ChanServ changed the topic of #ffmpeg to: Welcome to the FFmpeg USER support channel | Development channel: #ffmpeg-devel | Bug reports: https://ffmpeg.org/bugreports.html | Wiki: https://trac.ffmpeg.org/ | This channel is publically logged | FFmpeg 7.1.1 is released
Juest has quit [Ping timeout: 252 seconds]
<johnjaye>
ah it's not in the documentation so i didn't know about it
Cqoicebordel has quit [Quit: Konversation terminated!]
Cqoicebordel has joined #ffmpeg
Cqoicebordel has quit [Client Quit]
wiitux has quit [Quit: Leaving]
jensen1 has joined #ffmpeg
aaabbb has quit [Quit: Free Palestine]
rsx has quit [Quit: rsx]
cantelope has joined #ffmpeg
Kei_N has quit [Ping timeout: 245 seconds]
Kei_N has joined #ffmpeg
Kei_N has quit [Ping timeout: 268 seconds]
Kei_N has joined #ffmpeg
damian101 has joined #ffmpeg
bsFFFFFF has joined #ffmpeg
jensen1 is now known as arvl
lavaball has quit [Remote host closed the connection]
acidbunny has joined #ffmpeg
blb has quit [Quit: brb]
blb has joined #ffmpeg
damian101_ has joined #ffmpeg
damian101 has quit [Ping timeout: 248 seconds]
damian101 has joined #ffmpeg
damian101_ has quit [Ping timeout: 244 seconds]
devinheitmueller has joined #ffmpeg
damian101_ has joined #ffmpeg
damian101 has quit [Ping timeout: 248 seconds]
Imho has quit [Remote host closed the connection]
Imho has joined #ffmpeg
hussein1 has joined #ffmpeg
hussein1 has quit [Client Quit]
lavaball has joined #ffmpeg
DiJuMx has joined #ffmpeg
imho__ has joined #ffmpeg
Imho has quit [Ping timeout: 276 seconds]
ttys000_ has joined #ffmpeg
ttys000 has quit [Ping timeout: 265 seconds]
jmcantrell has joined #ffmpeg
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg
xx has quit [Ping timeout: 244 seconds]
xx has joined #ffmpeg
<dostoyevsky2>
I have this 8h mp4 file where `ffmpeg -i v.mp4 audio.mp3' just stops after 3:30h without any error message
damian101 has joined #ffmpeg
damian101_ has quit [Ping timeout: 265 seconds]
OGU has joined #ffmpeg
damian101 has quit [Ping timeout: 252 seconds]
lavaball has quit [Remote host closed the connection]
linkmauve has quit [Quit: Gateway shutdown]
<snoriman>
When I've got a AVFrame* that holds e.g. a YUV420P or another planar type, is the data to which `data` points a contigous chunk of memory? or is each plane allocated separately?
linkmauve has joined #ffmpeg
Shine_ has quit [Read error: Connection reset by peer]
<devinheitmueller>
They *can* be contiguous, but you can’t assume so, and need to treat them as non-contiguous.
<snoriman>
ok thanks devinheitmueller
iive has joined #ffmpeg
<devinheitmueller>
In all the use cases I’ve see, they are separate allocations.
<snoriman>
Ok then I will make sure to treat each plane as it's own "buffer"
<snoriman>
When I want to copy the `data`, how do I then determine the number of bytes per plane? for example in case I have yuv420p?
relue has quit [Ping timeout: 244 seconds]
<devinheitmueller>
There’s a linesize field specified in bytes, which you would multiply by height.
<snoriman>
but what if the plane height is different?
<devinheitmueller>
For 4/2/0 you would divide by 2.
<snoriman>
Is there a helper that gives me the height of a plane for a specific pixfmt and plane number?
relue has joined #ffmpeg
<devinheitmueller>
Been a while since I looked, so can’t say.
<devinheitmueller>
It is worth noting there are dedicated functions for copying an entire plane.
<jkqxz>
I don't think there is a specific helper, but pixdesc is what you want. See av_pix_fmt_desc_get(), the details you want are in the structure it returns.
