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
aaabbb has quit [Changing host]
aaabbb has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 244 seconds]
Anthony_ZO has joined #ffmpeg-devel
twelve has quit [Remote host closed the connection]
abdu71 has joined #ffmpeg-devel
cone-176 has quit [Quit: transmission timeout]
\\Mr_C\\ has quit [Remote host closed the connection]
ahmedhamed has quit [Quit: Connection closed for inactivity]
MyNetAz has quit [Remote host closed the connection]
MyNetAz 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
^Neo has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
cone-107 has joined #ffmpeg-devel
<cone-107>
ffmpeg Andreas Rheinhardt master:5247090b0daa: avcodec/pcm: Remove always-false check
<cone-107>
ffmpeg Andreas Rheinhardt master:e76daf6279a5: avcodec/pcm: Don't allocate LUT when unused
<cone-107>
ffmpeg Andreas Rheinhardt master:485e5388e956: avcodec/pcm: Cache sample_size value
<cone-107>
ffmpeg Andreas Rheinhardt master:f2168f4d869b: avcodec/pcm: Remove duplication from FFCodec define macros
<cone-107>
ffmpeg Andreas Rheinhardt master:c6d743650899: avcodec/pcm: Use dedicated init function for F16LE, F24LE
<cone-107>
ffmpeg Andreas Rheinhardt master:05ab7762101a: avcodec/pcm: Reindent after the previous commit
<cone-107>
ffmpeg Andreas Rheinhardt master:54d582e74951: avcodec/pcm: Avoid AVCodec.sample_fmts for decoders
MyNetAz has quit [Remote host closed the connection]
MyNetAz has joined #ffmpeg-devel
<cone-107>
ffmpeg Andreas Rheinhardt master:0facdafd1e10: avcodec/vlc: Merge VLCElem and RL_VLC_ELEM
<cone-107>
ffmpeg Andreas Rheinhardt master:3869862b0098: avcodec/mpeg12: Avoid temporary stack VLC array during RL VLC init
<cone-107>
ffmpeg Andreas Rheinhardt master:49d15dfc357b: avcodec/rl: Avoid temporary stack VLC array during RL VLC init
<cone-107>
ffmpeg Marvin Scholz master:81ca85a0872b: configure: allow mixed declarations and code for Objective-C
Anthony_ZO has joined #ffmpeg-devel
abdu71 has quit [Quit: Client closed]
minimal has quit [Quit: Leaving]
abdu71 has joined #ffmpeg-devel
jamrial has quit []
<cone-107>
ffmpeg Andreas Rheinhardt master:5ad86d5fbc46: avcodec: Mark init and close functions as av_cold
abdu71 has quit [Ping timeout: 240 seconds]
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 260 seconds]
MyNetAz has quit [Remote host closed the connection]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
MyNetAz has joined #ffmpeg-devel
<fflogger>
[newticket] Bi11: Ticket #11516 ([avformat] avformat/mov: Parse 'name' atom for title metadata) created https://trac.ffmpeg.org/ticket/11516
twelve has joined #ffmpeg-devel
cone-107 has quit [Quit: transmission timeout]
twelve has quit [Ping timeout: 260 seconds]
twelve has joined #ffmpeg-devel
twelve has quit [Ping timeout: 265 seconds]
twelve has joined #ffmpeg-devel
HarshK23 has joined #ffmpeg-devel
twelve has quit [Ping timeout: 265 seconds]
cone-125 has joined #ffmpeg-devel
<cone-125>
ffmpeg Danil Iashchenko master:a1c6ca168370: doc/filters: Shift CUDA-based filters to own section.
