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 7.0 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
thilo has quit [Ping timeout: 260 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
lexano has quit [Ping timeout: 264 seconds]
<compnn>
j-b, oh i dont mean to play games, i mean to troll. 'cant use ffmpeg due to software patents' 'cant use ffmpeg because vendor licenses, apple license' 'ffmpeg went to mars and (your software) didnt"
<compnn>
"if you're asking if you need ffmpeg, you dont need it"
<compnn>
whats with the requests to use ffmpeg on github / whatever ? someone should do that unofficially and just merge with git at periods. the whole reason why we arent hosted in usa is because software patents though. and microsoft runs github...
<compnn>
pretty sure some libraries were nuked off sourceforge for this very reason
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
lemourin has joined #ffmpeg-devel
arch1t3cht6 has joined #ffmpeg-devel
arch1t3cht has quit [Ping timeout: 264 seconds]
arch1t3cht6 is now known as arch1t3cht
<rcombs>
> As a broad guideline, divide work into as many discrete tasks as possible, at least 3x the number of cores and preferably more.
<rcombs>
^ new advice from apple on how best to split up tasks when on systems with heterogeneous cores; their recommendation is in line with my own performance testing
<rcombs>
like, have n_cores threads, but split a frame into n_cores*3 slices instead of n_cores slices, and assign slices to threads dynamically
System_Error has quit [Remote host closed the connection]
jamrial has quit []
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 260 seconds]
AbleBacon has quit [Read error: Connection reset by peer]
kekePower has quit [Remote host closed the connection]
kekePower has joined #ffmpeg-devel
okx has quit [Ping timeout: 260 seconds]
okx has joined #ffmpeg-devel
another| has quit [Ping timeout: 260 seconds]
another| has joined #ffmpeg-devel
kekePower has quit [Remote host closed the connection]
kekePower has joined #ffmpeg-devel
Wenbin_Chen__ has quit [Remote host closed the connection]
<cone-827>
ffmpeg Haihao Xiang master:3172a6722e94: lavfi/tonemap_vaapi: By default use bt709 for output frame
<cone-827>
ffmpeg Haihao Xiang master:db22804145a4: lavfi/tonemap_vaapi: Update the log
<cone-827>
ffmpeg Xinpeng Sun master:35ae44c61562: lavfi/tonemap_vaapi: Add support for HDR to HDR tone mapping
System_Error has joined #ffmpeg-devel
darkapex has quit [Ping timeout: 272 seconds]
darkapex has joined #ffmpeg-devel
kurosu has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
ramiro has quit [Ping timeout: 256 seconds]
ramiro has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<another|>
Lynne: I hear it's used a bit in Asia
Krowl has quit [Read error: Connection reset by peer]
<thardin>
why is srt:// such a pain?
<another|>
It might take off a bit once TV starts to use the newer standards
<JEEB>
broadcasters are basically the only major area which is investing into VVC
<JEEB>
"ooh shiny multi-layer coding" etc is what I saw during last year's NHK R&D
<JEEB>
+open days
<thardin>
vvc combines oodles of techniques into one codec, right? I read a little bit about it. the spec is pretty huge
<JEEB>
it's basically iterative development in the tree of AVC -> HEVC -> VVC
<JEEB>
you add a couple more motion vector possibilities, you make the block sizes possibly larger & coding tree unit more nimble etc
<JEEB>
and then I think they had some attempt at CfL
<thardin>
CfL is just KLT but not as good, right?
<frankplow>
VVC calls CfL cross-component. There's the CCLM which I believe is similar to AV1's then there's also cross-component ALF which I'm not too familiar with.
Krowl has joined #ffmpeg-devel
<thardin>
YUV itself is of course an attempt at decorrelation
<Lynne>
av1's cfl uses a single parameter (alpha in the spec) to fit the luma pixels to chroma via scaling
<Lynne>
only for ACs, DC is predicted as-is via chroma
<Lynne>
it's simple, doesn't cost the hardware much to do, and it gave about a percent BDrate improvement IIRC
<Lynne>
comparing it with KLT is like comparing a P90 with a slingshot
<thardin>
I still get the feeling that's where we're headed
<Lynne>
yeah, I absolutely think we should get rid of YUV entirely and just rely on the codec to do cross-correlation
<Lynne>
that's why I like JXL integrating colorspace in the codec itself
<Lynne>
as for the time domain, it would be cool if cameras no longer shot at 180 degree shutter but something much lower to reduce motion blur
<frankplow>
YUV is a perceptual thing as well as just decorrelation though
<Lynne>
then we could do proper motion field detection and maybe render framerate a decoder-time setting
<Lynne>
lolno, it's not perceptual until you subsample it
<frankplow>
Some distinction between a luma component and chroma component is needed for subsampling
<frankplow>
Yeah
<thardin>
subsampling should be done by the codec, not by the pixel format
<thardin>
it's a holdover from analog broadcast
<thardin>
another thing to explore would be whole-GOP transforms
<Lynne>
I just struck both out from what an optimal codec would have, given infinite compute
<thardin>
many years ago I experimented with 3D RLE for flat color essence (cartoons mostly). it was surprisingly effective
mkver has joined #ffmpeg-devel
cone-827 has quit [Quit: transmission timeout]
razor_ has joined #ffmpeg-devel
<wellsakus>
I've just uploaded the third patchset to evc. The earlier version v2 was uploaded by mistake and the latest changes are in v3
Livio has joined #ffmpeg-devel
lexano has joined #ffmpeg-devel
mkver has quit [Remote host closed the connection]
mkver has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Livio has quit [Ping timeout: 268 seconds]
<Lynne>
how many allocs do we generally do per a single avcodec encode in frame+out packet?