<devinheitmueller>
…and yes, ^ is a bit ugly… I was in a rush.
<snoriman>
ah thanks devinheitmueller and jkqxz! I'll have a look at av_pix_fmt_desc_get(), I indeed think that that's the one I need.
imho__ has quit [Ping timeout: 276 seconds]
ttys000 has joined #ffmpeg
ttys000_ has quit [Ping timeout: 248 seconds]
lavaball has joined #ffmpeg
System_Error has quit [Remote host closed the connection]
MarioMey has joined #ffmpeg
System_Error has joined #ffmpeg
OGU has quit [Ping timeout: 248 seconds]
lavaball has quit [Quit: lavaball]
xx has quit [Ping timeout: 244 seconds]
microlappy has joined #ffmpeg
sentriz has quit [Read error: Connection reset by peer]
sentriz has joined #ffmpeg
<BtbN>
The exception is actually CUDA pix fmts
<BtbN>
they _have_ to be one continous buffer
<BtbN>
some formats even with special constraints
hussein1 has joined #ffmpeg
DiJuMx has quit [Ping timeout: 276 seconds]
<devinheitmueller>
BtbN: Yeah, I figured some hwaccel formats probably required contiguous buffers, as I’ve seen that in the embedded space. But I couldn’t name any specifically.
<BtbN>
It's mostly because nvenc takes its input as one singular buffer
<BtbN>
so the frames have to be one, in the right layout
<devinheitmueller>
Yeah, that’s consistent with what I’ve seen with some embedded systems like TI Davinci.
meego has joined #ffmpeg
<meego>
Good morning. I'm optimizing the ebur128 filter (x1.17 w/ SIMD) and adding some tests. The EBUR128 official compliance test suite is about 380MB of WAV files.
<meego>
I'm guessing no ones wants all that merged into fate-suite. What is the better option: only implement a portion of those tests? or convert the files to some lossless codec before submitting them to fate-suite?
<BtbN>
380MB of extra files won't be accepted into fate
<meego>
oh wait, i'm on the user channel
<kepstin>
how big are they when compressed to flac, anyways?
<BtbN>
fate also isn't meant to do full compliance testing
<BtbN>
it's meant to detect regressions in the implementation
Fiji has joined #ffmpeg
<meego>
Mmh I'm not totally sure I understand the nuance. Are some tests happening outside of fate?
<meego>
is there a place for full compliance testing?
<BtbN>
There are compliance test suites, of course
<meego>
the files are 75MB in flac
Fiji_ has joined #ffmpeg
Fiji has quit [Read error: Connection reset by peer]
<BtbN>
Yeah, that's too much for fate
fling has quit [Remote host closed the connection]
fling has joined #ffmpeg
microlappy has quit [Quit: Konversation terminated!]
OGU has joined #ffmpeg
OGU has quit [Changing host]
OGU has joined #ffmpeg
microlappy has joined #ffmpeg
<meego>
so the right course of action would be to put the full test logic in tests/ somewhere, and download the sample files at test runtime. Correct?
MarioMey has quit [Quit: Leaving.]
<BtbN>
Just make an easily useable compliance test suite
<BtbN>
and add some very condenses samples to fate that are enough to catch regressions
<meego>
ok thanks
<meego>
i had missed that FATE was only a subset of tests. Everything makes more sense now.
<BtbN>
it's not a subset of tests
<BtbN>
it's regression tests for ffmpeg
<BtbN>
not compliance tests for a specific codec
wiitux has joined #ffmpeg
dxh has quit [Ping timeout: 252 seconds]
danielxhogan has quit [Ping timeout: 248 seconds]
Everything has joined #ffmpeg
danielxhogan has joined #ffmpeg
user_oreloznog has joined #ffmpeg
<meego>
understood, thanks
bsFFFFFF has quit [Remote host closed the connection]
danielxhogan has quit [Read error: Connection reset by peer]
danielxhogan has joined #ffmpeg
OGU has quit [Read error: Connection reset by peer]