<cone-125>
ffmpeg Lynne master:53ff9005b5bd: vulkan: add support for AV_PIX_FMT_GBRAP32
<cone-125>
ffmpeg Lynne master:23eb499b282a: ffv1dec: add support for hwaccels
<cone-125>
ffmpeg Lynne master:caff29dbb18f: FFHWAccel: add buffer_ref argument to start_frame
<cone-125>
ffmpeg Lynne master:6bad55eb17d0: ffv1: add a Vulkan-based decoder
<llyyr>
nice
MyNetAz has quit [Remote host closed the connection]
BradleyS has quit [Quit: quit]
BradleyS has joined #ffmpeg-devel
MyNetAz has joined #ffmpeg-devel
BradleyS has quit [Quit: quit]
BradleyS has joined #ffmpeg-devel
BradleyS has quit [Client Quit]
ngaullier has joined #ffmpeg-devel
BradleyS has joined #ffmpeg-devel
<thardin>
woop
<llyyr>
Lynne: there's a validation error when quitting mpv with hwdec=vulkan, doesn't happen all the time. "vkDestroyImage(): can't be called on VkImage 0xc700000000c7[../video/out/hwdec/hwdec_vulkan.c:316] that is currently in use by VkImageView"
<llyyr>
the last function in the backtrace is vulkan_frame_free in ffmpeg
<llyyr>
there's also just a bunch of validation errors in general when using ffmpeg binary to decode with vulkan
^Neo has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
<Lynne>
llyyr: git master? which codec?
<llyyr>
yes, h264
<Lynne>
I submitted a lot of false positives to the layer
<Lynne>
not our fault, also most drivers don't flag support correctly which causes validation layers to complain when its not even pur fault
<llyyr>
i see, well at least the quit thing seems like a real race condition
derpydoo has quit [Ping timeout: 252 seconds]
<Lynne>
its been there since the start
<Lynne>
me and philipl wondered how its even possible, but IIRC its a false positive
<Lynne>
or something in the mpv interop, since ffmpeg runs fine?
<cone-125>
ffmpeg Gyan Doshi master:9e857e1f8aee: ffbuild: read library linker objects from a file
<haasn>
ramiro: so, my new(er) swscale rewrite v4 is basically ready, except that it's slower even than the previous v3 prototype in all cases involving floats
<haasn>
mostly because I opted not to have "fused" implementations this time around
<haasn>
since I was hoping the improved per-op overhead would make fused implementations less necessary, but the problem mainly comes from the fact that floats do not fit into vector registers
<haasn>
unless I go down to chunk size 8, but then the compiler starts emitting horrible scalar code for reads and conversions
<haasn>
I think we are at a point where indeed it may be best to just rewrite the remaining problematic functions as asm
<haasn>
I think I will try to re-add fused implementations in some cases maybe
Lynne has quit [Read error: Connection reset by peer]
IndecisiveTurtle has joined #ffmpeg-devel
abdu64 has quit [Ping timeout: 240 seconds]
abdu64 has joined #ffmpeg-devel
<cone-125>
ffmpeg Zhao Zhili master:94beaf48723a: avcodec/videotoolboxenc: Add AYUV as a candidate pix_fmt for HEVC alpha
jamrial has joined #ffmpeg-devel
abdu64 has quit [Quit: Client closed]
abdu64 has joined #ffmpeg-devel
IndecisiveTurtle has quit [Quit: IndecisiveTurtle]
michaelni has quit [Quit: Leaving]
michaelni has joined #ffmpeg-devel
twelve has joined #ffmpeg-devel
<fflogger>
[editedticket] JackLau: Ticket #11123 ([ffmpeg] Incorrect first audio segment name when using segment_timeline while creating MPEG-DASH) updated https://trac.ffmpeg.org/ticket/11123#comment:1
abdu64 has quit [Quit: Client closed]
abdu64 has joined #ffmpeg-devel
twelve has quit [Ping timeout: 248 seconds]
ccawley2011 has joined #ffmpeg-devel
Anthony_ZO has quit [Ping timeout: 268 seconds]
odrling has quit [Remote host closed the connection]
odrling has joined #ffmpeg-devel
abdu81 has joined #ffmpeg-devel
abdu64 has quit [Ping timeout: 240 seconds]
<haasn>
does av_cmp_q() have well defined behavior on 1 / 0 ?
Traneptora has quit [Quit: Quit]
Traneptora has joined #ffmpeg-devel
<thardin>
has youtube changed something in their AAC encoder recently?
