michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
vipyne has joined #ffmpeg-devel
<jamrial>
michaelni: can you tell me the hash of the latest commit you fetched from my branch that you tested?
iive has quit [Quit: They came for me...]
<JEEB>
BtbN: good job with the trac work
<michaelni>
jamrial, should have been 4e45c3dac2a91b220a14ae8a277cd0d2d860c5f6
<jamrial>
michaelni: ok, cool
<JEEB>
btw re ambisonics, should we plug that into swresample or aresample so if the api user wants normal layout, we give that?
<fflogger>
[editedticket] MasterQuestionable: Ticket #11281 ([ffmpeg] ffmpeg create waveform overview with short audiofiles doesn't work) updated https://trac.ffmpeg.org/ticket/11281
<jamrial>
JEEB: there's a filter for that
<JEEB>
yea, but what I mean is that API user already has code to set XYZ channel layout via either swresample or the avfilter aresample filter, it could be made transparent. unless that is already there :)
vipyne has quit [Quit: Leaving.]
<JEEB>
but yea, will have to look at the patch set how much of that is already there.
thilo has quit [Ping timeout: 260 seconds]
thilo has joined #ffmpeg-devel
vipyne has joined #ffmpeg-devel
\\Mr_C\\ has quit [Remote host closed the connection]
<JEEB>
idea from VDD: SIMD tutorial/learning repository under github.com/FFmpeg
cone-784 has joined #ffmpeg-devel
<cone-784>
ffmpeg Brad Smith master:d6b2d08fc728: libavutil/ppc: Include the hardware feature flags like the other archs
<JEEB>
729
<JEEB>
whoops
<JEEB>
elenril: did the MV-HEVC work essentially already enable feeding a second frame to a possibly independent decoder state?
<elenril>
explain further.jpg
<JEEB>
separate nuh_layer_id alpha is what my context for my question is
<JEEB>
IIRC the alpha was completely separate independent stream, so I wonder if those parameter sets and frames could already be fed to the existing logic built for MV-HEVC
<elenril>
not verbatim, but it should not be very hard
<elenril>
...I would think
<JEEB>
ok
<JEEB>
and I wonder if the overall code structure could be reused for similar stuff for VPx. AV1 is in dav1d so if AV1 alpha is similarly specified that would end up there
<JEEB>
since my WIP branch for VP8 created a separate codec ctx and synchronizing that seemed to be a PITA with frame threading at least then
keith has quit [Remote host closed the connection]
keith_ is now known as keith
tufei__ has quit [Remote host closed the connection]
tufei__ has joined #ffmpeg-devel
welder has quit [Quit: WeeChat 3.8]
<fflogger>
[editedticket] osamu620: Ticket #10123 ([avcodec] j2k decoder fails on on conformance codestream p0_09.j2k when bitexact is used) updated https://trac.ffmpeg.org/ticket/10123
<JEEB>
tried to look up the location where avformat/mov.c reads track timescale and sets the AVStream timebase to look where it might be doing the field rate mishap , but couldn't quickly notice it
mkver has quit [Ping timeout: 260 seconds]
Krowl has joined #ffmpeg-devel
keith has quit [Remote host closed the connection]
keith has joined #ffmpeg-devel
Workl has joined #ffmpeg-devel
Workl2 has joined #ffmpeg-devel
<kurosu>
MMX & up: also valid for 3dnow/+/ext (may have written some of it)
Krowl has quit [Ping timeout: 252 seconds]
Workl has quit [Ping timeout: 260 seconds]
Krowl has joined #ffmpeg-devel
Workl2 has quit [Ping timeout: 252 seconds]
Workl has joined #ffmpeg-devel
Krowl has quit [Ping timeout: 265 seconds]
Workl has quit [Read error: Connection reset by peer]
MrZeus has quit [Read error: Connection reset by peer]
microchip_ has joined #ffmpeg-devel
putacho has quit [Ping timeout: 244 seconds]
Workl has quit [Read error: Connection reset by peer]
<jamrial>
michaelni: can you fetch the latest commit from my branch and test fate-pixfmt only?
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
microchip_ has quit [Read error: Connection reset by peer]
microchip_ has joined #ffmpeg-devel
microchip_ has quit [Client Quit]
microchip_ has joined #ffmpeg-devel
ccawley2011 has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
<beastd>
BtbN: Thanks for upgrading Trac!
<michaelni>
jamrial, on x86-32 failed fate-pixfmt-yuv444-gray10be fate-pixfmt-yuv444-gray10le fate-pixfmt-gbr-yuv420p10le and several others but if i re-run them they pass, if i run everything then i see failures again
<jamrial>
well, that's problem
<jamrial>
i didn't touch those paths
<jamrial>
cd
<michaelni>
iam confused by this too, x86-64 did pass
<beastd>
For MPlayer I have from the beginning reconfigured "Trac -> View tickets" to directly invoke a custom query instead of showing available reports because I find the extra indirection annoying and harmful to UX
<beastd>
The custom query selects all tickets not closed and orders by modified descending
<jamrial>
michaelni: i see -sws_flags +accurate_rnd+bitexact is added, so it's not something like non bitexact mmx code at play
<jamrial>
guess it should be run under ubsan/asan then
<michaelni>
i tried fate-pixfmt-yuv444-xv36le its clean under valgrind, iam missing something here
<BtbN>
Did your Intel 14th Gen CPU errode away some bits? :D
<michaelni>
aha
<fflogger>
[editedticket] llyyr: Ticket #11280 ([undetermined] Strange green flickering line added with a hevc_vaapi reencode) updated https://trac.ffmpeg.org/ticket/11280
<michaelni>
tests/ref/fate/pixfmt-yuv444-xv36le:15206400 tests/data/pixfmt/xv36le.yuv and tests/ref/fate/pixfmt-gbr-xv36le:15206400 tests/data/pixfmt/xv36le.yuv
<michaelni>
can someone show me the difference in the pathes ?
