exif: add InterColorProfile tag to tag list
Signed-off-by: Wu Jianhua <toqsxw@outlook.com>

Changed files

libavcodec/exif.c

Change #279770

Category	ffmpeg
Changed by	Wu Jianhua <toqsxwohnoyoudont@outlook.com>
Changed at	Fri 28 Aug 2026 20:50:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	df48dc624e7103ae99c59cb9d744cd17d317ec4d

Comments

tests/fate/image: add fate-tiff-icc
Signed-off-by: Wu Jianhua <toqsxw@outlook.com>

Changed files

tests/fate/image.mak
tests/ref/fate/tiff-icc

Change #279817

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Sat 29 Aug 2026 12:18:28
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1ae40482188283a7f955d001abb08c5d865cdae0

Comments

avfilter/dnn: extend ONNX Runtime backend with dnn_detect and multi-output support
This patch extends the existing ONNX Runtime DNN backend with object
detection support via dnn_detect and removes the single-output restriction
in dnn_processing.

Example usage:
  # SSD detection (single output, DirectML)
  ffmpeg -i input.mp4 -vf "scale=300:300,format=rgb24,dnn_detect=dnn_backend=onnx:model=ssd.onnx:model_type=ssd:device=dml:device_id=0" -f null -

  # YOLOv3 detection (two outputs, CPU)
  ffmpeg -i input.mp4 -vf "scale=416:416,format=rgb24,dnn_detect=dnn_backend=onnx:model=yolov3.onnx:output=yolo_13&yolo_26:model_type=yolov3:nb_classes=80:anchors=116&90&156&198&373&326&30&61&62&45&59&119" -f null -

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

doc/filters.texi
libavfilter/dnn/dnn_backend_onnx.c
libavfilter/dnn/dnn_io_proc.c
libavfilter/dnn_filter_common.c
libavfilter/vf_dnn_detect.c

Change #279831

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Sat 29 Aug 2026 16:53:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3386acd2f91d4a59be20eddf6d26abad682f4ca0

Comments

avutil/slicethread: fix build without threading support
The backwards compatibility wrappers for the deprecated
avpriv_slicethread_{create,execute}() dereference AVSliceThread, whose
definition only exists in the threaded branch of the file. A build
without pthreads/w32threads/os2threads therefore failed with
"dereferencing pointer to incomplete type 'AVSliceThread'".

Move the wrappers into the threaded branch and give the no-threads
branch stubs for the deprecated API matching those of create2/execute2.

Changed files

libavutil/slicethread.c

Change #279870

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dac4bc75862e3f635c674a58249fa066ebb6902c

Comments

configure: find an installed nvidia-video-filters package
The RTX Video filters that follow are generated code over data this tree does
not contain and cannot produce: the *_cuda_gen.h that describes each captured
kernel graph, and the cubins and weights it names, are extracted from the
proprietary NVIDIA libraries by an out-of-tree tool and installed into a prefix
that describes itself with an nvidia-video-filters.pc.  Nothing here ships or
extracts them.

The package existing is only half the answer.  Each feature comes from a
different NVIDIA library, so an install routinely holds some and not others, and
a filter whose data is absent must not be built rather than built to fail at run
time.  So the umbrella check only picks up the cflags, and each feature is
probed separately for both halves -- its generated header has to compile, and
the data directory named by the .pc has to exist -- giving one nvfdata_* per
feature for the filters to depend on.

Changed files

configure

Change #279871

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5b8190461c4eefbf63a77ee88ad20469872fa8bc

Comments

avfilter: add a shared core for the RTX Video CUDA filters
Seven of the filters that follow replay a captured NVIDIA kernel graph, and all
seven are 1:1 filters built the same way.  A <feature>_cuda_gen.h describes one
feature's graph -- which cubins to load, how big each scratch buffer is at a
given frame size, where the weights go, and every launch's grid, block and
argument block -- and the filter's job is to turn that into CUDA calls.  That
job is the same every time: gate the GPU architecture, load the modules, lay out
one contiguous arena, upload the weights, bind the input and output images, then
replay the launch list once per frame.  This is that job, written once.

What stays in each vf_*.c is what genuinely differs: its AVOptions, which
generated config the options select, the shape of its I/O binding, and the named
tunable offsets it patches into the argument blocks.

The generated headers are per-feature and expose everything as static, so the
core never includes them.  Each filter instead passes its tables in through
layout-compatible views and proves the cast with a static_assert, so a generator
change that broke the assumption fails the build rather than corrupting a
launch.

Two decisions here are load-bearing rather than tidiness, and are commented as
such in the source: the arena is one contiguous allocation because several
kernels do a tile/halo read a little past the logical end of their input buffer,
which is harmless inside an arena and becomes an illegal access once the heap
fragments; and cuSurfObjectCreate/Destroy are resolved out of libcuda directly,
once per process, because they are the one pair ffnvcodec's loader does not
export.

Changed files

libavfilter/rtx_cuda.c
libavfilter/rtx_cuda.h
libavfilter/rtx_dlpp_abi.h

Change #279872

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	404b97e55176a266600d878d661a8ef6ed7c38df

Comments

avfilter: add vsr_cuda, NVIDIA RTX Video Super Resolution
Upscales packed RGB CUDA frames with the RTX VSR network from the NGX SDK
snippet nvngx_vsr.dll, replayed directly on libcuda -- no DLL is loaded at run
time.  vsr_cuda_gen.h describes the captured graph as a function of the input
and output size; the filter evaluates it in config_output() and then hands the
result to the rtx_cuda core, which does the loading, allocation and replay.

What is specific to VSR and therefore lives here: the quality/scale option pair
that picks a generated config, the packed-RGB format selectors patched into the
argument blocks, and the input-texture/output-surface binding.

Quality 0-4 are the upscaling models (0 is a plain bicubic reference, 2x is fixed
for low/medium, and high/ultra upscale natively by 2x/3x/4x), 5 and 6 are
native-resolution restoration networks forced to 1x, and 7 is a pass-through.
Output sizes other than the network's native factor resample that result with the
graph's own bicubic compose.

Not every model exists in every extraction: the Windows nvngx_vsr.dll carries
0-4 at 2x, and the native Linux library adds the restoration and lossless models
and the 3x/4x heads.  The option ranges cover the union, and a model the
installed data has no network for is refused at config time by name.

The cubins and weights are not shipped; the filter is built only when an
installed nvidia-video-filters package provides the VSR data.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_vsr_cuda.c

Change #279873

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5664364466e924a14139df0af662a731856f7acd

Comments

avfilter: add isr_cuda, NVIDIA NGX Image Super Resolution
Runs the NGX Image Super Resolution network (the NvDLISR snippet
nvngx_dlisr.dll), a fidelity-oriented still-image upscaler at a fixed integer
2x, 4x or 8x.  It is a different network from vsr_cuda's and is structured
differently: it is resolution-independent, running on fixed 256x256 tiles with a
16-pixel overlap, so every per-tile launch carries identical grids, blocks and
argument bytes at every input size and only the device pointers differ.

Because the scale set is exactly what the snippet validates and there is no
resampling path in the graph, the output is always scale x the input -- there is
no w/h expression pair as on the VSR filters.

Input and output are plain linear device buffers rather than textures and
surfaces, so frames are staged through packed scratch; the conversion and tiling
kernels are CUDA-runtime functions living in the snippet's fat binaries rather
than as loose cubins.  Everything else -- module loading, the arena, the weights
upload and the output hwframe -- is the shared rtx_cuda core.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_isr_cuda.c

Change #279874

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	47d0aa27ab1029b24e64596773211f1c3e143eb2

Comments

avfilter: add vsr_drv_cuda, the driver RTX VSR network
The video super-resolution the NVIDIA driver itself runs (the DXVA/PPE plugin
nvaivpx.dll, ppe/features/AIVP), as opposed to the NGX SDK snippet that
vsr_cuda drives.  It is a newer and heavier network over the same DLPP kernels,
with its own quality levels and tunables, and it is worth having both: they do
not produce the same picture.

The driver numbers its networks 0-4, but they are not ordered by strength and
index 0 is a byte-identical duplicate of index 4.  Only 1-4 are exposed here:
one network reachable under two numbers is a usability trap, not a feature.  By
fidelity the order runs q1 (most faithful) > q3 > q4 > q2 (most aggressive
detail synthesis), so the default is q1.

Two things differ from vsr_cuda beyond the tables, and both are visible here:
each launch is issued with the kernel's own EIATTR_CBANK_PARAM_SIZE rather than
the captured driver argsize, because the driver over-reports by 8 bytes for the
two DLPP tex/surf kernels and cuLaunchKernel then fails with
CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES; and the graph has only two bindless slots in
total -- the input texture at dlpp_preProcess and the output surface at
dlpp_postProcess or ResampleAndComposeFP16 -- with no internal surfaces.

The graph super-resolves 2x internally: an exact isotropic 2x output stores
directly, anything else composes to the requested rectangle.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_vsr_drv_cuda.c

Change #279875

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ee8579987dbe5120e74199a99a883369ece37b44

Comments

avfilter: add dlpp_drv_cuda, the driver DLPP super-resolution network
The third super-resolution network in the driver stack: ppe/features/DLPP,
driven directly rather than through the AIVP plugin that vsr_drv_cuda uses.  It
exposes DLPP's own quality levels, which include two high-quality models the
AIVP path does not reach.

Quality 1 and 2 are the fixed-2x base models and behave like vsr_drv_cuda.
Quality 3/4 are the high-quality models, which upscale natively at a chosen
integer factor -- the scale option is the driver's own params[0x38] float, which
selects the pixel_shuffle2/3/4 super-resolution head.  For both, an output at
exactly the native factor stores directly and any other size resamples the
native result to the requested rectangle.

As on vsr_drv_cuda the driver's index 0 is a byte-identical duplicate -- of 1
here rather than of 4 -- and is likewise not exposed, so quality runs 1-4 and
defaults to 1.

It shares vsr_drv_cuda's two DLPP quirks: launches are issued with the kernel's
EIATTR_CBANK_PARAM_SIZE rather than the over-reported driver argsize, and the
whole graph has just the two bindless slots.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_dlpp_drv_cuda.c

Change #279876

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7ca18c5f07401a6c967124b3f3533122fa644f10

Comments

avfilter: add truehdr_cuda, NVIDIA RTX TrueHDR inverse tone mapping
Expands SDR to HDR with the RTX TrueHDR network from the NGX SDK snippet
nvngx_truehdr.dll.  Unlike the super-resolution filters it does not rescale:
the output resolution equals the input, while the neural core runs at an
internal resolution derived from the input aspect ratio (ff_rtx_nn_dims).

The two output formats are different colour spaces rather than two bit depths,
selected by argument words in the final drtm launch, so the filter tags the
output frame to match: rgbaf16le (the default) is scRGB -- linear light, Rec.709
primaries, 1.0 = 80 nits -- and x2bgr10le is display-ready HDR10/BT.2100 with a
PQ transfer and Rec.2020 primaries.  The four tunables (contrast, saturation,
middle grey, peak luminance) are orthogonal overrides in that same block.