<thardin>
I'm getting an issue when remuxing AAC to mpegts and then again to matroska, and writing the result to a pipe causing matroskaenc to complain about "Error parsing AAC extradata"
<JEEB>
and the mpegts works?
<JEEB>
because in theory if the mpegts works then all that needs to happen is that the extradata gets generated from the in-band stuff (if matroska requires extradata with parameter-set level of information in it)
<JEEB>
ADTS in mpeg-ts should just be packet-packet-packet so if there is any breakage I'd expect it to break there already
<thardin>
yeah but that only works if the matroska output is seekable
<thardin>
for some reason
<thardin>
see get_aac_sample_rates()
<thardin>
ah I see, it's because it wants to write headers but it can't until it gets its first packet. but that can be solved by delaying writing the header until the first audio packet
abdu64 has joined #ffmpeg-devel
<thardin>
that's messy though
<thardin>
what's extra weird is that it complains about this despite par->sample_rate being set
abdu81 has quit [Ping timeout: 240 seconds]
<haasn>
ugh
<haasn>
we have a royal mess surrounding AVFilterLink.frame_rate
<haasn>
not only do sources inconsistently set it to either 0/0, 0/1 or 1/0 in the absence of fps information
<haasn>
but many filters also explode on any of these three values
<JEEB>
thardin: with H.264 etc there's extract_extradata bsf or so
<haasn>
we also don't correctly propagate it between complex filter graphs
<thardin>
yeah but this is AAC
abdu47 has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
<JEEB>
yea, just noting that we may want something like that as well. otherwise check how movenc is handling it? since it has a delay_moov thingy
<thardin>
so if I understand correctly output_sample_rate is optional
<thardin>
yeah mxfenc does too I think
abdu64 has quit [Ping timeout: 240 seconds]
<thardin>
what I don't get is why this is suddenly a problem. I guess google changed something in youtube, or yt-dlp changed. might be connected to the AV1 deployment
<JEEB>
I don't think format 140 (highest bit rate AAC) has changed any time recently?
<JEEB>
(`yt-dlp -F URL` lists the formats)
<JEEB>
and -f lets you pick one of them when downloading
<thardin>
need to try with current master also. but this is 7.1-ish
<thardin>
here's a oneliner that demonstrates the issue: yt-dlp -f bestvideo+bestaudio/best https://youtube.com/watch?v=eVrYbKBrI7o -o - | ffmpeg -i - -map 0 -codec copy -f matroska - > /dev/null
<cone-125>
ffmpeg Michael Niedermayer master:efdf7512a584: gitattributes: End merge conflicts in Changelog
<cone-125>
ffmpeg Michael Niedermayer master:a8892c54003f: avcodec/ffv1: implement remap for encode/decode_plane()
<cone-125>
ffmpeg Michael Niedermayer master:d7fa6a388b89: avcodec/ffv1: fix remap without chroma planes
<cone-125>
ffmpeg Michael Niedermayer master:b57f71f716c2: avcodec/ffv1: Add YAF16 support
<cone-125>
ffmpeg Michael Niedermayer master:276854e1835e: avcodec/ffv1: Add GRAYF16 support
<cone-125>
ffmpeg Michael Niedermayer master:2471b2202375: avcodec/ffv1dec: Fix a YUVA issue with remaping
<thardin>
compiling master to see if the same issue exists there
<thardin>
yep, same issue in master
ccawley2011 has quit [Ping timeout: 248 seconds]
<thardin>
trac login is really slow
wyatt8740 has quit [Ping timeout: 252 seconds]
wyatt8750 has joined #ffmpeg-devel
<thardin>
oh it happens with AAC in general
<thardin>
maybe it's just that youtube enabled AAC? that they didn't have it before?
