BtbN 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 7.0 is released
Everything has quit [Quit: leaving]
jmcantrell has quit [Killed (calcium.libera.chat (Nickname regained by services))]
__jmcantrell__ is now known as jmcantrell
jmcantrell_ has joined #ffmpeg
jtgd has quit [Ping timeout: 248 seconds]
jtgd has joined #ffmpeg
rsx has quit [Quit: rsx]
five618480339176 has quit [Ping timeout: 248 seconds]
Mirarora has quit [Quit: Mirarora encountered a fatal error and needs to close]
xx has quit [Ping timeout: 264 seconds]
minimal has quit [Quit: Leaving]
Mirarora has joined #ffmpeg
Mirarora has quit [Quit: Mirarora encountered a fatal error and needs to close]
Mirarora has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 265 seconds]
rvalue- has joined #ffmpeg
rvalue has quit [Ping timeout: 276 seconds]
rvalue- is now known as rvalue
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg
bertieb has quit [Read error: Connection reset by peer]
bertieb has joined #ffmpeg
MightyBOB has joined #ffmpeg
Vonter has joined #ffmpeg
vlm has quit [Ping timeout: 252 seconds]
vlm has joined #ffmpeg
aaabbb has joined #ffmpeg
Suchiman_ has quit [Quit: Connection closed for inactivity]
System_Error has quit [Remote host closed the connection]
jmcantrell has quit [Quit: WeeChat 4.4.4]
jmcantrell_ is now known as jmcantrell
System_Error has joined #ffmpeg
System_Error has quit [Remote host closed the connection]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg
coldfeet has joined #ffmpeg
squeaktoy has quit [Ping timeout: 248 seconds]
coldfeet has quit [Quit: Lost terminal]
hjckr has joined #ffmpeg
averne_ has joined #ffmpeg
averne has quit [Ping timeout: 248 seconds]
averne_ is now known as averne
billchenchina has joined #ffmpeg
ndufresne has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Max SendQ exceeded]
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 265 seconds]
DauntlessOne4 has joined #ffmpeg
StephenLynx has joined #ffmpeg
delthas_ is now known as delthas
iliv has quit [Ping timeout: 252 seconds]
manwithluck has quit [Ping timeout: 252 seconds]
DEATH has quit [Ping timeout: 252 seconds]
manwithluck has joined #ffmpeg
DEATH has joined #ffmpeg
JesseTG has joined #ffmpeg
iliv has joined #ffmpeg
alexherbo2 has quit [Remote host closed the connection]
alexherbo2 has joined #ffmpeg
<JesseTG>
I'm getting an error message when trying to decode a video packet: "[rawvideo @ 000000000dfe83c0] Invalid buffer size, packet size 614400 < expected frame_size 1843200" I'm not sure why the packet's smaller than I'm asking for. What info can I find or provide that'll make this easier to diagnose?
robobub has joined #ffmpeg
Marth64 has joined #ffmpeg
Sl4yer has joined #ffmpeg
l4yer has quit [Ping timeout: 276 seconds]
billchenchina has quit [Ping timeout: 265 seconds]
bananicorn has quit [Remote host closed the connection]
EmleyMoor has quit [Ping timeout: 252 seconds]
Sl4yer has quit [Ping timeout: 265 seconds]
bananicorn has joined #ffmpeg
EmleyMoor has joined #ffmpeg
rsx has joined #ffmpeg
billchenchina has joined #ffmpeg
billchenchina has quit [Remote host closed the connection]
billchenchina has joined #ffmpeg
billchenchina has quit [Remote host closed the connection]
l4yer has joined #ffmpeg
l4yer has quit [Max SendQ exceeded]
l4yer has joined #ffmpeg
wobbol has quit [Ping timeout: 276 seconds]
<BtbN>
rawvideo expects exact frame sizes, as defined by resolution and pixel format
<BtbN>
It's even a stretch to call it a decoder
<BtbN>
What actually is your input?
<JesseTG>
A webcam.
<JesseTG>
In that case, 1280x720. The default resolution is 640x480, and that works as expected
<JesseTG>
What can I post that would be useful?
grufwub has left #ffmpeg [#ffmpeg]
<JesseTG>
https://pastebin.com/HUQ1hXXS Here's the code I'm writing. It's ~550 lines of C; it implements a ffmpeg-based video camera driver for (something close enough to) a game engine's camera abstraction
alexherbo2 has quit [Remote host closed the connection]
rex_victor is now known as rex
rv1sr has joined #ffmpeg
alexherbo2 has joined #ffmpeg
redeeman has joined #ffmpeg
<redeeman>
when I try -threads 64 using libsvtav1 it says maximum recommended is 16 - I get that it may not be making optimal use of it, but it shouldnt hurt encoding quality to do so, should it?
esu has quit [Quit: Esu]
ops has joined #ffmpeg
<ops>
hello
esu has joined #ffmpeg
<furq>
redeeman: it shouldn't make any difference to quality, you'll just get diminishing returns for speed
<ops>
if i'm doing x11grab, i can only capture one screen at a time, correct?