The input is bound as a PITCH2D texture rather than a CUDA array, which is what
the DLL does and what makes 10-bit x2bgr10le input byte-exact against it; the
format-agnostic texture read normalizes 8-bit rgb0/rgba and 10-bit alike.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_truehdr_cuda.c

Change #279877

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d8eb6779c3ba0713dc856430b71c816f5191fe22

Comments

avfilter: add truehdr_drv_cuda, the driver RTX TrueHDR network
The inverse tone mapping the NVIDIA driver itself runs (the DXVA/PPE plugin
nvaihdrx.dll, ppe/features/TrueHDR), as opposed to the NGX SDK snippet that
truehdr_cuda drives.  Like it, the output resolution equals the input and the
neural core runs at ff_rtx_nn_dims' internal resolution; unlike it, the graph
carries two internal surfaces -- postprocessing to debanding, and debanding to
drtm -- each a float32 array exposed to its producer as a surface and to its
consumer as a texture.

The tone curve comes entirely out of the final drtm launch's argument block.
The captured template runs ToneMapMode 0, a near-linear bypass that reads only
MaxLuminance; it is byte-exact against the driver but blows out midtones, so
tonemap=1 -- the default -- enables the driver's adaptive inverse tone map,
which also consumes the live per-frame calculate_pov scene statistic and is
shaped by the tunable floats.  This network is considerably brighter than the
SDK one, so preset=sdk names a tunable set that pulls its curve onto
truehdr_cuda's for a like-for-like comparison.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_truehdr_drv_cuda.c

Change #279878

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	44f50e6eb726319ceff5010f513b35a20a5c51a0

Comments

avfilter: add deepdvc_drv_cuda, NVIDIA RTX Dynamic Vibrance
Colour and vibrance enhancement with the driver's DeepDVC network (the DXVA/PPE
plugin nvaidvcx.dll, ppe/features/DeepDVC).  It is an HDRnet-style learned
enhancer: a small CNN predicts a coefficient grid which applyLUTToSurface then
applies at full resolution.

It is same-resolution and in-place, which shapes the binding: applyLUTToSurface
reads and writes one image handle and the downsample kernel reads that same
frame, so the whole graph binds a single frame texture.  Unified TEXMODE lets
one texture handle serve both the sample and the surface store -- a surface
handle there fails the TLD.

Both tunables, vibrance and gain, are a final blend strength inside
applyLUTToSurface, so vibrance=0 is an exact pass-through rather than an
approximate one.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_deepdvc_drv_cuda.c

Change #279879

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fe061a2d37ce40ac261ae20947ee446646c4944a

Comments

avfilter: add smoothmotion_cuda, NVIDIA Smooth Motion frame interpolation
Doubles the frame rate with the frame generator the NVIDIA driver ships as
"Smooth Motion" (in libnvidia-present.so).  The network is an FP8 conv/attention
U-Net; per interpolated frame it runs a fixed 25-launch graph -- downscale,
conv/attention backbone, depth_to_space, conv_out for the flow field, then
warp_coarse.  It emits the t=0.5 midpoint, so it is exact for 2x and the filter
offers nothing else.

The kernels are fully resolution-parametric: the driver runs the same graph with
dimensions, grids, scratch sizes and param-block offsets scaled to the native
frame size, without tiling and without resampling to a fixed internal
resolution.  smoothmotion_cuda_gen.h reproduces that scaling, so arbitrary input
sizes work.

This one does not use the rtx_cuda core.  It is a frame-rate filter rather than
a 1:1 one, it drives per-kernel multi-arch fatbins rather than a module list,
and it needs YUV input, so it carries its own CUDA kernels
(vf_smoothmotion_cuda.cu) to pack the supported 8/10/12/16-bit planar and
semi-planar layouts into the packed 4:4:4 the network samples, and to unpack the
result.  Packed inputs skip that step entirely.

The two input frames are bound as bindless textures and the output as a bindless
surface; the handles are written into the parameter block by name through
sm_fill_params, so the filter never patches the block by byte offset.

Changed files

configure
doc/filters.texi
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_smoothmotion_cuda.c
libavfilter/vf_smoothmotion_cuda.cu

Change #279880

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	11143c0eb094a14f8f3631f352f729e5b3103197

Comments

avfilter/smoothmotion_cuda: build on the shared RTX core
This was the last filter standing outside rtx_cuda.c, with its own copy of
almost everything the core exists to provide: a libcuda dlopen per filter
instance to reach cuSurfObjectCreate, its own module table and loader, its own
texture and surface creation, its own device binding and hwframe setup, and a
hand-written teardown -- which is where three of the bugs this series fixed
came from, all of them things the core had already got right for the other
seven.

What it adopts:

- FFRtxCuda replaces eighteen context members (the device ref, context and
  stream, the module table and its name-dedup array, the four scratch pointers,
  the output array and surface, the five pack buffers and their pitch, the four
  texture handles, and the libcuda handle with its two function pointers).
  ff_rtx_free_graph() replaces the 56-line teardown; init() disappears
  entirely, since resolving cuSurfObjectCreate is the core's job and it does it
  once per process rather than once per filter.
- The four scratch buffers become one contiguous ff_rtx_alloc_arena() with the
  trailing guard the core documents as load-bearing: a tile/halo read past a
  buffer's end is harmless inside one arena and an ILLEGAL_ADDRESS once separate
  allocations scatter across a fragmented heap.
- The output array and surface, the pack buffers and their input textures
  become FFRtxImages -- ff_rtx_image_array(FF_RTX_SURF|FF_RTX_LDST) and
  ff_rtx_image_pitch(FF_RTX_TEX|FF_RTX_CLAMP) build byte-identical descriptors
  to the ones this filter was writing out by hand.
- load_kernels() builds FFRtxModule/FFRtxFunc tables over sm_kernel_names[] and
  hands them to ff_rtx_load_modules(), which brings the bounds checks and the
  "no image for this GPU (cc x.y)" diagnostic with it.  The conversion PTX is
  parked one module id past the fatbins so it is unloaded with them.
- Weights go through ff_rtx_upload_weights(), so a weights.bin that does not
  match the graph is the same hard error it is everywhere else rather than a
  warning and a truncated copy.
- config_output() ends in ff_rtx_bind_device / ff_rtx_config_hwframes /
  ff_rtx_setup.  That also settles the output pool size: 4 like its siblings,
  where this filter alone passed 0.

What stays its own, because it genuinely differs: the launch convention (these
kernels take a single parameter struct through cuLaunchKernel's kernelParams,
not the packed argument buffer the core's replay uses) and the launch list,
which sm_fill_params refills every frame; the 25-format pack/unpack model, which
FFRtxPixFmt cannot express; and the straight-through RGB path's per-frame
textures over the source frames' own memory -- for which the core grew
ff_rtx_tex_over_pitch(), so the descriptor is written once rather than twice.

Verified bit-exact: framemd5 over fifteen pipelines -- every input family
(planar and semi-planar YUV, 4:2:2, packed 4:4:4, 8/10/16-bit, packed RGB,
rgba64, x2rgb10/x2bgr10), both `packed` modes and a non-default interpolation
window -- is unchanged, as is the output of the other seven filters over the
core changes.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SSFz5UKyF9dSGYVeKioi4n

Changed files

libavfilter/Makefile
libavfilter/rtx_cuda.c
libavfilter/rtx_cuda.h
libavfilter/vf_smoothmotion_cuda.c

Change #279881

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Fri 07 Aug 2026 17:45:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0db5be95d695cf71ea021fe227703c8ae37c4d0d

Comments

avfilter/rtx_cuda: clean up comments

Changed files

libavfilter/rtx_cuda.c
libavfilter/rtx_cuda.h
libavfilter/vf_deepdvc_drv_cuda.c
libavfilter/vf_dlpp_drv_cuda.c
libavfilter/vf_isr_cuda.c
libavfilter/vf_smoothmotion_cuda.c
libavfilter/vf_truehdr_cuda.c
libavfilter/vf_truehdr_drv_cuda.c

Change #279882

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Thu 20 Aug 2026 03:25:44
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1f7d511a250ae9f5d0974f2bbd706ac8c67a26f8

Comments

avfilter/smoothmotion_cuda: add field_align, defaulting to exact
The generated table can now emit either flow-field geometry.  The driver's own
is inconsistent at any dimension that is not a multiple of 40: the network built
the field at a fixed 10 px per texel, spanning 2.5*L0 pixels, and the warp reads
it back as if it spanned the frame, so the flow lands up to 2.5*L0 - dim pixels
(worst case 39) from where it was computed -- a ramp, worst at the right/bottom
edge.  Motion boundaries are what pay for it: measured at 1600x900, 3.2 dB of
Y-PSNR frame-wide and 6.9 dB on the leading edge of a moving object.

field_align=exact uses the geometry the network actually produced and is the
default, since it emits bytes identical to the driver's at every multiple-of-40
size -- which is every display mode Smooth Motion runs at in the driver, and
where the whole table was validated.  field_align=driver reproduces the driver
at the other sizes too, and is what comparing against a capture needs.

See docs/smoothmotion/field-alignment.md in the rtx-video-re tree for the
measurements and how to repeat them.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AGq45TEwHmDoHnSzCYJuP7

Changed files

libavfilter/vf_smoothmotion_cuda.c

Change #279883

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fd131e79130380a5555679a847ebb38b686f433d

Comments

avfilter/vf_nvoffruc: Add filter for nvidia's Optical Flow FRUC library
The NvOFFRUC library provides a GPU accelerated interpolation feature
based on nvidia's Optical Flow functionality. It's able to provide
reasonably high quality realtime interpolation to increase the frame
rate of a video stream - as opposed to vf_framerate that just does a
linear blend or vf_minterpolate that is anything but realtime.

As interesting as that sounds, there are a lot of limitations that
mean this filter is mostly just a toy.

1. The licensing is useless. The library and header and distributed as
   part of the Optical Flow SDK which has a proprietary EULA, so anyone
   wanting to build the filter must obtain the SDK for both build and
   runtime and the resulting binaries will be nonfree and
   unredistributable.

2. The NvOFFRUC.h header doesn't even compile in pure C without
   modification.

3. The library can only handle NV12 and "ARGB" (which realy means any
   single plane, four channel, 8 bit format). This means it can't help
   with our inevitable future dominated by 10+ bit formats.

4. The pitch handling logic in the library is very inflexiable, and it
   assumes that for NV12, the UV plane is contiguous with the Y plane.
   This actually ends up making it incompatible with nvdec output for
   certain frame sizes. To avoid constantly fighting edge cases, I took
   the brute force approach and copy the input and output frames
   to/from CUarrays (which the library can accept) to give me a way to
   ensure the correct layout is used.

