TheSilentLink has quit [Quit: Good Bye! My bouncer has probably crashed or lost connection to the internet...]
vlm has quit [Quit: Leaving]
mven91 has joined #ffmpeg
mven9 has quit [Ping timeout: 252 seconds]
mven91 is now known as mven9
RhineDevil has joined #ffmpeg
RhineDevil has quit [Client Quit]
gvg_ has quit [Ping timeout: 252 seconds]
gvg_ has joined #ffmpeg
emmanuelux_ has joined #ffmpeg
iive has quit [Quit: They came for me...]
emmanuelux has quit [Ping timeout: 264 seconds]
Estrella has joined #ffmpeg
Vonter has quit [Ping timeout: 256 seconds]
Vonter has joined #ffmpeg
Estrella has quit [Ping timeout: 256 seconds]
gvg has quit [Ping timeout: 252 seconds]
gvg_ has quit [Ping timeout: 246 seconds]
hbbs has quit [Quit: bye]
lucasta has quit [Quit: Leaving]
hbbs has joined #ffmpeg
fossdd has quit [Read error: Connection reset by peer]
thilo has quit [Ping timeout: 276 seconds]
hussein1 has quit [Ping timeout: 240 seconds]
minimal has quit [Quit: Leaving]
fossdd has joined #ffmpeg
thilo has joined #ffmpeg
hussein1 has joined #ffmpeg
FH_thecat has quit [Ping timeout: 245 seconds]
lolok has quit [Remote host closed the connection]
sagax has joined #ffmpeg
javabean has quit [Ping timeout: 246 seconds]
gchound has quit [Quit: WeeChat 4.1.2]
javabean has joined #ffmpeg
lemourin has quit [Read error: Connection reset by peer]
lemourin has joined #ffmpeg
javabean has quit [Ping timeout: 255 seconds]
javabean has joined #ffmpeg
wsc3 has joined #ffmpeg
wsc3 is now known as gchound
javabean has quit [Ping timeout: 268 seconds]
navi has quit [Quit: WeeChat 4.0.4]
squeaktoy has quit [Ping timeout: 260 seconds]
squeaktoy has joined #ffmpeg
gvg has joined #ffmpeg
javabean has joined #ffmpeg
gvg_ has joined #ffmpeg
hussein1 has quit [Remote host closed the connection]
hussein1 has joined #ffmpeg
Dotz0cat has quit [Ping timeout: 268 seconds]
gchound has quit [Quit: WeeChat 4.1.2]
sagax has quit [Remote host closed the connection]
jagannatharjun has joined #ffmpeg
marcj has quit [Read error: Connection reset by peer]
marcj has joined #ffmpeg
javabean has quit [Ping timeout: 264 seconds]
javabean has joined #ffmpeg
troyt has quit [Quit: AAAGH! IT BURNS!]
theobjectivedad has quit [Remote host closed the connection]
\\Mr_C\\ has joined #ffmpeg
theobjectivedad has joined #ffmpeg
bitoff has joined #ffmpeg
bitoff has quit [Remote host closed the connection]
bitoff has joined #ffmpeg
troyt has joined #ffmpeg
bitoff has quit [Remote host closed the connection]
bitoff has joined #ffmpeg
bitoff has joined #ffmpeg
upekkha has quit []
upekkha has joined #ffmpeg
hpkn has quit [Remote host closed the connection]
Dotz0cat has joined #ffmpeg
bitoff has quit [Ping timeout: 246 seconds]
FH_thecat has joined #ffmpeg
mosasaur has joined #ffmpeg
sagax has joined #ffmpeg
jagannatharjun has quit [Quit: Connection closed for inactivity]
bitoff has joined #ffmpeg
bitoff has quit [Ping timeout: 255 seconds]
lusciouslover has quit [Ping timeout: 245 seconds]
lusciouslover has joined #ffmpeg
AbleBacon has quit [Read error: Connection reset by peer]
waleee has joined #ffmpeg
<mosasaur>
seems I've been misinformed, there's handbrake as the ultimate ffmpeg gui, however it's probably that's not what I was asking for, a *simple* gui. But there's also hope, maybe someone will build something with pysimplegui, and maybe that will be me, if I find the time and energy
<CounterPillow>
or you could stop being a scrub and learn
<JEEB>
it's not an "ffmpeg GUI" (meaning a GUI for the command line app), it's a GUI application that utilizes FFmpeg's APIs underneath. but as such, it seems to be a pretty alright encoding tool
<JEEB>
the ffmpeg CLI tries to do way too many things, so trying to pack all that into a consise and nice GUI isn't going to be good, so you'd probably end up with N different GUIs that all attempt to handle one part of what in theory can be done with the CLI.