<michaelni>
anyway i have to leave, i can do more checking in 2h or so
<jamrial>
there should be none. i don't see an unscaled path for those
<michaelni>
are you sure that 2 tests dont use the same path ? so if they run at the same time with make -j32 they race
<jamrial>
oh, i see what you mean
blb has quit [Remote host closed the connection]
blb has joined #ffmpeg-devel
<jamrial>
michaelni: force pushed a new version, thanks
<fflogger>
[editedticket] Disjt: Ticket #11281 ([avfilter] "showwavespic" filter failed with very short audio (< 200 ms)?) updated https://trac.ffmpeg.org/ticket/11281
<fflogger>
[editedticket] Disjt: Ticket #11281 ([avfilter] "showwavespic" filter failed with very short audio (< 200 ms)?) updated https://trac.ffmpeg.org/ticket/11281
Workl has quit [Read error: Connection reset by peer]
elvis_a_presley has quit [Quit: smoke-bomb ; grapple-hook]
elvis_a_presley has joined #ffmpeg-devel
<fflogger>
[editedticket] Disjt: Ticket #11281 ([avfilter] "showwavespic" filter failed with very short audio (< 200 ms)?) updated https://trac.ffmpeg.org/ticket/11281
Traneptora has joined #ffmpeg-devel
<fflogger>
[editedticket] bernd_b: Ticket #11280 ([undetermined] Strange green flickering line added with a hevc_vaapi reencode) updated https://trac.ffmpeg.org/ticket/11280
<jamrial>
michaelni: looks like there's some x86 assembly that's faulty. if you compile with --disable-asm (or use cpuflags 0 i assume) it will fail on x86 too
<fflogger>
[editedticket] Disjt: Ticket #11281 ([avfilter] "showwavespic" filter failed with very short audio (< 200 ms)?) updated https://trac.ffmpeg.org/ticket/11281
<jamrial>
force pushed a version without those tests
MetaNova has joined #ffmpeg-devel
<jamrial>
the difference is visually indistinguishable with and without x86 asm, so it could be rounding somewhere or something
<jamrial>
funny how once you start adding a lot of swscale conversion tests, bugs surface
___nick___ has quit [Ping timeout: 252 seconds]
___nick___ has joined #ffmpeg-devel
<michaelni>
jamrial, latest version passes x86-32 / 64 mips and arm (didnt try mingw)
<jamrial>
ok, thanks
<michaelni>
np, happy to be able to help improve swscale :)
<jamrial>
now we need to find what asm function is at fault for the above tests :p
<jamrial>
and either fix it, or make sure it's not used if bitexactness is requested
<cone-519>
ffmpeg James Almer master:a686d34fea6e: swscale/swscale_unscaled: add unscaled conversion for AYUV/VUYA/UYVA
<cone-519>
ffmpeg James Almer master:78ba06928a69: swscale/x86/rgb2rgb: add optimized versions of the remaining shuffle_bytes functions
<cone-519>
ffmpeg James Almer master:165f1241299c: vsrc_testsrc: add support for x2rgb10le to rgbtestsrc
cone-519 has joined #ffmpeg-devel
<cone-519>
ffmpeg James Almer master:430bec86665f: fate/filter-video: add more rgbtestsrc tests
<cone-519>
ffmpeg James Almer master:febc9e8162af: swscale/output: add full chroma interpolation support for x2rgb10
<cone-519>
ffmpeg James Almer master:5ccc3f0fca1f: swscale/swscale_unscaled: add unscaled hbd planar RGB to x2rgb10le
<cone-519>
ffmpeg James Almer master:c029a2f7ddea: swscale/swscale_unscaled: add unscaled rgb to planar rgba
<cone-519>
ffmpeg James Almer master:5f5421ec66f5: swscale/swscale: prevent integer overflow in chrRangeToJpeg16_c
<cone-519>
ffmpeg James Almer master:40878dfb11ea: fate/pixfmts: test conversion of high bitdepth formats
gnafu has quit [Quit: "Rebooting for kernel update."]
<cone-519>
ffmpeg Sean McGovern master:c325f9c619b3: libavutil/ppc: defines involving bit shifts should be unsigned
___nick___ has quit [Ping timeout: 255 seconds]
MrZeus has joined #ffmpeg-devel
<fflogger>
[newticket] Disjt: Ticket #11283 ([undetermined] ffmpeg aloop audio filter not working correctly in 48khz stereo wav) created https://trac.ffmpeg.org/ticket/11283
System_Error has quit [Remote host closed the connection]
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 252 seconds]
System_Error has joined #ffmpeg-devel
cone-519 has quit [Quit: transmission timeout]
<any1>
I really appreciate doc/APIchanges. It's very helpful and surprisingly well managed.
<pal>
thardin et al.: do we care that the patch that fixes J2K encoder quantization also changes the compression ration achieved for a specific value of the (arbitrary) Q parameter?
<pal>
... this does not break any standard/recommendations but might be surprising to anyone who had assumed that mechanics underlying the Q parameter for the J2K encoder would never change.