5. The library is stateful in an unhelpful way. It is called with one
   input frame, and one output buffer and always interpolates between
   the passed input frame and the frame from the previous call. This
   both requires special handling for the first frame, and also
   prevents generating more than one intermediate frame. If you want
   to do 3x or 4x etc interpolation, this approach doesn't work.

   So, again, I brute forced it by treating every interpolation as a
   new session - calling it twice with each input frame, even if the
   first frame happens to be the same as the last frame we called it
   with. This allows us to generate as many intermediate frames as we
   want, but it presumably consumes more GPU resources.

6. The library always creates a `NvOFFRUC` directory with an empty log
   file in it in $PWD. What a niusance.

But with all those caveats and limitations, it does actually work. I
was able to upsample a 24fps file to 144fps (my monitor limit) with
respectable results. In some situations, it starts bogging down, and
I'm not entirely sure where those limits are - certainly I can see it
consuming a significant percentage of GPU resources for large scaling
factors.

The implementation here is heavily based on vf_framerate with the
blending function ripped out and replaced by NvOFFRUC. That means we
have all the nice properties in terms of being able to do non-integer
scaling, and downsampling via interpolation as well.

Is this mergeable? No - but it was an interesting exercise and maybe
folks in narrow circumstances may find some genuine use from it.

Changed files

configure
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_nvoffruc.c

Change #279884

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	25a955f91a9ce493a521d622238f4b952c258516

Comments

avfilter/vf_nvoffruc: add support for `source_fps` expressions
The current code can only accept absolute values for the requested fps,
but it's often useful to specify a multiple of the video frame rate.

I've borrowed the code from `vf_fps` to implement this. You can read the
documentation for that filter to see how to use it, but the TLDR is you
can now use `fps=source_fps*2` or a similar expression. It also supports
the `ntsc`, `pal`, `film`, and `ntsc_film` aliases from `vf_fps`.

Changed files

libavfilter/vf_nvoffruc.c

Change #279885

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	43187a7ac5ae56553dfa35ade8adcf1067954ff0

Comments

avfilter/vf_nvoffruc: Drop use of internal.h
This was removed in ffmpeg.

Changed files

libavfilter/vf_nvoffruc.c

Change #279886

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a7a40351bfc6929227ec2defeeb6d3ec32c7ae7d

Comments

avfilter/vf_nvoffruc: update to use FFFilter

Changed files

libavfilter/allfilters.c
libavfilter/vf_nvoffruc.c

Change #279887

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d3327f9ae9cbba83976735c611c8cd518fbf2cca

Comments

avfilter/nvoffruc: add Windows build support
This was provided by [Usulyre](https://github.com/Usulyre)

Changed files

libavfilter/vf_nvoffruc.c

Change #279888

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7a065cfd577634f67941a5eedbd7363d53345b73

Comments

avfilter/nvoffruc: add diff for NvOFFRUC.h
Although I cannot distribute the header, I can include the diff that has to be
applied to make ffmpeg (or for that matter, anything) compile with it.

This diff includes fixes from [Usulyre](https://github.com/Usulyre). Thanks!

Changed files

libavfilter/nvoffruc-ffmpeg.diff
libavfilter/vf_nvoffruc.c

Change #279889

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0abbcdbdc4a3d6668bf1fc6022b2e2dffafa7922

Comments

avfilter/nvoffruc: add fallback for new timebase being unexpressible
In some situations the derived timebase can be unexpressible using an
AVRational, if the denominator is larger than an int. Today this causes the new
timebase to collapse to 0/1, which is unusable. In such situations, fall back
to using 1/frame_rate as this is significantly better than nothing.

Changed files

libavfilter/vf_nvoffruc.c

Change #279890

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d42a94ccd889c7caf14479912bd64f5b5fb7dad9

Comments

vf_nvoffruc: update hw_frames_ctx on passthrough frames
Fixes frame context mismatch when passthrough or cloned frames
don't inherit the output link's hw_frames_ctx, causing hwdownload
to fail with 'No frames received' errors.

Changed files

libavfilter/vf_nvoffruc.c

Change #279891

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ebcbde82969f6bdf57c6b369676500580c41dedc

Comments

avfilter/vf_nvoffruc: Add the NvOFFRUC.h header
This is not redistributable

Changed files

libavfilter/NvOFFRUC.h

Change #279892

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a9a90eaa647d4b764244f72d2383e447228c0502

Comments

avfilter/nvoffruc: Add P010 support
This assumes the use of my patched library with P010 support, of course.

Changed files

libavfilter/NvOFFRUC.h
libavfilter/vf_nvoffruc.c

Change #279893

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b4937a4fbb4f72132b00186055d60818599c7bdc

Comments

lavu/hwcontext: reverse order of teardown of frames context and pool
This avoids a validation failure saying an image is still in use by a
command buffer when the image is destroyed. Reversing the order
ensures the command pool and buffers are drained before images are
destroyed.

It's unclear if this is truly an error - as all the indications are
that the buffer in question is properly submitted and completed at
the time the image is freed.

This change obviously would impact all hwcontexts, but at a glance,
no context implementation seems to rely on the pool being freed before
the frames context.

Changed files

libavutil/hwcontext.c

Change #279894

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3d92d5b14421af94c404a17ecef1d8adc83cfb67

Comments

nvdec: Attempt to support h.263 decoding

Changed files

libavcodec/h263dec.c
libavcodec/nvdec.c
libavcodec/nvdec_mpeg4.c

Change #279895

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c9edae942e428419bebb79068037614945cda70c

Comments

avformat/hls: Try and mark discontinuous packets

Changed files

libavformat/hls.c

Change #279896

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:37:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f5b6ea3b65cc9d39556b4a51ca85b4125b6a5569

Comments

avcodec: add bink2 video decoder

Changed files

configure
libavcodec/Makefile
libavcodec/allcodecs.c
libavcodec/bink2.c
libavcodec/bink2f.c
libavcodec/bink2g.c
libavcodec/codec_desc.c
libavcodec/codec_id.h
libavformat/bink.c

Change #279897

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	af0d0cbd57ceb66d219355da30b3678b096a27f5

Comments

avutil/hwcontext_vulkan: fix the GBRAPF16 Vulkan format mapping
AV_PIX_FMT_GBRAPF16 is a floating point format, as the name suggests, so it should
be mapped to VK_FORMAT_R16_SFLOAT.

The comparable GBRPF16 and GBRAPF32 both already use SFLOAT.

Changed files

libavutil/hwcontext_vulkan.c

Change #279898

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4d8037a8d5adeb327f8836fdb2047aeba56cbcb6

Comments

avutil/hwcontext_vulkan: enable VK_NV_optical_flow
Enable VK_NV_optical_flow if it is present, as an optional extension. The
extension was added in 1.3.230 which is older than the minimum required
header release, so we don't need #ifdef guards.

Changed files

libavutil/hwcontext_vulkan.c

Change #279899

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	acddf16353d1755052dc49391e906e139a9bf4c0

Comments

avfilter/fruc_vulkan: copy vf_framerate.c verbatim to vf_fruc_vulkan.c
The fruc_vulkan filter will re-use the frame timing logic from vf_framerate -
it works so we don't need to do something different. To maximise reviewability,
this commit simply copies the file into place, so that all the changes in the
subsequent commits can be more easily understood.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279900

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b16c028002fca314c41b30e32f16db0462603fa3

Comments

avfilter/fruc_vulkan: wire up filter as a copy of vf_framerate
This is the minimal level of renames required to get the new filter to function
as a clone of vf_framerate. The context type is inlined from the header, but
there are no actual functional changes.

Changed files

configure
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_fruc_vulkan.c

Change #279901

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b58f426684d5cba48fe39d8b91ef36cdb1fd1f80

Comments

avfilter/fruc_vulkan: evaluate the output frame rate as an expression
This change enhances the `fps` parameter to support the `source_fps`
expressions from `vf_fps`, which are highly useful if you want to request a
multiple of the base frame rate.

The NTSC/PAL constants from `vf_fps` are not brought over. These are very
unlikely to be useful targets for a filter that is increasing the framerate of
a video.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279902

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9f60ec3010a8d93d15f6afd4f436cae82726814b

Comments

avfilter/fruc_vulkan: fall back to a plain timebase when reduction degenerates
The output timebase is derived, as in vf_fps / vf_framerate, by reducing the
source timebase against the requested frame rate. That reduction can collapse to
a zero time base - av_reduce() bounds its result, so the numerator can underflow
to zero, leaving 0/1 - when the source timebase shares no useful factors with
the requested rate, leaving an unusable output timebase that breaks the
downstream PTS rescaling.

Detect the degenerate result and fall back to the plain 1/fps timebase derived
directly from the requested rate, so config_output() always yields a valid
timebase.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279903

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0dd13033a18ce7091ac5313c10ccbd0b1a83ed9a

Comments

avfilter/fruc_vulkan: turn the filter into a Vulkan filter
This change introduces the basic Vulkan filter machinery, while preserving the
original vf_framerate timing logic. At this stage, the filter simply picks the
temporally closest input frame for the output.

Note that while input frames are not modified in anyway, they must be
deep-copied for output. This ensures that they match the output context, which
may differ from the input context in important way, and avoids extending the
lifetime of hardware decoded frames that are part of the decoder's DPB.

This requires exposing vkCmdCopyImage in the Vulkan function loader, and a
compute queue family and execution pool to record the copy on.

The filter depends on vulkan_1_4 for its Vulkan-Headers requirement; while this
is not required by any of the code in this commit, the final optical-flow
pipeline will need VK_KHR_maintenance9, which first appeared with Vulkan 1.4.
To avoid confusion, the requirement is asserted from the beginning.

Changed files

configure
libavfilter/Makefile
libavfilter/vf_fruc_vulkan.c
libavutil/vulkan_functions.h

Change #279904

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2133f67b2f10262f641d1e9c7e06b8c3f42a9d51

Comments

avutil/vulkan: select nearest mipmap filtering for nearest samplers
ff_vk_init_sampler() always used VK_SAMPLER_MIPMAP_MODE_LINEAR for samplers with
normalized coordinates. But it is possible to have an image view that uses
normalized coordinates while not supporting linear filtering (eg: R16G16_SINT).

The simple solution is to always use nearest mipmap filtering if the image
filter is nearest.