<JEEB>
they did
<JEEB>
as I said, format 140 is one of the older ones
wyatt8750 has quit [Read error: Connection reset by peer]
wyatt8740 has joined #ffmpeg-devel
abdu47 has quit [Ping timeout: 240 seconds]
<thardin>
guess I can use --format-sort to avoid picking AAC then, for now
<fflogger>
[newticket] Tjoppen: Ticket #11517 ([avformat] Remuxing AAC in mpegts to non-seekable matroska fails with "Error parsing AAC extradata, unable to determine samplerate") created https://trac.ffmpeg.org/ticket/11517
<thardin>
opus has a similar issue it seems
<toots5446>
michaelni: what's your recommendation to get those FATE samples to an acceptable state? I can changed the hardcoded encoder if you want. What are the reproduction steps for the wine/qemu cross-compilation? Is it possible to get a definite list of requirements so I can work on all of them at once? Thanks
undermine has joined #ffmpeg-devel
wyatt8750 has joined #ffmpeg-devel
wyatt8740 has quit [Ping timeout: 260 seconds]
abdu has joined #ffmpeg-devel
av500 has joined #ffmpeg-devel
wyatt8750 has quit [Ping timeout: 248 seconds]
abdu has quit [Quit: Client closed]
abdu has joined #ffmpeg-devel
wyatt8740 has joined #ffmpeg-devel
twelve has joined #ffmpeg-devel
ccawley2011 has joined #ffmpeg-devel
Marth64 has joined #ffmpeg-devel
psykose has quit [Remote host closed the connection]
psykose has joined #ffmpeg-devel
Lynne has joined #ffmpeg-devel
Lynne has quit [Remote host closed the connection]
Lynne has joined #ffmpeg-devel
twelve has quit [Quit: Quit]
twelve has joined #ffmpeg-devel
abdu76 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
ccawley2011 has quit [Ping timeout: 268 seconds]
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
abdu85 has joined #ffmpeg-devel
abdu76 has quit [Ping timeout: 240 seconds]
<michaelni>
toots5446, bitexact flag should prevent versions to be stored in the output
<michaelni>
and check how other excutables are executed
<toots5446>
kk
abdu85 has quit [Ping timeout: 240 seconds]
twelve has quit [Ping timeout: 252 seconds]
^Neo_ has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 252 seconds]
twelve has joined #ffmpeg-devel
cone-125 has quit [Quit: transmission timeout]
<BBB>
mkver: not sure if you did anything, but gmail (for me) started marking your messages as spam again
<mkver>
I didn't do anything.
<BBB>
hm... ok. annoying...
<wbs>
BBB: I've noticed the same; a couple of patches from mkver, plus a couple of patches from other authors on other mailing lists have been mislabeled as spam in the last few days
<BBB>
manual lauss, also
<BBB>
seems to be just them, mostly
<toots5446>
michaelni: are you able to provide reproduction steps you are using to run the tests with mingw? I would like to make sure that the fix is confirmed before sending back. I am quite familiar with mingw.
twelve has quit [Remote host closed the connection]
<JEEB>
running on linux and running the tests through wine
wyatt8740 has quit [Ping timeout: 272 seconds]
wyatt8740 has joined #ffmpeg-devel
<toots5446>
I see --target-exec=wine is used
<toots5446>
should work the same on my m4 machine
ngaullier has quit [Remote host closed the connection]
Guest10 has joined #ffmpeg-devel
Guest10 has quit [Client Quit]
<toots5446>
michaelni: I think that I'm also gonna update the mechanism for attaching metadata update to frames to use a AVTree
<toots5446>
Frames can span more than one packets (see with vorbis I belive) so it can't be synchronous and packets may be submitted in non-linear order.
<toots5446>
So storing <PTS, metadata> seems more appropriate. I believe that you raised this concern before.
azureglades has joined #ffmpeg-devel
abdu85 has joined #ffmpeg-devel
azureglades has quit [Client Quit]
MyNetAz has quit [Remote host closed the connection]
<philipl>
And speaking of image views, I'm seeing a stream of validation failures related to the storage format and image view format not being the same (RGBA ordering)
<kasper93>
looks like format is not checked before, which apparently is UB if because it looks like is not know to driver
<Lynne>
philipl: you should send it anyway
<kasper93>
On another note, how acceptable would be libcurl integration in place of ffmpeg's dated http.c? There are multiple reasons to do so, starting of newer http versions support, SOCKS proxy support, compression support and many many more limitations.