<JEEB>
this is already visible with the CLI as well, so many things almost work - or work until you hit one of those corner cases that the CLI's architecture doesn't handle.
<JEEB>
so having the answer be multiple tools built for specific use cases via the API sounds like the theoretical good point.
lusciouslover has quit [Ping timeout: 245 seconds]
sagax has quit [Excess Flood]
Ogobaga has quit [Quit: Konversation terminated!]
Ogobaga has joined #ffmpeg
<mosasaur>
I have been working with ffmpeg-python, or was it python-ffmpeg, idk, it was some time ago, but what I got from that was that it should be possible to write stuff in a form that look more streamlined, even if it's just because of the significant whitespace. So you end up with code that is more readable, which would be good if the one reading it would be you, a few years in the future. Unfortunately my tasks seldom repeat, I always have to change the
<mosasaur>
names of source and output directories, filenames, formats and what not. But if it was approached like some collective system, where everyone could post their recipes, and have them ranked according to peer review and coding style, maybe it could work on some basic level, after a way is found to address the naming problem
<mosasaur>
maybe a wikipedia style with templates?
YuGiOhJCJ has quit [Remote host closed the connection]
kepstin has quit [Remote host closed the connection]
kepstin has joined #ffmpeg
damo22 has joined #ffmpeg
<damo22>
my phone records video in 3840x2160, whats the cheapest way to downsample to 1080p ?
<CounterPillow>
setting your phone's recording resolution to 1080p
<damo22>
i tried -vf scale=1920x1080:flags=neighbor
<AMM>
digging up a computer from the trash and stealing electricity from somewhere
<CounterPillow>
the used scale filter is barely going to matter if you're then feeding it into an encoder
<CounterPillow>
when doing playback, bilinear on the gpu is literally free
<damo22>
i see
<damo22>
im an audio guy i know almost nothing about video
<damo22>
if i was doing it i imagine i could discard every second pixel
<damo22>
but it probably doesnt work like that
lavaball has joined #ffmpeg
rsx has quit [Quit: rsx]
<furq>
it does work like that if you meant three out of every four pixels
<furq>
but there's no reason to actually do that if you're reencoding anyway
<furq>
just use whatever scaler looks best
<damo22>
what is a pixel actually, is it a 4-tuple of RGBA?
<CounterPillow>
depends on the pixel format
<furq>
for video it's probably yuv
<damo22>
oh
<furq>
and probably 4:2:0 so only one u and v pixel for every four y pixels
<damo22>
maybe CounterPillow's idea is best, set my phone to capture in the format i want
rsx has joined #ffmpeg
<furq>
that's usually the best idea
<furq>
the second clause not the first
BillTorvalds has joined #ffmpeg
<damo22>
why does apple refuse to open webm?
<damo22>
that means i need to encode two kinds of video, mp4 and webm
<CounterPillow>
because they're part of the patent pools trying to rent seek off the alternatives iirc
<furq>
webm is google
<furq>
and the people who run everything in our lives are dreadful
<damo22>
indeed
<furq>
you don't need to bother with webm though
<furq>
everything plays h264/aac in mp4
<CounterPillow>
except a default installation of fedora :^)
<furq>
like i said, everything plays h264/aac in mp4
<damo22>
i found that opus is better than mp3 and possibly aac
<furq>
it is but not enough to really care about
<CounterPillow>
yes opus is superior to aac, but it's a few kilobits per second of savings at same quality at best
<damo22>
and if i want lossless audio, i cant embed flac into a mp4
<furq>
it's better than even the best aac encoder but good luck noticing the difference at 128k or above
<damo22>
:(
<furq>
you can't put lossless audio in webm either
<damo22>
are you sure
<furq>
pretty sure it's vorbis or opus only
<damo22>
hmm ok
<damo22>
put wav in mp4?
<furq>
you can put alac in mp4
<furq>
you shouldn't but you can
<damo22>
mine sounds like a horrible idea
<damo22>
alac is apple proprietary again?
<furq>
more or less
<furq>
ffmpeg has an encoder
<furq>
but i assume device support is bad outside of apple
<damo22>
h264 encoding needs a license?
<furq>
in theory
fling has quit [Remote host closed the connection]
<damo22>
lol
<CounterPillow>
in practice MPEG-LA only cares if you're making money
fling has joined #ffmpeg
<CounterPillow>
and even then, they're too boomer to understand most ways people make money
<galad>
I don't thin alac has got any patent, and apple released an open source decoder and encoder many years ago
<damo22>
do we even have a patent unencumbered format that can hold video and lossless audio
<furq>
mkv
<damo22>
i see, but devices wont support it?