Changed files

libavutil/vulkan.c

Change #279905

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d6c932b563599588a67541541ccd0257a86853a5

Comments

avfilter/fruc_vulkan: add the initial optical-flow pipeline
Add the Vulkan optical-flow machinery on top of the filter skeleton:
* a grayscale-extraction compute shader
* the VK_NV_optical_flow session and its persistent input/flow images
* the cross-queue (compute -> optical flow -> compute) submission with
  binary-semaphore ordering
* an interpolation compute shader dispatched per output frame which blends each
  new frame instead of repeating the nearest source frame

Notable characteristics:

* At this stage we are doing a simple proportional blend that doesn't actually
  look at the optical flow information. That will happen in the next commit
* The optical flow is stored in a weird nvidia specific fixed point image
  format. Yay?
* Optical flow is calculated based purely on Luma. For YUV format, this is
  trivially obtained, but for RGB input, we should put some effort into getting
  accurate Luma. But we only need relative Luma differences, so we don't have
  to use the exactly correct colourspace conversion; I've just hard-coded
  BT.709
* We're doing host-synchronisation for initial simplicity, but this will get
  replaced in subsequent changes
* The accepted format set is deliberately narrow to begin with: planar and
  semi-planar YUV, gray, and packed RGB whose sampled channels are already
  in component order. Subsequent changes will expand the support set.
* We rely on VK_KHR_maintenance9 to enable implicit ownership transfers for
  images between the compute and optical flow queue families.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/Makefile
libavfilter/vulkan/fruc_grayscale.comp.glsl
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279906

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	70b82911557ea5182f9c082115fedc4a794aeb00

Comments

avfilter/fruc_vulkan: expose optical flow perf level and grid size
Add the two options that tune the NV optical flow session:

* perf: performance level (slow/medium/fast), previously hard-coded to slow.
  Trades motion-estimation quality for speed.
* grid: output grid size in pixels (auto/1/2/4/8), previously always the finest
  grid the device supported. A coarser grid computes fewer flow vectors, so it
  is faster and uses less memory at the cost of detail.

These settings are important as you can't handle high resolution input in
realtime at the highest quality settings. For example, 2160p24 movie content
needs reduced settings (eg: pref=medium,grid=2) for the filter to run
comfortably.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c

Change #279907

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c3a1b5e0ada46dec6a5257748da15e763f4a4637

Comments

avfilter/fruc_vulkan: warp frames with the computed optical flow
Replace the placeholder temporal blend in the interpolation shader with a
motion-compensated algorithm. This is finally the point where we are actually
using the optical flow information.

For each output pixel the source position on the f0 and f1 grids is recovered
by fixed-point Picard iteration before the samples are blended. The number of
iterations and the relaxation factor are established through empirical testing
on a couple of ad-hoc samples, representing highly uniform and non-uniform
motion respectively.

The approach here is hardly profound, and not meant to be. There are for sure
many more advanced algorithms for doing motion interpolation from optical flow,
and we can happily replace it with, or add, something more advanced once all
the machinery is in place.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279908

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0b8913d087b3157351409d856e42715a07f8f74f

Comments

avutil/vulkan: add ff_vk_exec_add_dep_signal_sem
Add a helper to register a timeline-semaphore signal with an explicit value on
an execution context, mirroring ff_vk_exec_add_dep_wait_sem() on the wait side.

The fruc_vulkan filter will use this to chain timeline semaphores across its
pipelined cross-queue submissions, where an interpolation must signal a
specific (generation-keyed) value rather than the implicit binary signal.

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #279909

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4dc6c1a6288381ea2cc3d8ae7f1936e6697fb3f8

Comments

avfilter/fruc_vulkan: pipeline the cross-queue submissions
The grayscale, optical flow and interpolation stages are currently serialised
with host fence waits (ff_vk_exec_wait) after each of the the grayscale and
optical flow submissions.

We can remove these waits and manage everything with timeline semaphores. In
particular, we must correctly handle the case where more than one frame is
being interpolated per optical flow pair. In these situations, we have multiple
interpolations waiting on the optical flow generation, and then we have the
next optical flow generation waiting on all the previous interpolations.

This requires the ff_vk_exec_add_dep_signal_sem() we added in the previous
commit.

We also introduce a second context to the optical flow exec pool so that the
next command buffer can be recorded without a host-wait for the previous one
to retire - but of course, optical flow is serial so the queued one won't
actually execute until the previous one completes.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279910

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3f1a8edbb8102f4d9d1d5e36cd5028e526f41a3b

Comments

avfilter/fruc_vulkan: double-buffer optical flow resources
The grayscale inputs, flow images and optical flow session were a single shared
instance, so each source pair's optical flow execution had to wait for the
previous pair's interpolations to finish reading the flow images before it could
overwrite them, serialising the optical flow engine against the compute queue.

Double-buffer the per-pair resources across two slots (the optical flow session
bakes in its image bindings, so each slot owns its own session, grayscale and
flow images), indexed by the pair generation. One pair's optical flow execution
can then run while the previous slot's flow images are still being sampled by
the earlier pair's interpolations. The write-after-read fence on slot reuse is
tracked per slot (interp_done) instead of against the single shared instance.

As the optical flow engine is a single serial unit, more slots can't increase
throughput.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279911

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	df77e84cb4f4b3041859692d361af58b5734bcaf

Comments

avfilter/fruc_vulkan: suppress spurious flow in featureless regions
In textureless regions the optical flow engine has no data to track and
its regularizer invents a smooth, often large, and internally
self-consistent flow field. Backward warping then traces a background
pixel along that invented flow onto a nearby moving object and pulls
object pixels into what should be empty background, leaving a ghost
outline at a distance from the object where nothing is actually moving.
Forward/backward flow consistency does not catch this because the
invented flow round-trips cleanly.

I have no idea if this is a common limitation in optical flow implementations
but it's definitely happening with the nvidia hardware, so this change
attempts to introduce a heuristic to work around it.

When the luma at a given location is sufficiently identical in both frames,
and the optical flow magnitude is sufficiently large, we will now assume that
the location is static and the flow is bogus. Genuine motion will usually lead
to luma differences, and low magnitude flow will be respected either way.

Again, this is all highly empirical, and the constants are justified purely by
by fiddling around with specific samples.

Changed files

libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279912

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	ef7317dc4c3154acd86c1d7c6881fa575f816724

Comments

avfilter/fruc_vulkan: support planar RGB and additional packed formats
We have previously only supported some packed RGB and planar YUV, so the logic
for extracting the luma was not flexible enough to handle other scenarios.

To generalise this for planar RGB, we need to bind every colour plane to the
grayscale shader and derive luma from it.

Additionally, we add support packed 4:4:4 YUV formats.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_grayscale.comp.glsl
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279913

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:05
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1250049b6f14d06c318f89a6c6d857591649accd

Comments

avfilter/fruc_vulkan: don't recalculate warping for every plane
The warp calculation is currently done for every plane, but doesn't depend on
the plane contents, just the dimensions; so this is wasted work when plane
sizes don't vary (eg: RGB, 4:4:4 YUV).

This will not have a visible impact on performance at realistic interpolation
rates as the optical flow generation is the limiting factor until you are
doing something silly like 30->900fps.

But wasted work is wasted work, so best to stop doing it.

Changed files

libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279914

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:22
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	6dcef57f0c902da2a8e11b839d6af390d981ae96

Comments

Merge branch 'fruc_vulkan_pr'

Changed files

no files

Change #279915

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 01:38:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	63ee38cd7f181691c5c374f894768e0dff63a7e9

Comments

Merge branch 'nvidia-video-filters'

Changed files

no files

Change #279916

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aea4deed0c0e61042d64ded8e48df721aaf6aea7

Comments

avutil/hwcontext_vulkan: fix the GBRAPF16 Vulkan format mapping
AV_PIX_FMT_GBRAPF16 is a floating point format, as the name suggests, so it should
be mapped to VK_FORMAT_R16_SFLOAT.

The comparable GBRPF16 and GBRAPF32 both already use SFLOAT.

Changed files

libavutil/hwcontext_vulkan.c

Change #279917

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e7976dd12f4907c7b020425a193fd88ab5a914f9

Comments

avutil/hwcontext_vulkan: enable VK_NV_optical_flow
Enable VK_NV_optical_flow if it is present, as an optional extension. The
extension was added in 1.3.230 which is older than the minimum required
header release, so we don't need #ifdef guards.

Changed files

libavutil/hwcontext_vulkan.c

Change #279918

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c6d56731b7ee5e4bc2e069707b3c31af49b8a83d

Comments

avfilter/fruc_vulkan: copy vf_framerate.c verbatim to vf_fruc_vulkan.c
The fruc_vulkan filter will re-use the frame timing logic from vf_framerate -
it works so we don't need to do something different. To maximise reviewability,
this commit simply copies the file into place, so that all the changes in the
subsequent commits can be more easily understood.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279919

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	91a51a797f78611b177006a2ea68c5250e6ec67d

Comments

avfilter/fruc_vulkan: wire up filter as a copy of vf_framerate
This is the minimal level of renames required to get the new filter to function
as a clone of vf_framerate. The context type is inlined from the header, but
there are no actual functional changes.

Changed files

configure
libavfilter/Makefile
libavfilter/allfilters.c
libavfilter/vf_fruc_vulkan.c

Change #279920

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	642135d007a6f00b408dd3e87c1685727716802f

Comments

avfilter/fruc_vulkan: evaluate the output frame rate as an expression
This change enhances the `fps` parameter to support the `source_fps`
expressions from `vf_fps`, which are highly useful if you want to request a
multiple of the base frame rate.

The NTSC/PAL constants from `vf_fps` are not brought over. These are very
unlikely to be useful targets for a filter that is increasing the framerate of
a video.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279921

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0cb9bd6da359465ed0a3af37e4e40b29b34d4b2e

Comments

avfilter/fruc_vulkan: fall back to a plain timebase when reduction degenerates
The output timebase is derived, as in vf_fps / vf_framerate, by reducing the
source timebase against the requested frame rate. That reduction can collapse to
a zero time base - av_reduce() bounds its result, so the numerator can underflow
to zero, leaving 0/1 - when the source timebase shares no useful factors with
the requested rate, leaving an unusable output timebase that breaks the
downstream PTS rescaling.

Detect the degenerate result and fall back to the plain 1/fps timebase derived
directly from the requested rate, so config_output() always yields a valid
timebase.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279922

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	67d341f49eba27a8e52404bdce8915ad07a1c12f

Comments

avfilter/fruc_vulkan: turn the filter into a Vulkan filter
This change introduces the basic Vulkan filter machinery, while preserving the
original vf_framerate timing logic. At this stage, the filter simply picks the
temporally closest input frame for the output.

Note that while input frames are not modified in anyway, they must be
deep-copied for output. This ensures that they match the output context, which
may differ from the input context in important way, and avoids extending the
lifetime of hardware decoded frames that are part of the decoder's DPB.

This requires exposing vkCmdCopyImage in the Vulkan function loader, and a
compute queue family and execution pool to record the copy on.

The filter depends on vulkan_1_4 for its Vulkan-Headers requirement; while this
is not required by any of the code in this commit, the final optical-flow
pipeline will need VK_KHR_maintenance9, which first appeared with Vulkan 1.4.
To avoid confusion, the requirement is asserted from the beginning.

Changed files

configure
libavfilter/Makefile
libavfilter/vf_fruc_vulkan.c
libavutil/vulkan_functions.h

Change #279923

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	73a353c826633d8414483aab374f5a2c4038b094

Comments

avutil/vulkan: select nearest mipmap filtering for nearest samplers
ff_vk_init_sampler() always used VK_SAMPLER_MIPMAP_MODE_LINEAR for samplers with
normalized coordinates. But it is possible to have an image view that uses
normalized coordinates while not supporting linear filtering (eg: R16G16_SINT).

