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 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
thilo has quit [Ping timeout: 276 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 264 seconds]
System_Error has joined #ffmpeg-devel
psykose has quit [Remote host closed the connection]
psykose has joined #ffmpeg-devel
mkver has quit [Ping timeout: 244 seconds]
minimal has quit [Quit: Leaving]
cone-880 has joined #ffmpeg-devel
<cone-880>
ffmpeg Michael Niedermayer master:142eb90f07a4: postproc/postprocess_template: fix handling of first row of dering_C
<cone-880>
ffmpeg Michael Niedermayer master:0118f392be91: postproc/postprocess_template: Fix reading uninitialized pixels in dering_C()
<cone-880>
ffmpeg Michael Niedermayer master:716c3986c63b: fate: add stripetest
<cone-880>
ffmpeg Michael Niedermayer master:342869ad7ccb: tests: Add libpostproc blocktest
<cone-880>
ffmpeg Michael Niedermayer master:fb36f170deb2: avcodec/ffv1enc: Eliminate fabs()
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 252 seconds]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
jamrial has quit []
marth64_ has joined #ffmpeg-devel
Marth64[m] has quit [Ping timeout: 268 seconds]
MisterMinister has joined #ffmpeg-devel
Marth64[m] has joined #ffmpeg-devel
marth64_ has quit [Ping timeout: 252 seconds]
LainIwakura has quit [Ping timeout: 240 seconds]
System_Error has quit [Remote host closed the connection]
LainIwakura has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
LainIwakura has quit [Ping timeout: 240 seconds]
cone-880 has quit [Quit: transmission timeout]
abdu has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 264 seconds]
System_Error has joined #ffmpeg-devel
paulk has quit [Read error: Connection reset by peer]
paulk has joined #ffmpeg-devel
paulk has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
<fflogger>
[newticket] joe: Ticket #11558 ([build system] ISO C99 and later do not support implicit function declarations) created https://trac.ffmpeg.org/ticket/11558
Kwiboo- is now known as Kwiboo
abdu has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
abdu has joined #ffmpeg-devel
kepstin has quit [Remote host closed the connection]
kepstin has joined #ffmpeg-devel
kepstin has quit [Remote host closed the connection]
kepstin has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
<frankplow>
I think something might be up with the patchwork account system. If I try to log in, it tells me that my username/password is incorrect (although I believe I have the correct password), then if I try to reset my password I never receive the email
abdu has quit [Ping timeout: 240 seconds]
System_Error has quit [Ping timeout: 264 seconds]
<fflogger>
[editedticket] thebombzen: Ticket #11558 ([build system] ISO C99 and later do not support implicit function declarations) updated https://trac.ffmpeg.org/ticket/11558#comment:2
<cone-804>
ffmpeg Mark Thompson master:de42e06b5b20: lavc: APV metadata bitstream filter
<cone-804>
ffmpeg Mark Thompson master:0c79a091e4dc: lavc: APV codec ID and descriptor
<cone-804>
ffmpeg Mark Thompson master:483cadf8d77d: lavc: APV decoder
<cone-804>
ffmpeg Mark Thompson master:821717c3fe1c: lavc/cbs: APV support
<cone-804>
ffmpeg Mark Thompson master:324330a11e15: lavf: APV demuxer
<cone-804>
ffmpeg Mark Thompson master:d03c99441dc5: lavc/apv: AVX2 transquant for x86-64
<cone-804>
ffmpeg Mark Thompson master:b511c767d059: lavf: APV muxer
<cone-804>
ffmpeg Mark Thompson master:8ea11dcd846f: Changelog, doc: Add entries for new APV features
<jkqxz>
On topic for that GA discussion, I guess some small samples for APV would be a good idea.
<jkqxz>
Do we have favourite content for that sort of purpose?