<galad>
mkv is a container, then it depends on what you put inside
<galad>
anyway, the last h.264 patents will expire in a couple of years
<damo22>
so i assume most people compromise and just use mp4?
<galad>
mp4 is another container
<galad>
it's so old, I don't think anyone claimed patents on it
<furq>
if you want browsers to play it then your choices are h264/aac in mp4 or vp9/opus in webm
<galad>
it's based on mov, that's almost 30 years old
<damo22>
right, i want browsers to play it
TheSilentLink has joined #ffmpeg
<damo22>
but safari wont play webm
<galad>
and there is no royalty on distributing h.264 on the web
<furq>
it's not really a question of what the container supports in that case
<galad>
so it's safe to just use mp4 with h264 and aac on your website
<furq>
most browsers won't play mp3 in mp4 even though they play them both separately
<damo22>
ok
fling_ has joined #ffmpeg
Ogobaga has quit [Quit: Konversation terminated!]
<damo22>
transcoding video from mp4 to webm is a computationally expensive exercise, i bet google made hardware ASICs to do it
Ogobaga has joined #ffmpeg
<furq>
they did
<galad>
yes, and the quality of youtube video is really poor
fling has quit [Ping timeout: 240 seconds]
<furq>
it got worse since they made those vp9 hardware encoders
<CounterPillow>
ye
<damo22>
my old thinkpad cant even play the video my phone created
<CounterPillow>
for like a day they also managed to ship broken video streams, that was fun
<damo22>
(in realtime)
<furq>
they at least have a relatively high rate h264 stream for ios now which is nice
<furq>
and by for ios i mean for ios and yt-dlp
billchenchina has joined #ffmpeg
<damo22>
i wouldnt mind a vp9/opus hardware encoder
<damo22>
maybe my phone has one
fling_ is now known as fling
<damo22>
whats the compression ratio, like 20x from mp4 -> webm?
<CounterPillow>
that question makes no sense
<CounterPillow>
mp4 and webm are containers
<CounterPillow>
and lossy codecs can usually be any bitrate
<damo22>
oh
<CounterPillow>
what you should be asking is how two video codecs compare to each other at the same quality
<damo22>
h264 and vp9 at the same quality then
<damo22>
vp9 is how many times smaller?
<CounterPillow>
Usual numbers I've heard thrown around are 50% but with a good H.264 encoder like x264 at high profile you're probably closer to 20%
<damo22>
ok
<damo22>
i had a 600MB h264/aac and used ffmpeg to convert it to vp9/opus and i got it down to about 35MB and it looked about the same quality to my eyes
<damo22>
not much noticable difference
<damo22>
it was a screencast
<damo22>
possibly vp9 shrinks down a lot when many pixels are the same between frames
<CounterPillow>
you're not comparing apples to apples there
<CounterPillow>
I can take any garbage CBR VP9 and compare it to a crf H264 and H264 will "win" even though it's the worse codec
<damo22>
what is crf
fling has quit [Remote host closed the connection]
fling has joined #ffmpeg
<CounterPillow>
a rate control mechanism that tries to adjust the bitrate to achieve a similar measure of quality each frame. In contrast, CBR throws a fixed amount of bits per second at the encoder as a budget
zsoltiv has joined #ffmpeg
<AMM>
(constant rate factor, constant bit rate)
<furq>
"h264" is kind of meaningless as well
<furq>
pretty much any device that can record has some terrible asic h264 encoder
<furq>
which will compare extremely badly to x264
<furq>
whereas vp9 that's not from youtube will almost certainly be from libvpx
<CounterPillow>
true
<CounterPillow>
though Intel does ship a VP9 hardware encoder so technically a lot of people have access to something that is not libvpx
<CounterPillow>
not that they'd use it though
junaid_ has joined #ffmpeg
alexherbo2 has joined #ffmpeg
billchenchina has quit [Remote host closed the connection]
billchenchina has joined #ffmpeg
sss1 has quit [Ping timeout: 240 seconds]
sss1 has joined #ffmpeg
junaid_ has quit [Remote host closed the connection]
foul_owl has quit [Ping timeout: 276 seconds]
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
navi has joined #ffmpeg
foul_owl has joined #ffmpeg
alexherbo2 has quit [Ping timeout: 250 seconds]
alexherbo2 has joined #ffmpeg
bitoff has quit [Ping timeout: 276 seconds]
deetwelve has joined #ffmpeg