The simple solution is to always use nearest mipmap filtering if the image
filter is nearest.

Changed files

libavutil/vulkan.c

Change #279924

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	2f00078367ee7e297233b042e32ce4221bd84119

Comments

avfilter/fruc_vulkan: add the initial optical-flow pipeline
Add the Vulkan optical-flow machinery on top of the filter skeleton:
* a grayscale-extraction compute shader
* the VK_NV_optical_flow session and its persistent input/flow images
* the cross-queue (compute -> optical flow -> compute) submission with
  binary-semaphore ordering
* an interpolation compute shader dispatched per output frame which blends each
  new frame instead of repeating the nearest source frame

Notable characteristics:

* At this stage we are doing a simple proportional blend that doesn't actually
  look at the optical flow information. That will happen in the next commit
* The optical flow is stored in a weird nvidia specific fixed point image
  format. Yay?
* Optical flow is calculated based purely on Luma. For YUV format, this is
  trivially obtained, but for RGB input, we should put some effort into getting
  accurate Luma. But we only need relative Luma differences, so we don't have
  to use the exactly correct colourspace conversion; I've just hard-coded
  BT.709
* We're doing host-synchronisation for initial simplicity, but this will get
  replaced in subsequent changes
* The accepted format set is deliberately narrow to begin with: planar and
  semi-planar YUV, gray, and packed RGB whose sampled channels are already
  in component order. Subsequent changes will expand the support set.
* We rely on VK_KHR_maintenance9 to enable implicit ownership transfers for
  images between the compute and optical flow queue families.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/Makefile
libavfilter/vulkan/fruc_grayscale.comp.glsl
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279925

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	71e5dfbf890e16dcfad53c9d059ee0c32f949a2b

Comments

avfilter/fruc_vulkan: expose optical flow perf level and grid size
Add the two options that tune the NV optical flow session:

* perf: performance level (slow/medium/fast), previously hard-coded to slow.
  Trades motion-estimation quality for speed.
* grid: output grid size in pixels (auto/1/2/4/8), previously always the finest
  grid the device supported. A coarser grid computes fewer flow vectors, so it
  is faster and uses less memory at the cost of detail.

These settings are important as you can't handle high resolution input in
realtime at the highest quality settings. For example, 2160p24 movie content
needs reduced settings (eg: pref=medium,grid=2) for the filter to run
comfortably.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c

Change #279926

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	283565c4c905e3726b8a262e4429a9083c1b2146

Comments

avfilter/fruc_vulkan: warp frames with the computed optical flow
Replace the placeholder temporal blend in the interpolation shader with a
motion-compensated algorithm. This is finally the point where we are actually
using the optical flow information.

For each output pixel the source position on the f0 and f1 grids is recovered
by fixed-point Picard iteration before the samples are blended. The number of
iterations and the relaxation factor are established through empirical testing
on a couple of ad-hoc samples, representing highly uniform and non-uniform
motion respectively.

The approach here is hardly profound, and not meant to be. There are for sure
many more advanced algorithms for doing motion interpolation from optical flow,
and we can happily replace it with, or add, something more advanced once all
the machinery is in place.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279927

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c22044c2502c8aeb8db7814164c2b5d31663fc04

Comments

avutil/vulkan: add ff_vk_exec_add_dep_signal_sem
Add a helper to register a timeline-semaphore signal with an explicit value on
an execution context, mirroring ff_vk_exec_add_dep_wait_sem() on the wait side.

The fruc_vulkan filter will use this to chain timeline semaphores across its
pipelined cross-queue submissions, where an interpolation must signal a
specific (generation-keyed) value rather than the implicit binary signal.

Changed files

libavutil/vulkan.c
libavutil/vulkan.h

Change #279928

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	fa7eebb1c989cc5b5f4f4eabc6c0b3e7e2ad43db

Comments

avfilter/fruc_vulkan: pipeline the cross-queue submissions
The grayscale, optical flow and interpolation stages are currently serialised
with host fence waits (ff_vk_exec_wait) after each of the the grayscale and
optical flow submissions.

We can remove these waits and manage everything with timeline semaphores. In
particular, we must correctly handle the case where more than one frame is
being interpolated per optical flow pair. In these situations, we have multiple
interpolations waiting on the optical flow generation, and then we have the
next optical flow generation waiting on all the previous interpolations.

This requires the ff_vk_exec_add_dep_signal_sem() we added in the previous
commit.

We also introduce a second context to the optical flow exec pool so that the
next command buffer can be recorded without a host-wait for the previous one
to retire - but of course, optical flow is serial so the queued one won't
actually execute until the previous one completes.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279929

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d1f157dcc1609eeecc7c8eba8657072ca1172c2b

Comments

avfilter/fruc_vulkan: double-buffer optical flow resources
The grayscale inputs, flow images and optical flow session were a single shared
instance, so each source pair's optical flow execution had to wait for the
previous pair's interpolations to finish reading the flow images before it could
overwrite them, serialising the optical flow engine against the compute queue.

Double-buffer the per-pair resources across two slots (the optical flow session
bakes in its image bindings, so each slot owns its own session, grayscale and
flow images), indexed by the pair generation. One pair's optical flow execution
can then run while the previous slot's flow images are still being sampled by
the earlier pair's interpolations. The write-after-read fence on slot reuse is
tracked per slot (interp_done) instead of against the single shared instance.

As the optical flow engine is a single serial unit, more slots can't increase
throughput.

Changed files

libavfilter/vf_fruc_vulkan.c

Change #279930

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:53
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a87d92d97523ef43354742c5f9cd12a59c904068

Comments

avfilter/fruc_vulkan: suppress spurious flow in featureless regions
In textureless regions the optical flow engine has no data to track and
its regularizer invents a smooth, often large, and internally
self-consistent flow field. Backward warping then traces a background
pixel along that invented flow onto a nearby moving object and pulls
object pixels into what should be empty background, leaving a ghost
outline at a distance from the object where nothing is actually moving.
Forward/backward flow consistency does not catch this because the
invented flow round-trips cleanly.

I have no idea if this is a common limitation in optical flow implementations
but it's definitely happening with the nvidia hardware, so this change
attempts to introduce a heuristic to work around it.

When the luma at a given location is sufficiently identical in both frames,
and the optical flow magnitude is sufficiently large, we will now assume that
the location is static and the flow is bogus. Genuine motion will usually lead
to luma differences, and low magnitude flow will be respected either way.

Again, this is all highly empirical, and the constants are justified purely by
by fiddling around with specific samples.

Changed files

libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279931

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f6d7c08944f8425b134f3a79b3569b5dadcf42ec

Comments

avfilter/fruc_vulkan: support planar RGB and additional packed formats
We have previously only supported some packed RGB and planar YUV, so the logic
for extracting the luma was not flexible enough to handle other scenarios.

To generalise this for planar RGB, we need to bind every colour plane to the
grayscale shader and derive luma from it.

Additionally, we add support packed 4:4:4 YUV formats.

Changed files

doc/filters.texi
libavfilter/vf_fruc_vulkan.c
libavfilter/vulkan/fruc_grayscale.comp.glsl
libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279932

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	964d4786ecea13ddf9354d8c4b0606db33a6067e

Comments

avfilter/fruc_vulkan: don't recalculate warping for every plane
The warp calculation is currently done for every plane, but doesn't depend on
the plane contents, just the dimensions; so this is wasted work when plane
sizes don't vary (eg: RGB, 4:4:4 YUV).

This will not have a visible impact on performance at realistic interpolation
rates as the optical flow generation is the limiting factor until you are
doing something silly like 30->900fps.

But wasted work is wasted work, so best to stop doing it.

Changed files

libavfilter/vulkan/fruc_interpolate.comp.glsl

Change #279933

Category	ffmpeg
Changed by	Philip Langdale <philiplohnoyoudont@overt.org>
Changed at	Sun 30 Aug 2026 02:07:54
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d20bae84e0d857d756a7fa8dbe150f7e3da549bd

Comments

avfilter: version bump for vf_fruc_vulkan
Also update Changelog

Changed files

Changelog
libavfilter/version.h

Change #279934

Category	ffmpeg
Changed by	iSold Leo <sohnoyoudont@qwqlog.com>
Changed at	Sun 30 Aug 2026 04:50:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d65fd5581d7e28dc063a35ccb7ef88ee0b7417f5

Comments

avdevice/avfoundation: keep format and frame rate range paired
Formats with the same dimensions may expose different frame rates. The
selection loop previously updated selected_format for every size match but
kept selected_range from an earlier format. This could apply a range to a
format that does not support it and make device configuration fail.

Only update selected_format together with its matching range. Track a
size-only match separately so the muxed-device fallback remains unchanged.

Reported-by: MW-de <mw-de@noreply.code.ffmpeg.org>
Suggested-by: MW-de <mw-de@noreply.code.ffmpeg.org>
Fixes: https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/22694
Signed-off-by: iSold Leo <s@qwqlog.com>

Changed files

libavdevice/avfoundation.m

Change #279938

Category	ffmpeg
Changed by	罗乾林 <luoqianlinohnoyoudont@sansi.com>
Changed at	Sun 30 Aug 2026 05:23:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e5d295b3cd467fc713300e208ba2dacc7ff603a0

Comments

avcodec/mediacodec_wrapper: avoid leaking setBuffer backing storage

Changed files

libavcodec/mediacodec_wrapper.c

Change #279939

Category	ffmpeg
Changed by	罗乾林 <luoqianlinohnoyoudont@sansi.com>
Changed at	Sun 30 Aug 2026 05:23:09
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b32f8d1c2377079302d23f82d555d13deda68c57

Comments

avcodec/mediacodecdec: free HEVC SEI parsed from extradata

Changed files

libavcodec/mediacodecdec.c

Change #279978

Category	ffmpeg
Changed by	Andreas Menze <andreas.menzeohnoyoudont@hr.de>
Changed at	Sun 30 Aug 2026 16:27:03
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7bced195dabc09959de95eaa0cd22b2b3df8a7b4

Comments

avformat/wavdec: cache avio_size() before footer seeks
Over HTTP, seeking to a trailing chunk can make avio_size() stale, so the PCM
duration guard fails and duration is N/A. Cache size before seeking.

Fixes #24122

Changed files

libavformat/wavdec.c

Change #279983

Category	ffmpeg
Changed by	James Almer <jamrialohnoyoudont@gmail.com>
Changed at	Sun 30 Aug 2026 18:03:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	818cecc6e1afab932cf4d40ef0d7b8cd40311a17

Comments

tests/api/movenc: add a test for fragmented output with no samples
Covers the fix from 9f35e220ffbba21c88356eb2bbfa3679ede93793.