<kasper93>
I implemented it for mpv, but I could port to lavf if such addition would be not rejected.
delewis has joined #ffmpeg-devel
delewis_ has quit [Remote host closed the connection]
<JEEB>
I think it was already brought up at some point, so there are at least some people who feel like the minimum viable product HTTP implementation probably is not going to stretch to various additional things.
<Lynne>
definitely okay, I've been wanting that too
<JEEB>
BtbN did earlier note that it's good to have a consensus probe be done since apparently some people could NAK libcurl addition for "we already have it"
<JEEB>
of course there are things specifically that http.c does not have
<Lynne>
you'd get true http2 support with connection reuse, and maybe even http3
MyNetAz has joined #ffmpeg-devel
<JEEB>
and that are unlikely to happen in the minimum viable product http.c like yes, HTTP/2 and HTTP/3 over QUIC
<JEEB>
then the usual things like libcurl init being non-threadsafe were brought up
<JEEB>
and someone noted that "callbacks suck"
<JEEB>
Marth64 did note that they were interested in proto'ing such a thing
<JEEB>
and then apparently thread safety for init got improved in 7.84.0
twelve has quit [Remote host closed the connection]
<BtbN>
you don't need http2 for connection re-use though, even pipelining is a http1.1 thing.
<JEEB>
yea
<BtbN>
http2 specificylly to me seems like an insane mountain of complexity for not really much practical gain
twelve has joined #ffmpeg-devel
<philipl>
Lynne: will do
<kasper93>
BtbN: http2 is just one of the thing that we would get for free, the point is here, that things like #7158 or #5776 are not really getting resolved in current implementation.
<Marth64>
Hello and sorry for being absent for past 2 weeks, have been in a big project and also moving my office
<Marth64>
JEEB: This was the idea of using curl I think? I'm not a HTTP/2 fan myself but thought curl could be a good coverage for other things
<Marth64>
beyond limited HTTP we have now
<Marth64>
I agree kasper93 on the principle that libcurl can provide all this for free
abdu27 has joined #ffmpeg-devel
abdu85 has quit [Ping timeout: 240 seconds]
abdu25 has joined #ffmpeg-devel
abdu33 has joined #ffmpeg-devel
abdu27 has quit [Ping timeout: 240 seconds]
abdu25 has quit [Ping timeout: 240 seconds]
abdu95 has joined #ffmpeg-devel
Mirarora has joined #ffmpeg-devel
abdu33 has quit [Ping timeout: 240 seconds]
abdu95 has quit [Ping timeout: 240 seconds]
abdu95 has joined #ffmpeg-devel
abdu57 has joined #ffmpeg-devel
abdu95 has quit [Ping timeout: 240 seconds]
abdu77 has joined #ffmpeg-devel
abdu97 has joined #ffmpeg-devel
abdu57 has quit [Ping timeout: 240 seconds]
twelve has quit [Ping timeout: 252 seconds]
abdu77 has quit [Ping timeout: 240 seconds]
Guest9 has joined #ffmpeg-devel
Guest9 has quit [Client Quit]
cone-956 has joined #ffmpeg-devel
<cone-956>
ffmpeg Manuel Lauss master:46da5605b273: MAINTAINERS: add Manuel Lauss as sanm.c/smush.c Maintainer
mkver has quit [Quit: Leaving]
abdu62 has joined #ffmpeg-devel
abdu97 has quit [Ping timeout: 240 seconds]
twelve has joined #ffmpeg-devel
Thul has joined #ffmpeg-devel
Thulinma has quit [Read error: Connection reset by peer]
abdu93 has joined #ffmpeg-devel
abdu62 has quit [Ping timeout: 240 seconds]
abdu24 has joined #ffmpeg-devel
abdu75 has joined #ffmpeg-devel
abdu93 has quit [Ping timeout: 240 seconds]
abdu24 has quit [Ping timeout: 240 seconds]
<Traneptora>
do we have any internal functions to convert a hexdump into a binary dump?