<Sean_McG>
is there a way to get that number from a profiling tool?
<Sean_McG>
also, hi/good $TIME_OF_DAY
<Lynne>
I think mkver did tests, because of the refstruct refactor which saved a ton
Krowl has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
<mkver>
Lynne: valgrind tells you this.
<mkver>
(Due to the threading stuff, the number is unfortunately random nowadays.)
<Lynne>
valgrind would, but we do a lot of other allocations on init, and I'm interested in a single non-threaded encode operation
<Lynne>
or decode, either would work
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
cone-038 has joined #ffmpeg-devel
<cone-038>
ffmpeg Andreas Rheinhardt master:94132dc4aa4d: avcodec/ac3enc: Avoid allocation for mdct_window
<cone-038>
ffmpeg Andreas Rheinhardt master:79d1814b719a: avcodec/ac3enc: Avoid indirections, allocations of small arrays
<cone-038>
ffmpeg Andreas Rheinhardt master:3b93b1af139d: avcodec/ac3enc: Avoid allocation for windowed_samples
<cone-038>
ffmpeg Andreas Rheinhardt master:da460fb95cba: avcodec/ac3enc: Combine cpl_coord buffers
<cone-038>
ffmpeg Andreas Rheinhardt master:5b9fab0f077d: avcodec/ac3enc: Combine loops
<cone-038>
ffmpeg Andreas Rheinhardt master:7f35c999f61b: avcodec/ac3enc: Avoid copying samples
<Dmitri_Ovch>
Hi, when I transcode a video in which the resolution changes using this command
<Dmitri_Ovch>
I get an error reinitializing the filter graph. even if there is a filter that works with this format.
<Dmitri_Ovch>
does the automatic insertion of the scale filter work only with the host memory,
<Dmitri_Ovch>
or is it possible to specify that ffmpeg need to use this filter in this case?
Krowl has joined #ffmpeg-devel
<JEEB>
the auto-scaler insertion logic is only pointing at the scale vf
<JEEB>
until that logic gets extended, you need to basically explicitly add the relevant scale filter and possibly even `-noauto_conversion_filters` so that it wouldn't even attempt to insert something.
<Dmitri_Ovch>
Thanks
<JEEB>
I had some ongoing work making vf_libplacebo work with d3d11 surfaces, but it's really annoying due to the default d3d11 flags not letting you sample the image :P
elenril has left #ffmpeg-devel [WeeChat 3.8]
jarthur has joined #ffmpeg-devel
mkver has quit [Ping timeout: 252 seconds]
<haasn>
pain
mkver has joined #ffmpeg-devel
Livio has joined #ffmpeg-devel
<cone-038>
ffmpeg Andrew Sayers master:6927457443e6: avutil/frame: remove comment about avcodec_get_frame_class()
kierank has left #ffmpeg-devel [#ffmpeg-devel]
Krowl has quit [Read error: Connection reset by peer]
tufei_ has joined #ffmpeg-devel
tufei has quit [Remote host closed the connection]
Livio has quit [Ping timeout: 260 seconds]
<cone-038>
ffmpeg James Almer master:5b9db32ccc94: avformat/mov: ignore old infe box versions
<cone-038>
ffmpeg James Almer release/7.0:ed55219eddc9: avformat/mov: ignore old infe box versions
razor_ has quit [Ping timeout: 260 seconds]
Raz- has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
chaosengage has joined #ffmpeg-devel
iive has joined #ffmpeg-devel
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 268 seconds]
j45_ is now known as j45
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 268 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
Workl has joined #ffmpeg-devel
Krowl has quit [Ping timeout: 246 seconds]
kurosu has quit [Quit: Connection closed for inactivity]
chaosengage has quit [Quit: Leaving]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
Workl has quit [Read error: Connection reset by peer]
AbleBacon has joined #ffmpeg-devel
cone-038 has quit [Quit: transmission timeout]
Krowl has joined #ffmpeg-devel
compnn has quit [Read error: Connection reset by peer]
compnn has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
Raz- has quit [Ping timeout: 255 seconds]
___nick___ has joined #ffmpeg-devel
Raz- has joined #ffmpeg-devel
Raz- has quit [Ping timeout: 252 seconds]
___nick___ has quit [Ping timeout: 264 seconds]
Krowl has quit [Read error: Connection reset by peer]
mkver has quit [Ping timeout: 256 seconds]
mkver has joined #ffmpeg-devel
cone-076 has joined #ffmpeg-devel
<cone-076>
ffmpeg Frank Plowman master:bb9e4ff35568: lavc/vvc: Skip enhancement layer NAL units
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 256 seconds]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
Teukka has quit [Read error: Connection reset by peer]