Signed-off-by: James Almer <jamrial@gmail.com>

Changed files

tests/api/api-movenc-test.c
tests/ref/fate/api-movenc

Change #280073

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 12:51:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	28ab668edddfa37b09e94b33c19ff4a6ea7fd5a1

Comments

avfilter/vf_scdet: expand format list
This filter works by running the SAD function on the first plane, which
works for any (native endian) format that starts with an unshifted luma
plane. This notably includes formats like nv12, yuva420p, and grayscale
formats, as well as luma-alpha formats. The last addition is debatable,
but the precedent here is the existing support for RGBA, which also
dilutes scene scores by including the alpha component in the SAD
calculation). There were also some notable missing packed RGBA formats,
such as ARGB, as well as 16-bit variants.

This is an exhaustive listing as of the current set of formats.
Tested by looping over `format=<pixfmt>,scdet` and verifying that the
scene SAD score doesn't (meaningfully) change, except for expected
deviations (e.g. larger scores for full range vs limited range, and
lowered scores for packed alpha formats).

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/vf_scdet.c

Change #280074

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 12:51:35
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	eeca9d9280a92ef1536c154346c126cfb9d0ed6a

Comments

avfilter/vf_scdet: add support for MSB-aligned pixel formats
This is trivial since the SAD result is already safe to 16 bits, and
the result is already normalized by the bit depth.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/vf_scdet.c

Change #280076

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 12:56:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	dec1ffb86d8aa0bad663308fe129253287fba259

Comments

avcodec/j2kenc: fix edge case when packet ends on 0xFF
ISO/IEC 15444-1:2002 (E), Annex B.10.1 "Bit-stuffing routine":

> The last byte in the packet header shall not be an 0xFF value (thus the
> single zero bit stuffed after a byte with 0xFF must be included even if
> the 0xFF would otherwise have been the last byte).

This matches the behavior seen in the jpeg2000 decoder (jpeg2000_flush),
and the behavior of openjpeg.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavcodec/j2kenc.c

Change #280077

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 12:56:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	86416861a717eab0999512075fcb220e06955887

Comments

avcodec/j2kenc: add buffer size check before flushing
Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavcodec/j2kenc.c

Change #280078

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 12:56:08
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f88b741dbf242864e0c47b58e59dcbd67207c74b

Comments

avcodec/j2kenc: init bitstream after writing SOP
Otherwise, *sop remains uninitialized if the sop is written.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavcodec/j2kenc.c

Change #280106

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e768ade128e67a559efe8ae6e4e74acd26209ffc

Comments

avfilter/adelay: parse fractional second delays correctly
the initial integer scan stores '.' as the suffix for values such as 1.5s, causing them to be interpreted as milliseconds. inspect the actual final suffix instead.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_adelay.c

Change #280107

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	25c2c2338780344df7a8f0c97c81e86513a530f2

Comments

avfilter/adelay: reject delays outside int64 range
converting a non-finite or too large sample count to int64_t is undefined. validate the value before conversion.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_adelay.c

Change #280108

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	7f9b5b057cea21f24dc9e0a25e93fd3a07d970cd

Comments

avfilter/aderivative: saturate integer differences
full-scale transitions overflow or wrap integer sample formats. use saturating subtraction for the integer paths.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_aderivative.c

Change #280109

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4021d0d193932c65cc03a210da7a44c773ce36a5

Comments

avcodec/wavpackenc: handle INT32_MIN samples
abs() cannot represent the magnitude of INT32_MIN. labs() has the same issue on platforms where long is 32 bits. use FFABSU for the full signed sample range.

Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavcodec/wavpackenc.c

Change #280110

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	71f9ea2d840ee69482c12c79c9987a641331c1f8

Comments

avfilter/aecho: reject non-finite delays and decays
Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_aecho.c

Change #280111

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	31e92a56caad583fbe3e868947382f38e8bb6abc

Comments

avfilter/firequalizer: reject overflowing delays
Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_firequalizer.c

Change #280112

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0ba7514f995eb2d598cc574c7b459cb64e122c86

Comments

avfilter/acrossover: reject non-finite parameters
Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_acrossover.c

Change #280113

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	1d49dc80a992b596fbe44d3d90e1d7b10b300091

Comments

avfilter/aiir: reject non-finite parameters
Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_aiir.c

Change #280114

Category	ffmpeg
Changed by	Ayoub Nabil Boubagrat <237098474+ayoubnabilohnoyoudont@users.noreply.github.com>
Changed at	Mon 31 Aug 2026 15:42:52
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	728655e2e7665e9059778953797a98dc6bb9d145

Comments

avfilter/aphaser: clip integer output
Signed-off-by: Ayoub Nabil Boubagrat <237098474+ayoubnabil@users.noreply.github.com>

Changed files

libavfilter/af_aphaser.c

Change #280115

Category	ffmpeg
Changed by	Paulo Cabral Sanz <paulosanzohnoyoudont@poli.ufrj.br>
Changed at	Mon 31 Aug 2026 15:44:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c2394a0ff339484e54f757114dd0e7502fdf2fab

Comments

avformat/binka: reject truncated packet reads
binka_read_packet() calls avio_read() into a buffer from av_new_packet()
and ignores the return value. packet_alloc() only zeroes the 64-byte
padding; the payload is left uninitialized.

pkt_size is avio_rl16() + 4 (max 65539). On a truncated file the demuxer
still returns 0, so av_read_frame delivers up to 64 KiB of uninitialized
heap per packet. ffmpeg -c copy writes those bytes into the output.

Use ffio_read_size() so a short read fails the packet with
AVERROR_INVALIDDATA. Same class as 29f513a (fsb) and 4b47405 (genh).

Signed-off-by: Paulo Cabral Sanz <paulosanz@poli.ufrj.br>

Changed files

libavformat/binka.c

Change #280116

Category	ffmpeg
Changed by	yongdev <yongdevohnoyoudont@google.com>
Changed at	Mon 31 Aug 2026 15:47:25
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c9e36046a338638279782cba4fba3299bf65f46b

Comments

libswscale: use unaligned read/write macros in planarCopyWrapper
When performing unscaled planar copies or format conversions on 16-bit
buffers in planarCopyWrapper, srcPtr2 and dstPtr2 are not guaranteed to
be 8-byte aligned (e.g. subsampled chroma planes with non-multiple-of-8
widths/strides or odd slice offsets).

AV_RN64A / AV_WN64A and AV_RN32A / AV_WN32A cast pointers to av_alias64
or av_alias32 which have __attribute__((aligned(8))) or aligned(4).
Dereferencing unaligned pointers with aligned attribute triggers Undefined
Behavior traps (SIGILL / UD1 instruction with Clang -fsanitize=alignment).
Replace them with AV_RN64 / AV_WN64 and AV_RN32 / AV_WN32.

Signed-off-by: yongdev <yongdev@google.com>

Changed files

libswscale/swscale_unscaled.c

Change #280137

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 19:22:59
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	486f03997ce4b4ecf61aba0ec14b3aa61f39fc02

Comments

avcodec/aacenc: add option to allow encoding non-PCE 7.1(wide)
PCE is not compatible with all downstream sources. Some browsers, for example,
will refuse to play an AAC file with a PCE channel layout. So there is a
legitimate use case for encoding non-PCE files if required.

However, due to the historical work-around where FFmpeg produced out-of-spec
files that mislabelled 7.1 as 7.1(wide), we should gate this behind a new
option.

I considered tying it to -strict 1 (which also disables the equivalent
work-around in aacdec), but this may lead to users unknowingly opting into
this behavior due to enabling -strict 1 out of habit (or for other reasons).

This way, a user who really knows what they're doing can use
-af aformat=channel_layouts=7.1(wide) -aac_allow_71wide yes to force
producing a spec-compliant non-PCE AAC file.

I decided to keep the option undocumented for now, matching the existing
`aac_pce` option (also undocumented).

The channel IDs were taken by comparing the AAC bitstream channel order:
 FC, FLC, FRC, FL, FR, BL, BR, LFE

Aginst the FFmpeg channel order for 7POINT1_WIDE_BACK:
 0=FL 1=FR 2=FC 3=LFE 4=BL 5=BR 6=FLC 7=FRC

This gives us {2, 6, 7, 0, 1, 4, 5, 3}.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavcodec/aacenc.c
libavcodec/aacenc.h
libavcodec/aacenctab.h

Change #280140

Category	ffmpeg
Changed by	Timo Rothenpieler <timoohnoyoudont@rothenpieler.org>
Changed at	Mon 31 Aug 2026 20:09:20
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4efbbd15438413bc2224f15edc4cbb080cd25896

Comments

forgejo/inject-pr-samples: properly flush and close injected sample before hashing

Changed files

.forgejo/inject-pr-samples.py

Change #280146

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	a28964729781a79e3668f441fde4a9025e537d3e

Comments

avcodec/h264dec: add data partitioning fields to H264SliceContext
A partitioned slice carries its residual in NAL units of its own, so
the decoder needs a bit reader per partition next to the one for
partition A. They are values rather than pointers because the slice
context is memcpy'd when slices are reordered.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

libavcodec/h264dec.h

Change #280147

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	f017d0eefb27af25d260c34bc96aeaae3bbd8159

Comments

avcodec/h264_slice: parse slice_id from partition A
7.3.2.9.1 appends slice_id to the slice header of a partition A. It is
what partitions B and C refer back to, so it has to be read and range
checked before they can be matched to their A. 7.4.2.9.1 bounds it by
PicSizeInMbs, which is half the frame size for a field picture and for
MBAFF.

Data partitioning belongs to the Extended profile, which does not allow
CABAC, so a PPS asking for it is rejected here.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

libavcodec/h264_slice.c

Change #280148

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0e00e1b3d8c940cba30c084af045859ec8b44ff5

Comments

avcodec/h264: decode slice data partitions B and C
Syntax elements of category 2 stay in partition A, category 3 (intra
residual) moves to partition B and category 4 (inter residual) to
partition C, so CAVLC has to read the residual of each macroblock from
the partition that matches its type. A partition that never arrived is
reported instead of being read from A.

B and C are taken from the NAL units immediately following their A and
matched to it by slice_id. Arbitrary slice order may separate them
(7.4.1.2.5); that is not supported, and such a stream reports the
missing partition instead. Attaching runs while the slice is queued,
before any slice is executed, so no partition can arrive too late to
be used. ff_h264_queue_decode_slice() reports which context it queued,
as a slice opening a second field is moved into slot zero once the
pending field has been decoded.