<beastd>
jkqxz: good point. unfortunately not aware of anything specific. looking around what others with existing tests used could be helpful. maybe others here that recently added formats have better advice
<JEEB>
the virtualdub color cube used to be a thing, it's utilized in the utvideo encoder tests for example
<beastd>
IIRC also sometimes our lavfi test sources were leveraged for generating short video/audio content
<jamrial>
yeah, but there's no native encoder
<beastd>
and cutting down existing files (where it is possible) was also done
<JEEB>
(the virtualdub stuff was originally encoded with virtualdub I think, and it's a lossless clip that can then be utilized as input for encoding something else with an external or internal encoder)
<jkqxz>
I'm making things with the reference encoder. Ideally for this it would be small 4:2:2 10-bit natural content.
<JEEB>
for natural content the classics are the SVT samples from 2006
<jkqxz>
I doubt it. That's a once-per-tile calculation.
<jamrial>
per tile per plane, but yes, figured it would not make much of a difference
<fflogger>
[editedticket] BlueWindy: Ticket #9996 ([ffmpeg] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:13
<jkqxz>
Does it show up in a profile?
<jamrial>
haven't tried
ccawley2011_ has joined #ffmpeg-devel
<jkqxz>
For me it's already unrolled completely to 3 instructions per entry.
<jkqxz>
Perf got 2 samples which landed in it out of 10904 for a medium-complexity stream.
<jkqxz>
So maybe 0.02%?
<jkqxz>
Hmm. In that same function there are a lot of samples hitting division instructions, presumably sub_h/sub_w. Why hasn't it noticed they can be shifts?
<jamrial>
may be worth changing sub_w and sub_h to either 0 or 1, and then explicitly use shifts
<jkqxz>
Yeah.
<jamrial>
and probably because it can't know that pix_fmt_desc->log2_chroma_{w,h} are either 0 or 1
<jamrial>
at compile time
<jkqxz>
A nice swedish waterfall: <https://0x0.st/8WNW.apv>. Two tiles (320x180), varied frequencies, some metadata, 120KB. Sensible?
<jamrial>
sure
<jamrial>
better than the 8mb 4k testsuite samples :p
<jamrial>
is this hdr footage? what's the source?
ccawley2011_ has quit [Ping timeout: 265 seconds]
<jkqxz>
SVT open content.
<jkqxz>
And yes, the colour description is missing.
<jkqxz>
Luckily someone has already written a bsf to fix that easily: <https://0x0.st/8WNx.apv>.
<jamrial>
jkqxz: i don't see it in trace_headers
<jkqxz>
The second file is fixed. It's 2020/PQ (9/16/9).
<jamrial>
oh, i was looking for mdcv metadata
<jkqxz>
Playing it in mpv the difference is obvious. The one without the colour description is very faded.
<jamrial>
yes
ccawley2011_ has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
witchymary has quit [Remote host closed the connection]
witchymary has joined #ffmpeg-devel
ccawley2011_ has quit [Ping timeout: 252 seconds]
cone-804 has quit [Quit: transmission timeout]
<jamrial>
jkqxz: color info in the frame header is weirdly aligned. wonder why they didn't use a few of the reserved bits to align it...
abdu has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
abdu has joined #ffmpeg-devel
<mkver>
jkqxz: What guarantees that there are not multiple APV_PBU_PRIMARY_FRAME in a packet?
<jkqxz>
qq(An AU consists of one "primary frame", zero or more "non-primary frame"s, zero or more "alpha frame"s, zero or more "depth frame"s, zero or more "preview frame"s, zero or more "metadata"s, and zero of more "filler"s.)
<jkqxz>
Returning one of them doesn't seem unreasonable in this case. Would you like a warning to go with it?
<mkver>
I mean: What guarantees it in your code, not in the spec.
<jkqxz>
Nothing guarantees it here. The decoder will return the last one and nothing else cares.
<mkver>
If you call apv_decode() when you already have a frame, you should reach the "pic->data[*]!=NULL in get_buffer_internal\n" error message, which is good enough for me.
ccawley2011_ has joined #ffmpeg-devel
LainIwakura has joined #ffmpeg-devel
<another|>
<@jkqxz> Is there a 10-bit foreman? He would work.
<another|>
Are you implying that 8-bit foreman is lazy?
<jkqxz>
I have never seen him doing any work. He just stands around.