alexherbo2 has quit [Remote host closed the connection]
alexherbo2 has joined #ffmpeg
Vonter has quit [Remote host closed the connection]
Vonter has joined #ffmpeg
<sss1>
hi all, you probably all already tired of my question, but i still not figured out what i am doing wrong, the problem is what at certain dimensions i have screwed images, i guess it's something with alignment, https://ncl.dark-alexandr.net:44367/s/z5ZbDAfTmm93ACp - here is samples and code used, sorry for annoing )
kurosu has joined #ffmpeg
MootPoot has quit [Quit: Connection closed for inactivity]
Kei_N has quit [Ping timeout: 256 seconds]
Kei_N has joined #ffmpeg
lusciouslover has joined #ffmpeg
billchenchina has quit [Remote host closed the connection]
mosasaur has quit [Quit: Leaving]
billchenchina has joined #ffmpeg
alexherbo2 has quit [Ping timeout: 250 seconds]
lusciouslover has quit [Ping timeout: 245 seconds]
bitoff has joined #ffmpeg
billchenchina has quit [Ping timeout: 268 seconds]
pikapika_lunar has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
pikapika has joined #ffmpeg
bitoff has quit [Ping timeout: 268 seconds]
<galad>
sss1: you are memcpy the frame data into the AVFrame, but you do not check the AVFrame linesize
<galad>
AVFrame linesize is usually aligned to 32 or 64
<galad>
you should make a for cycle
<galad>
copy a line
<galad>
and increase the destination pointer by adding the linesize
<galad>
and repeat
Ogobaga has quit [Quit: Konversation terminated!]
Ogobaga has joined #ffmpeg
zumba_addict has joined #ffmpeg
<sss1>
galad: is this true also for rgbXX pixel format ?, now i am copying whole rgb buffer into data[0] in avframe, is this correct ?
<sss1>
or it's true only for ycbcr pixel format ?
<galad>
check the AVFrame linesize, if it's different than the width then yes
<BtbN>
doesn't matter, linesize is linesize.
shibboleth has quit [Remote host closed the connection]
<sss1>
galad: BtbN thx for advice, i will check
waleee has quit [Ping timeout: 252 seconds]
psykose has joined #ffmpeg
junaid_ has joined #ffmpeg
hightower2 has joined #ffmpeg
<kepstin>
while you could allocate your own buffers for a frame without aligning the line size, this isn't really supported and is likely to cause problems. some filters or codecs use cpu vector instructions which assume aligned data and will either run slower or break if a line isn't aligned. (and may read/write slightly past the image width, including on the last line)
zumba_addict has quit [Quit: Client closed]
Muimi has joined #ffmpeg
l4yer has quit [Remote host closed the connection]
Muimi has quit [Quit: Going offline, see ya! (www.adiirc.com)]
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg
lusciouslover has joined #ffmpeg
rsx has quit [Quit: rsx]
Muimi has joined #ffmpeg
hightower2 has quit [Ping timeout: 255 seconds]
hightower2 has joined #ffmpeg
waleee has joined #ffmpeg
ircgpk has joined #ffmpeg
ZedHedTed has quit [Quit: leaving]
<sss1>
galad: so i got linesize 4128 for rgb24 frame with resolution of 1364x768, currently i am copying width* height * 3bytes from source rgb buffer, is it incorrect ?, as i know source rgb buffer is not aligned
ircgpk has quit [Quit: Leaving]
ircgpk has joined #ffmpeg
kurosu has quit [Quit: Connection closed for inactivity]
<BtbN>
Again: Use the linesize when copying
<BtbN>
And generally, when accessing the buffer
ircgpk_ has joined #ffmpeg
ircgpk has quit [Ping timeout: 268 seconds]
<sss1>
BtbN: it seems i have problem with interperting it's value, i have just one linisize and one buffer vars in example above, in this case linesize is pixels or bytes or some other type ?
ircgpk_ is now known as ircgpk
<BtbN>
it's literally the size of each line of pixels, in bytes
<BtbN>
read its doxy for details
<sss1>
can you post link please ?
<BtbN>
probably in the frame header in avutil
<sss1>
i see, so width*3 looks like it
<BtbN>
what?
<BtbN>
Again: use the linesize when accessing the buffer. It's literally why it exists
<BtbN>
If you don't, you'll likely just read garbage after the first line
<sss1>
yes yes, i mean looks like i got it now, how to use it
<sss1>
need to do few tests
<sss1>
is here any helper to extract bytes per pixel from configured avframe ?
<BtbN>
not really. What kind of helper would you expect?