Attaching and routing land together on purpose: enabling the partition
A path on its own would decode partitioned streams incorrectly instead
of reporting them unsupported. hwaccel formats are not offered for a
partitioned picture and CODEC_FLAG2_CHUNKS is refused, as both assume
one self-contained slice NAL.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

libavcodec/h264_cavlc.c
libavcodec/h264_slice.c
libavcodec/h264dec.c
libavcodec/h264dec.h

Change #280149

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3e59635d1e04bbf29e10e276a027ab9aa1204471

Comments

avcodec/h264_parser: recognise partition A as a picture start
parse_nal_units() takes slice_type and the POC fields from the first
VCL NAL of a picture, but recognised only plain and IDR slices. A
partition A opens with a slice header too, so until now a partitioned
stream left the parser reporting no picture type at all. It needs the
same leading bytes as any other VCL NAL, so it gets the same source
length cap rather than having its whole RBSP extracted.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

libavcodec/h264_parser.c

Change #280150

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	74f1a4a16027331f0e0c95feee6330f29ba18387

Comments

avcodec: add FF_BUG_H264_DP_NNZ
9.2.1 makes an inter neighbour unavailable for the coeff_token nC
derivation under constrained intra prediction with slice data
partitioning, and step 7 then drops it from the (nA + nB + 1) >> 1
average. JM instead treats it as available with nN = 0, so it still
contributes. Every partitioned stream in the wild comes from JM or a
derivative, so the decoder has to be able to do both.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

doc/APIchanges
doc/codecs.texi
libavcodec/avcodec.h
libavcodec/options_table.h
libavcodec/version.h

Change #280151

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e13b8370910351d9601dd9ead7d3ce77fff6869b

Comments

avcodec/h264: derive nC for partitioned constrained-intra slices
An inter neighbour must not contribute its coefficient count to nC for
an intra macroblock when constrained intra prediction and data
partitioning are both in use; otherwise decoding partitions A and B
alone would depend on data that lives in partition C.

The neighbour is kept available with nN = 0, which is what JM does and
what the streams it produces expect. FF_BUG_H264_DP_NNZ selects that
and is autodetected on the first partition A, while -bug 0 drops the
neighbour from the average as 9.2.1 has it. Streams without data
partitioning decode identically either way.

This replaces a FIXME older than any data partitioning support.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

libavcodec/h264_mvpred.h
libavcodec/h264dec.c

Change #280152

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5212830ee809d09fec41b220871cb9eeab87716c

Comments

Changelog,doc: note H.264 data partitioning support
The decoder now handles nal_unit_type 2 to 4. There is no encoder
support.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

Changelog
doc/general_contents.texi

Change #280153

Category	ffmpeg
Changed by	guy-with-a-why <aaronohnoyoudont@iterationlab.io>
Changed at	Mon 31 Aug 2026 20:13:21
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3b5b6298fbf62b4dabf6f8c393ee4deb584246a9

Comments

fate/h264: add slice data partitioning tests
Three JM-encoded streams: all three partitions present, partition C
legitimately absent, and constrained_intra_pred_flag set, which is the
case that exercises the nC derivation of 9.2.1.

The last stream is decoded both ways. -bug 0 selects the literal 9.2.1
reading, which desynchronises the residual of a JM-encoded stream; the
concealed output then depends on the thread count, so that test pins
threads to 1.

Signed-off-by: guy-with-a-why <aaron@iterationlab.io>

Changed files

tests/fate/h264.mak
tests/ref/fate/h264-data-partitioning
tests/ref/fate/h264-data-partitioning-ab
tests/ref/fate/h264-data-partitioning-cip
tests/ref/fate/h264-data-partitioning-cip-strict

Change #280158

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cf7f75ebf19b52a83003cac5581e2a1e9d40cab3

Comments

avformat/shared: don't acquire more blocks on already errored caches
Correctly mirrors the logic used to decide whether to write back to the
cache or not.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280159

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3d8488bed763e01e5d833cd253333ee6c0e4b010

Comments

avformat/shared: clean up PENDING state on error/early exit
This makes sure threads that fail for reasons other than the underlying I/O
failing clean up after their own PENDING state on failure, unless another
thread updated the block state in the meantime.

Subsumes the existing "is_race" condition, which is inverted to "acquired"
that is 1 exactly when we were the thread that set the PENDING state.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280160

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	e7965fac92fa5184e1f9ba19fdcf37dff3f68aad

Comments

avformat/shared: correctly early-exit on read past last block end
Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280161

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	5c07f7d8db791ef9d0f2208be3faf2328aec6ca8

Comments

avformat/shared: early exit on prior spacemap I/O failure
The code implicitly assumes that the caller won't re-call read()/seek()
after observing a prior AVERROR(EIO). However, this is not necessarily a
future-proof guarantee, so it's safer to explicitly re-error in this case.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280162

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	75cf5752574b6f178d12b16cdc82e5d9521affde

Comments

avformat/shared: hard-error on -cache_verify data mismatch
Makes this debug option way more useful. Also matches the existing behavior
on CRC mismatches.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

doc/protocols.texi
libavformat/shared.c

Change #280163

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4f7a381a0cfdeefb342d37d5640b2349e0da0b22

Comments

avformat/shared: allow interruption during BLOCK_PENDING read loop
Also allows nonblocking calls.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280164

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0a83c9660dccf803e260d6e3ca2ab47380457838

Comments

avformat/shared: don't consider EINTR a hard failure
Don't trigger `write_err` for this transient failure.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280165

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	83412412a492911609a61c4782e3d9364d9c2b2f

Comments

avformat/shared: don't re-read spacemap block size per call
This would otherwise trigger OOB/UB if the spacemap header is corrupted.
Instead, read it once during init, verify it there, and then cache that
file for the remainder of the process.

Sponsored-by: nxtedition AB
Reported-by: Mateusz Gierblinski <mateusz.gierblinski@curios-it.eu>
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280166

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	aa701bd99f336bfa0e9b1ea94c5824e3ea88441c

Comments

avformat/shared: don't re-read spacemap filesize per call
See previous commit for justification. This commit also addresses another
edge case where we neglected to bounds check the stored filesize against
the int64_t limits.

Sponsored-by: nxtedition AB
Reported-by: Mateusz Gierblinski <mateusz.gierblinski@curios-it.eu>
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280167

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	03a934512d0875d6c7ece802b4c4b9678182332b

Comments

avformat/shared: don't try opening spacemap if cache file failed
This avoids potentially clobbering `errno` if the s->mapfd fails
for a different reason, and is just cleaner in general.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280168

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Mon 31 Aug 2026 21:10:45
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	089a48eb36b2c0d185867522a1fc2613b5bbb96f

Comments

avformat/shared: add Spacemap header size assertion
Ensures the padding is correctly updated after additions.

Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavformat/shared.c

Change #280228

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Mon 31 Aug 2026 23:36:16
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	b1f564bda248dbe8ac37cb743966fdd7ee4b490c

Comments

hwcontext_vulkan: add support for selecting a device via text-form UUID
We had everything we need for years but somehow never did it.

Changed files

libavutil/hwcontext_vulkan.c

Change #280289

Category	ffmpeg
Changed by	Lynne <devohnoyoudont@lynne.ee>
Changed at	Tue 01 Sep 2026 12:49:37
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	72e50c4ca1f40d812f202b5608763f0a779b2bf1

Comments

hwcontext_vulkan: use memcmp instead of strncmp when testing the dev UUID
Found by Myrsloik, cheers

Changed files

libavutil/hwcontext_vulkan.c

Change #280290

Category	ffmpeg
Changed by	younengxiao <steven.xiaoohnoyoudont@amd.com>
Changed at	Tue 01 Sep 2026 12:59:14
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	44161610904f75b99b3f3ffd6dea4aa096b25516

Comments

avfilter/dnn: fix teardown hang in ONNX backend with nireq >= 2
The ONNX Runtime backend only ever runs one synchronous request, so
clamp nireq to 1 at load time (matching the actual allocation)

Signed-off-by: younengxiao <steven.xiao@amd.com>

Changed files

libavfilter/dnn/dnn_backend_onnx.c

Change #280291

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d6820fb8b3b49a7b2cc3fb5d3b379745ebcc97b9

Comments

avfilter/vf_colordetect: add support for full range alpha offset
This allows us to add a threshold for alpha comparisons to the full range
path as well.

The AArch64 changes were LLM-assisted, since I'm not familiar with that
architecture. However, the logic checks out to me.

Also adjusts the checkasm test to make sure there are no rounding differences
between the C and assembly implementations. Tested correct across 100k seeds
on x86.

checkasm:
 - CPU: AMD Ryzen 9 9950X3D 16-Core Processor (00B40F40)
 - Timing source: x86 (rdtsc)
 - Bench duration: 100000 µs per function (432788090 cycles)
 - Random seed: 4076092999

Benchmark results: (old)
  name                              cycles (vs ref)
  detect_alpha_8_full_c:            4670.7
  detect_alpha_8_full_avx2:          856.6 ( 5.45x)
  detect_alpha_8_full_avx512icl:     667.5 ( 7.00x)
  detect_alpha_16_full_c:           2705.7
  detect_alpha_16_full_avx2:         693.1 ( 3.90x)
  detect_alpha_16_full_avx512icl:    546.4 ( 4.94x)

Benchmark results: (new)
  name                              cycles (vs ref)
  detect_alpha_8_full_c:            7988.5
  detect_alpha_8_full_avx2:         1078.0 ( 7.41x)
  detect_alpha_8_full_avx512icl:     876.7 ( 9.11x)
  detect_alpha_16_full_c:           3734.7
  detect_alpha_16_full_avx2:         814.6 ( 4.58x)
  detect_alpha_16_full_avx512icl:    613.8 ( 6.08x)

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/aarch64/vf_colordetect_neon.S
libavfilter/vf_colordetect.c
libavfilter/vf_colordetectdsp.h
libavfilter/x86/vf_colordetect.asm
tests/checkasm/vf_colordetect.c

Change #280292

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	8b7e4b74b216cfc86a359e149aee1be7870c98ab

Comments

avfilter/vf_colordetect: precompute alpha detection fields
In particular, we also store s->mpeg_range explicitly due to the upcoming
commit, which will allow these values to deviates between the range detection
and alpha mode detection passes.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/vf_colordetect.c

Change #280293

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	0dc120b360b604c3a1505644f0863fb23c35e08c

Comments

avfilter/vf_colordetect: add threshold option
This is useful to filter false positives due to encoding noise in lossily
compressed sources.

The reason this is limited to 0.05 (= 5%) is twofold:

1. Higher values would mostly defeat the purpose of this filter, since a
   deviation of >5% is enough to fully blur the distinction between JPEG and
   MPEG range, let alone premultiplied and straight alpha. There's no useful
   signal to extract with such a high tolerance for noise.

2. Higher values would overflow the `offset` value in the 8-bit SIMD. This
   could also be solved by clamping, but the first point is the salient one
   anyways.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

doc/filters.texi
libavfilter/vf_colordetect.c

Change #280294

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9ee7e00bfa9be83d72c572c017030d1b8e5212e4

Comments

avfilter/vf_colordetect: increase threshold default to 0.001
I split this off from the commit adding the option to distinguish between the
change in code (which should be a no-op by default) and the change in behavior
(which is a deliberate deviation) when bisecting.

0.001 = 0.1% permits just one 10-bit code point of deviation, while
requiring tight bounds for 8-bit content.