<ops>
i can't do -i :0.0 -i :0.1 -i :0.2, etc?
<redeeman>
furq: thats what I thought, thanks
<ops>
each -i would require its own ffmpeg process?
<redeeman>
it seems to use between 16 and 40 cores, mostly around 20
<furq>
ops: you need to specify -f x11grab before each -i
<ops>
awesome, so how do i specify separate output files for each -i?
<ops>
so i would just do -map 1:v:0 with additional -hls_segment_filename and -f hls arguments?
<furq>
you'd need to duplicate everything after -i
<ops>
ok, will experiment
wziko has quit [Ping timeout: 248 seconds]
alexherbo2 has quit [Remote host closed the connection]
alexherbo2 has joined #ffmpeg
alexherbo2 has quit [Remote host closed the connection]
nasso has quit [Quit: nasso]
MightyBOB has joined #ffmpeg
nedril_ is now known as Nedry
lemourin0 has joined #ffmpeg
lemourin0 is now known as lemourin
lemourin has quit [Killed (platinum.libera.chat (Nickname regained by services))]
harshilkhanna has joined #ffmpeg
Tano has quit [Quit: WeeChat 4.4.2]
yans has joined #ffmpeg
haritz has joined #ffmpeg
<JesseTG>
...oh, hell. I'd forgotten to set the `video_size` parameter in the dict passed to `avformat_open_input`.
<Nedry>
Ffmpeg together with Yt-dlp on Windows 10 can't maintain yt livestream download for 4 hours straight, it cuts or hangs at around 2-3hrs long run
<Nedry>
I wonder if it can maintain long running dowload on other OSes
olndrxyz has joined #ffmpeg
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Max SendQ exceeded]
olndrxyz has quit [Read error: Connection reset by peer]
<Nedry>
I made a Windows cmd standalone cronejob script for it to re enact once in every 2hrs
<Nedry>
And the targeted live streams gets downloaded with 2hrs duration on each of them
wziko has joined #ffmpeg
wziko has quit [Max SendQ exceeded]
rsx has quit [Quit: rsx]
wziko has joined #ffmpeg
JesseTG has quit [Quit: Leaving]
<Nedry>
If anybody wants my cronscript for windows, pm me, i'll copy paste in the pm
alexherbo2 has joined #ffmpeg
Tano has joined #ffmpeg
wziko has quit [Ping timeout: 265 seconds]
harshilkhanna has quit [Quit: Client closed]
harshilkhanna has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 252 seconds]
harshilkhanna has quit [Quit: Client closed]
Rena has quit [Quit: $WITTY_QUIT_MESSAGE]
olndrxyz has joined #ffmpeg
Rena has joined #ffmpeg
<olndrxyz>
Hi, does libtheora have any advantadge compared to webm/vp8-vp9 today? I read in a email thread from 14 years ago that libtheora was much faster at both encoding & decoding than anything available for vp8 at the time. Is that still the same today with vp8-vp9?
<JEEB>
heck, you have AV1 now
<JEEB>
consider that theora was VP3 on steroids
<furq>
if there is any advantage then those would be the only ones
<JEEB>
then you had VP5,6,7,8
<furq>
but i doubt anyone has even benchmarked it for years
<JEEB>
and then google did VP9, and AOM did AV1
<furq>
theora is more or less irrelevant
<JEEB>
yea
<JEEB>
720p 10bit 4:2:0 was ~17fps with SVT-AV1 and preset 5, which made me smile
<JEEB>
and this was on a Zen2 AMD laptop
<furq>
i think every browser has dropped theora support now
<furq>
which is probably the last nail in the coffin
<furq>
i wonder what wikipedia are doing about that
YuGiOhJCJ has joined #ffmpeg
<galad>
wasm or javascript decoder, it's fast enough
<JEEB>
yea, that should keep those old lowres videos playable
<olndrxyz>
JEEB: but I don't have the hardware. afaik AV1 is heavy on resources...
rvalue is now known as WastedGymMembers
<furq>
vp8 encoding is fast and vp8/vp9 decoding are fast
WastedGymMembers is now known as DntBuyGymPlan
<JEEB>
olndrxyz: on what level? av1 decoding with dav1d is quite well optimized at this point. slower than vp9, but quite a lot of stuff should be possible
<JEEB>
(vp9 with ffvp9 specifically)
<olndrxyz>
furq: could google, as the owner of the patents, change the vp8-vp9 license and screw us? for example the Xiph.org Foundation at the time negotiated an irrevocable free license to the VP3 codec with On2.