<sss1>
something to interpert pixel format
<BtbN>
You can query the memory layout of each pixel format via lavu pix_fmt_desc, if you want to write ultra generic code
<BtbN>
But it gets messy quickly, given how many formats there are
<sss1>
i see, thx
cahilnebilir12 has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
minimal has joined #ffmpeg
<sss1>
BtbN: one more question, what shoul i do with alignment size ?, just skip it in destination buffer or ... ?
<BtbN>
if all you want to do is run swscale, you don't really need to bother with an AVFrame at all.
<BtbN>
It has an API to operate on buffers directly
<sss1>
where can i read about it ?
<BtbN>
in the swscale docs
<sss1>
but ycbcr needs some array of three buufers anyway, no ?
<BtbN>
Yeah, buf swscale does not overly care if that's in an AVFrame or some other random buffers
ZedHedTed has joined #ffmpeg
<sss1>
and i will need to deal with alignment anyway ?
<BtbN>
alignment?
<sss1>
linesize
<BtbN>
You will need to provide swscale with the accurate linesize of all buffers you provide
<sss1>
so it will work with unaligned linesize ?
lavaball has quit [Remote host closed the connection]
clarkh has quit [Read error: Connection reset by peer]
bitoff has joined #ffmpeg
bitoff has quit [Remote host closed the connection]
bitoff has joined #ffmpeg
Haxxa has quit [Quit: Haxxa flies away.]
Haxxa has joined #ffmpeg
lucasta has joined #ffmpeg
ircgpk has quit [Remote host closed the connection]
ircgpk has joined #ffmpeg
ircgpk has quit [Remote host closed the connection]
<BtbN>
I don't know if swscale has a minimum alignment. It might be slower if you don't align by a certain amount
Pie-jacker875 has joined #ffmpeg
<Pie-jacker875>
I'm running fedora rpmfusion build ffmpeg-6.0.1-3.fc39 which apparently has backported av1 vaapi but I'm not sure how to use it. I have an RX 7900 XT.
<Pie-jacker875>
the av1 wiki page does not yet have information on vaapi
<sss1>
i think a lot of copypasted logic can be dropped later
BillTorvalds has quit [Quit: Leaving]
<sss1>
BtbN: but i have ugly empty (green) line at right and/or bottom at certain resolutions
<sss1>
does not scaling should scretch at all dimensions ?
<BtbN>
You are probably doing something wrong then
<sss1>
i guess so, but what ?
<BtbN>
either taking the linesize for width, seeing the random garbage on the right, or dealing with a codec that rounds up to multiples of 8 or 16, and ignoring that info
<sss1>
no, it's not garbage, it's zeroes i guess
<sss1>
so it valid empty data
<sss1>
looks green if encoded into video
<Pie-jacker875>
Perhaps I assumed I could use it because the patch notes I found weren't specific, but is av1 vaapi only for intel?
<Pie-jacker875>
I do see VAProfileAV1Profile0: VAEntrypointEncSlice when I run vainfo
<sss1>
Pie-jacker875: av1 encoding is svt only for intel, and decoding works on both vaapi and svq
<sss1>
fc
<sss1>
k
<sss1>
qsv i mean
<sss1>
qsv
<Pie-jacker875>
Okay I'm blind and I see it now
<Pie-jacker875>
I see it's partial, would be useful to know what features work but I realize if it's going to work eventually there's probably no desire to write documentation for it yet
lavaball has joined #ffmpeg
ivanich has joined #ffmpeg
bitoff has quit [Ping timeout: 240 seconds]
<sss1>
Pie-jacker875: i am using av1-qsv daily, works fine
<sss1>
encoding i mean
junaid_ has quit [Remote host closed the connection]
<Pie-jacker875>
I don't have intel hardware
<Pie-jacker875>
I've been itching a bit to experiment with my navi 31 GPU's hardware encoder but I don't have the proprietary drivers installed
<Pie-jacker875>
just last week I attempted to install the proprietary drivers so I could use AMF and it screwed up my system
<sss1>
good thing about intel, is what everything is working with open-source software
<Pie-jacker875>
yeah I'm pretty jealous
<sss1>
unlike most others
<sss1>
they selling intel aXXX gpus for cheap prices
<Pie-jacker875>
perhaps if I had a stronger power supply I would throw an ARC card in my system
<sss1>
i using a380, one of the cheapest
<sss1>
$150 i think
<sss1>
or less
alexherbo2 has joined #ffmpeg
<Pie-jacker875>
hey apparently it does work
<Pie-jacker875>
it's going to take me a while to understand what this command is doing but copypasting it nearly verbatim made it do something