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 8.0 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<jamrial>
Gramner: why not in a separate asm file like you did for itxfm 16bpp?
<Gramner>
no particular reason
<jamrial>
it would be better imo. help with build paralelization
<Gramner>
compared to itx the number of lines is much lower i guess
<jamrial>
i think there's still some vp9 asm files that alone take more than an entire library that BBB was going to look into splitting or rewritting sometime
<Gramner>
itx is the worst offender for that iirc
fengdaolong has joined #ffmpeg-devel
ocrete has quit [Read error: Software caused connection abort]
ocrete has joined #ffmpeg-devel
c_14 has quit [Server closed connection]
c_14 has joined #ffmpeg-devel
Kimapr_ has quit [Remote host closed the connection]
Kimapr_ has joined #ffmpeg-devel
skinkie has quit [Server closed connection]
skinkie has joined #ffmpeg-devel
MetaNova has quit [Read error: Software caused connection abort]
MetaNova has joined #ffmpeg-devel
Venemo has quit [Server closed connection]
Venemo has joined #ffmpeg-devel
jamrial has quit []
<Lynne>
haasn: I also like to use mpv and the CLI
<Lynne>
and like kepstin mentioned, whitepoint is not directly the color temperature in kelvin
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 244 seconds]
funkylab_ has quit [Read error: Software caused connection abort]
Chagall has quit [Read error: Software caused connection abort]
<indecisiveturtle>
Lynne: When you get a chance could you give the prores encoder a quick try. Last time your feedback had helped me solve a lot of misc bugs
Kimapr_ has quit [Remote host closed the connection]
Kimapr_ has joined #ffmpeg-devel
<BBB>
do I have approve permissions on forgejo?
<BBB>
and if I do (which I assume I do), why don't I see an "approve" button in the interface?
<wbs>
BBB: if not, tell BtbN to give you the right membership
<wbs>
BBB: the approve button is when you're looking at the diffs, in the top right there's a "Finish review" button, which contains approve/comment/request changes buttons
<BBB>
jamrial: and yes I can look into vp9 itx at some point
<BBB>
it's more a christmas break project I guess
<BBB>
wbs: ah! tnx. i guess I'm used to it being in the front-page interface also
<wbs>
BBB: yeah - this is more like where it's hidden on github
Kimapr_ has quit [Remote host closed the connection]
Kimapr_ has joined #ffmpeg-devel
DauntlessOne4985 has quit [Remote host closed the connection]
<indecisiveturtle>
averne: Your decoder worked at the things I've thrown at it, tried various mps values and using non standard dimentions that need macroblock reduction on edge slices and result was okay on both my vega 8 and gtx 1650
<Gramner>
jamrial: vp9mc.o (all instruction sets incl. avx512) is ~20kB and vp9itxfm.o is ~250kB (with the separate avx512 code being ~16kB). so I don't think compilation speed should be much of an issue for mc
<jamrial>
yeah, it's fine as is
Kimapr_ has quit [Remote host closed the connection]
Kimapr_ has joined #ffmpeg-devel
<BBB>
lol ~30sec to assemble vp9itxfm.o, yikes
<BBB>
sorry earth for what I have done to you over all these years
<Gramner>
just redo the build system to start the compilation of the largest files first :P
linkmauve has joined #ffmpeg-devel
<averne>
indecisiveturtle: Great, thanks! And yeah weird dimensions should be working. It's nice that you've tested on AMD, I've only been working on Intel/NV recently. I'll give your encoder a try tomorrow, too busy today sorry
<indecisiveturtle>
np
Kimapr_ has quit [Remote host closed the connection]
Kimapr_ has joined #ffmpeg-devel
DauntlessOne4985 has joined #ffmpeg-devel
DodoGTA has quit [Quit: DodoGTA]
DodoGTA has joined #ffmpeg-devel
<JEEB>
thank you for all the people for bringing up the behavior etc on that one GA thread. I was afraid I would have to explain all the explicit comments of not being willing to follow community behavior rules as well as all that other spam and vitrol etc after finishing with $dayjob .