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

doc/filters.texi
libavfilter/vf_colordetect.c

Change #280295

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	4bf82ad8df504cd73373f0c2e1fa1129a792c997

Comments

avfilter/x86/vf_colordetect: split into offset and non-offset variants
This partially addresses a a performance regression otherwise introduced by
f37b9a3f, which matters especially for 8-bit content where the default
threshold value rounds down to an offset of 0.

The numbers are still slower than before because the scalar tail handling
code is slower by ~200 cycles, since we benchmark across an unaligned
width.

checkasm:
 - CPU: AMD Ryzen 9 9950X3D 16-Core Processor (00B40F40)
 - Timing source: x86 (rdtsc)
 - Bench duration: 100000 µs per function (460518890 cycles)
 - Random seed: 2935750295

Benchmark results:
  name                                  cycles (vs ref)
  detect_alpha_8_full_c:                7877.4
  detect_alpha_8_full_avx2:             1063.4 ( 7.41x)
  detect_alpha_8_full_avx512icl:         872.1 ( 9.03x)
  detect_alpha_8_full_off_c:            7890.3
  detect_alpha_8_full_off_avx2:         1085.5 ( 7.27x)
  detect_alpha_8_full_off_avx512icl:     873.0 ( 9.04x)
  detect_alpha_16_full_c:               3700.9
  detect_alpha_16_full_avx2:             797.7 ( 4.64x)
  detect_alpha_16_full_avx512icl:        612.1 ( 6.05x)
  detect_alpha_16_full_off_c:           3701.8
  detect_alpha_16_full_off_avx2:         801.2 ( 4.62x)
  detect_alpha_16_full_off_avx512icl:    684.8 ( 5.41x)

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/vf_colordetect.c
libavfilter/vf_colordetectdsp.h
libavfilter/x86/vf_colordetect.asm
libavfilter/x86/vf_colordetect_init.c
tests/checkasm/vf_colordetect.c

Change #280296

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	d241c80e59e26eac47d448f54bc3629b75b89f34

Comments

avfilter/x86/vf_colordetect: avoid calling C code for tail handling
Instead, we can just call the SIMD function a second time. This will reprocess
a few overlapping pixels, but that is safe since this function is read-only
and idempotent.

We only need to bail to the C function if the overall width is smaller than
a single SIMD vector.

checkasm:
 - CPU: AMD Ryzen 9 9950X3D 16-Core Processor (00B40F40)
 - Timing source: x86 (rdtsc)
 - Bench duration: 100000 µs per function (462406519 cycles)
 - Random seed: 555761032

Benchmark results:
  name                                  cycles (vs ref)
  detect_alpha_8_full_c:                7884.5
  detect_alpha_8_full_avx2:              439.9 (17.88x)
  detect_alpha_8_full_avx512icl:         262.8 (30.00x)
  detect_alpha_8_full_off_c:            7884.0
  detect_alpha_8_full_off_avx2:          456.3 (17.27x)
  detect_alpha_8_full_off_avx512icl:     267.3 (29.49x)
  detect_alpha_8_limited_c:            13452.5
  detect_alpha_8_limited_avx2:          1163.5 (11.56x)
  detect_alpha_8_limited_avx512icl:      650.9 (20.67x)
  detect_alpha_16_full_c:               3696.1
  detect_alpha_16_full_avx2:             426.4 ( 8.67x)
  detect_alpha_16_full_avx512icl:        262.6 (14.07x)
  detect_alpha_16_full_off_c:           3698.7
  detect_alpha_16_full_off_avx2:         454.7 ( 8.13x)
  detect_alpha_16_full_off_avx512icl:    267.1 (13.85x)
  detect_alpha_16_limited_c:            8215.9
  detect_alpha_16_limited_avx2:         1288.4 ( 6.38x)
  detect_alpha_16_limited_avx512icl:     691.3 (11.88x)
  detect_range_8_c:                     1411.0
  detect_range_8_avx2:                   281.6 ( 5.01x)
  detect_range_8_avx512icl:              168.7 ( 8.36x)
  detect_range_16_c:                    1334.4
  detect_range_16_avx2:                  228.6 ( 5.83x)
  detect_range_16_avx512icl:             139.7 ( 9.50x)

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

libavfilter/x86/vf_colordetect_init.c

Change #280297

Category	ffmpeg
Changed by	Niklas Haas <gitohnoyoudont@haasn.dev>
Changed at	Tue 01 Sep 2026 13:15:06
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	adacc96aa789602d6cc36350f0b9f27f59ece540

Comments

tests/checkasm/vf_colordetect: benchmark on aligned width
This gives a more faithful representation of the SIMD speedup, as the
unaligned checkasm test case size is not representative of typical inputs.

checkasm:
 - CPU: AMD Ryzen 9 9950X3D 16-Core Processor (00B40F40)
 - Timing source: x86 (rdtsc)
 - Bench duration: 100000 µs per function (432431813 cycles)
 - Random seed: 2646589325

Benchmark results:
  name                                  cycles (vs ref)
  detect_alpha_8_full_c:                7335.9
  detect_alpha_8_full_avx2:              396.2 (18.51x)
  detect_alpha_8_full_avx512icl:         191.0 (38.41x)
  detect_alpha_8_full_off_c:            7324.1
  detect_alpha_8_full_off_avx2:          404.0 (18.13x)
  detect_alpha_8_full_off_avx512icl:     199.1 (36.74x)
  detect_alpha_8_limited_c:            12848.8
  detect_alpha_8_limited_avx2:          1088.7 (11.80x)
  detect_alpha_8_limited_avx512icl:      563.7 (22.79x)
  detect_alpha_16_full_c:               3360.0
  detect_alpha_16_full_avx2:             391.8 ( 8.57x)
  detect_alpha_16_full_avx512icl:        190.5 (17.64x)
  detect_alpha_16_full_off_c:           3362.8
  detect_alpha_16_full_off_avx2:         404.9 ( 8.31x)
  detect_alpha_16_full_off_avx512icl:    200.3 (16.76x)
  detect_alpha_16_limited_c:            7771.7
  detect_alpha_16_limited_avx2:         1210.8 ( 6.42x)
  detect_alpha_16_limited_avx512icl:     603.6 (12.88x)
  detect_range_8_c:                     1233.1
  detect_range_8_avx2:                   254.9 ( 4.84x)
  detect_range_8_avx512icl:              134.0 ( 9.20x)
  detect_range_16_c:                    1172.2
  detect_range_16_avx2:                  177.9 ( 6.58x)
  detect_range_16_avx512icl:              96.0 (12.20x)

Sponsored-by: nxtedition AB
Signed-off-by: Niklas Haas <git@haasn.dev>

Changed files

tests/checkasm/vf_colordetect.c

Change #280306

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 01 Sep 2026 13:45:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	cd352f0b1d1ad7a74460b6e83132bac8b94cafa0

Comments

tests/fate/filter-video: Detect peaks in ebur128 filter test
Reviewed-by: Niklas Haas <ffmpeg@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

tests/fate/filter-video.mak
tests/ref/fate/filter-metadata-ebur128

Change #280307

Category	ffmpeg
Changed by	Andreas Rheinhardt <andreas.rheinhardtohnoyoudont@outlook.com>
Changed at	Tue 01 Sep 2026 13:45:56
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	c27482a18d7e63a2680638929359a82790eb7631

Comments

avfilter/x86/f_ebur128: Avoid unnecessary movq
For all supported x86 ABIs, floating point values are returned
in xmm0, not in rax.

Reviewed-by: Niklas Haas <ffmpeg@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Changed files

libavfilter/x86/f_ebur128.asm

Change #280380

Category	ffmpeg
Changed by	Michael Niedermayer <michaelohnoyoudont@niedermayer.cc>
Changed at	Wed 02 Sep 2026 00:49:12
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	905ef92f022ac658caf3e50244deeb2d9e6f262d

Comments

avformat/sierravmd: require a 6bit palette in the probe
vmdvideo_decode_init() loads a 768 byte VGA palette from header offset
28 and scales each entry by 4, so every one of these bytes is 6bit in a
file that carries a video stream. Rejecting a larger value keeps raw
video from being mistaken for VMD: a yuv420p10le file starting with
0x2e 0x03 matches the header size field and then scores the same 50 as
the rawvideo demuxer picked by the .yuv extension, which leaves the
format undetermined and fails the whole conversion.

The 26 files with a video stream on samples.ffmpeg.org/game-formats/
sierra-vmd/ all pass; the 9 audio only ones there hold unrelated data
at offset 28 and are exempted by the width and height test.

Fixes: https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/22284

Changed files

libavformat/sierravmd.c

Change #280389

Category	ffmpeg
Changed by	Zuxy Meng <zuxy.mengohnoyoudont@gmail.com>
Changed at	Wed 02 Sep 2026 04:47:55
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3bec079933ca6d5e50e9e307a597b27cab47e288

Comments

checkasm/h264: add weight tests
Also fixed a bug in handling negative weights in x86 10-bit asm
implementation found by the newly added tests.

Weight isn't enabled for neon as it can fail at certain edge cases.

Bi-weight test is only enabled on x86 as other archs cannot pass it
yet.

Signed-off-by: Zuxy Meng <zuxy.meng@gmail.com>

Changed files

libavcodec/x86/h264_weight_10bit.asm
tests/checkasm/h264dsp.c

Change #280411

Category	ffmpeg
Changed by	zhanghongyuan <zhanghongyuanohnoyoudont@uniontech.com>
Changed at	Wed 02 Sep 2026 09:22:39
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	3c517709ffcf5633bea8ee4fae4757b667c46e23

Comments

fftools/cmdutils: fix typo in tile_grid usable_only check
The second condition duplicated the nb_tiles check instead of
verifying that offsets is allocated.

Signed-off-by: zhanghongyuan <zhanghongyuan@uniontech.com>

Changed files

fftools/cmdutils.c

Change #280412

Category	ffmpeg
Changed by	tangsha <tangshaohnoyoudont@kylinos.cn>
Changed at	Wed 02 Sep 2026 09:23:32
Repository	https://git.ffmpeg.org/ffmpeg.git
Project	ffmpeg
Branch	master
Revision	9fc8c785e2747c87121ec28f8f10ceab0562384b

Comments

avutil/encryption_info: reject init info with zero-sized key IDs
av_encryption_init_info_alloc() only allocates key_ids when key_id_size is
non-zero, but stores num_key_ids regardless. If num_key_ids is non-zero and
key_id_size is 0, the returned object has a NULL key_ids array alongside a
non-zero num_key_ids. av_encryption_init_info_free(),
av_encryption_init_info_get_side_data() and
av_encryption_init_info_add_side_data() then loop over num_key_ids and
dereference key_ids[i], leading to a null pointer dereference.

Reject the num_key_ids > 0 && key_id_size == 0 combination instead of
returning an inconsistent object.

Signed-off-by: tangsha <tangsha@kylinos.cn>

Changed files

libavutil/encryption_info.c