<JEEB>
so I'm happy that others still recalled that stuff
DauntlessOne4985 has quit [Remote host closed the connection]
DodoGTA has quit [Quit: DodoGTA]
DauntlessOne4985 has joined #ffmpeg-devel
zsoltiv has joined #ffmpeg-devel
desmond-netint has quit [Ping timeout: 258 seconds]
zsoltiv has quit [Remote host closed the connection]
MetaNova has quit [Remote host closed the connection]
zsoltiv has joined #ffmpeg-devel
MetaNova has joined #ffmpeg-devel
zsoltiv_ has joined #ffmpeg-devel
zsoltiv_ has quit [Read error: Connection reset by peer]
zsoltiv_ has joined #ffmpeg-devel
zsoltiv__ has joined #ffmpeg-devel
zsoltiv has quit [Read error: Connection reset by peer]
zsoltiv__ is now known as zsoltiv
kode54 has quit [Read error: Connection reset by peer]
kode54 has joined #ffmpeg-devel
DodoGTA has joined #ffmpeg-devel
paulk has quit [Ping timeout: 255 seconds]
<fjlogger>
[FFmpeg/FFmpeg] Issue #20362 opened: Tee muxer makes the DASH muxer generated manifests with invalid video codec value for h264/avc1 streams (https://code.ffmpeg.org/FFmpeg/FFmpeg/issues/20362) by pov
paulk has joined #ffmpeg-devel
paulk has joined #ffmpeg-devel
DodoGTA has quit [Quit: DodoGTA]
DodoGTA has joined #ffmpeg-devel
ngaullier has quit [Remote host closed the connection]
DauntlessOne4985 has quit [Remote host closed the connection]
DauntlessOne4985 has joined #ffmpeg-devel
microlappy has joined #ffmpeg-devel
rom1v has quit [Server closed connection]
rom1v has joined #ffmpeg-devel
DauntlessOne4985 has quit [Read error: Connection reset by peer]
microlappy has quit [Remote host closed the connection]
<haasn>
ffmpeg currently likes to buffer a _lot_ of frames even for inputs that are not being used yet (e.g. `ffmpeg -i A -i B -i C -filter_complex [0][1][2] concat`), especially when using frame threading. I was thinking about ways we can improve this to have it buffer less aggressively near the start of a filtergraph; i.e. just pulling a single frame to configure the filter graph and then 'pausing' the input (not feeding it more
<haasn>
packets)
<haasn>
until the dec->filter queue actually runs dry
<haasn>
I saw also that ffmpeg_dec.c unconditionally buffers 8 video frames in between the decoder thread and the filter thread, why 8 here? wouldn't 2 or 3 be more than sufficient?
<JEEB>
that sounds like an oddly specific number
<JEEB>
kind of like the HEVC reference count for short terms (since it then IIRC only lets you use the other 8 for something completely different)
<Compn>
8 frame buffer ought to be enough for everyone...
DauntlessOne4985 has joined #ffmpeg-devel
MisterMinister has joined #ffmpeg-devel
mkver has quit [Ping timeout: 260 seconds]
<ePirat>
BtbN, did you disable the automatic new label removal again?
<BtbN>
The thing is, there is no workflow logging an error or it trying to remove the label. So it must either not run for the event or run for it but not try to remove it.
<BtbN>
I also wonder if I can just scrap the isOrgMember check. I don't think non-member even CAN label, not even their own issues
<ePirat>
yeah
<BtbN>
Not sure if you can or can't label your own issue though
<ePirat>
can you make it always log something when it runs?
<BtbN>
It already does log something whenever it runs and does anything
<BtbN>
so it not logging anything just means it decided to not do anything
<BtbN>
The action might just plain be "labeled"
<ePirat>
also it seems you dont log some errors (410, 404)
<ePirat>
so maybe it somehow triggers these?
<BtbN>
Well yes, cause they mean the issue/pr didn't have the label anymore