<olndrxyz>
JEEB: I have to transcode some videos and I own a t420 and a Pixel 4A. Are the available software encoders/decoders good enough for my hardware?
olndrxyz has quit [Read error: Connection reset by peer]
<JEEB>
tbh I would just go H.264 with x264 if you want good enough compression and speed. the mobile device has a hardware decoder, and the encoder is old enough that it's blazing fast. not that newer encoders wouldn't run, of course
<JEEB>
but if you just need quick encoding for your mobile device just go H.264 :P it has CABAC and x264 is stil a hilariously good GPL encoder for it
<JEEB>
`-preset faster` goes wroom-wroom, and both the mobile SoC as well as browsers support it.
lolok has quit [Quit: lolok]
<Marth64>
I don't know if I'm just stubborn but I usually end up just falling back to libx264 or h264 in general when I get frustrated or blocked by limits. It's great for what it is <=1080p.
<Marth64>
it just werks
<JEEB>
even higher for me, until I needed 10bit that plays in browsers
<JEEB>
also for 2160p it's really easy to get into the level 6.x trap with x264 since it actually implemented it
<JEEB>
except... levels 6.x actually lengthenened the available motion vector length :DD
<JEEB>
you can guess how hw decoders coped with that
<JEEB>
just setting `-level 5.2` should handle that to limit it
<JEEB>
ebin, my 2021 zen3 AMD laptop does SVT-AV1 at ~35fps with same input clip and parameters that the 2020 laptop did ~17fps with :)
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
<Marth64>
good to know thx for sharing :)
<olndrxyz>
JEEB: H264 is still a proprietary format, I want to use a free format.
<JEEB>
you might be thinking of a different format
<JEEB>
from initial format publication in 2003 to current year. there is a standard ITU pre-publishing timeout for the latest version, but that seems common for ITU specs
<JEEB>
what these cutting edge (at the time of publication) ITU/ISO commonly developed formats were not designed as was an attempt at being royalty free. but that's *very* different from being proprietary/closed. it's just a different process where everyone pinky-promises that they will license IPR related to implementing the format in a FRAND manner.
<JEEB>
(we could see with H.265 from 2013 that this sort of development ended up in a mess, but that's once again not related to a format being proprietary or closed)
<JEEB>
heck, as a funny anecdote I was on the HEVC development mailing list as a complete rando and the archives are public, while I can bet you on whether you can get some publicly available archive of the AV1 standardization process :)
<olndrxyz>
JEEB: yes the specs are available but if you want to serve your files to a large audience you have to pay some fees afaik. And they can still change the terms.
<JEEB>
at least MPEG-LA specifically always kept *serving* free video over interwebs without royalties
<JEEB>
which is partially why H.264 got such adoption for web video
<JEEB>
also it still stands that you indeed were not talking about proprietary, but another word or wording. most likely "designed to be (hopefully) royalty free".
DntBuyGymPlan has quit [Quit: ChristmasMaintenance]
<JEEB>
why certain distributions don't themselves distribute H.264 decoders or encoders is because unlike content distribution, distribution of binaries has royalties attached to it according to various places that announced they were licensing. also that is why getting nicely built and signed binaries of openh264 from Cisco then became completely OK :)
<JEEB>
(also looking at that expired or expiring patents listing, next year might become interesting with regards to features defined in 2005 ed of H.264, which is what fits most users' interests)
alexherbo2 has quit [Remote host closed the connection]
<olndrxyz>
JEEB: yes I meant royalty free precisely.
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
<JEEB>
but yea, nowadays (the past N years) f.ex. debian or ubuntu don't seem to care about this stuff and just distribute binaries to users, so they're taking care or risk of that side, and then if you want to distribute video over a HTTP server or so, for H.264 content that the user does not pay for over the interwebs according to my understanding was always free. and pretty much everyone and their dog has decode
<JEEB>
capability
<furq>
you only have to pay fees for hosting h264 content if you're charging for it
<furq>
if that wasn't clear
<furq>
if you're not charging then unless you're one of those agpl guys there's no real reason to not just use x264
<JEEB>
why would AGPL be relevant?
Blacker47 has quit [Quit: Life is short. Get a V.90 modem fast!]
<furq>
it's still very good and everything can decode it
<furq>
just philosophically
<furq>
you know what those guys are like
alexherbo2 has joined #ffmpeg
<olndrxyz>
why do vlc and other media players show no metadata other than the title when I use the ffmpeg command: `ffmpeg -i input.opus -codec copy -metadata title="Foo" -metadata author="Bar" -y output.opus` ?
<furq>
author isn't really a common tag
<furq>
i don't have vlc installed so idk if it's predefined there
<furq>
but at least my music player doesn't have a field for it
<JEEB>
opus is OGG so you can check what OpusTags get written
<furq>
i just ran the same thing and it writes both tags
<furq>
i assume for vorbis comment formats it doesn't do any tag remapping
<JEEB>
if there is something like boxdumper or AtomicParsley for OGG
<furq>
well there's ffprobe
<furq>
or opusinfo
<furq>
i still don't understand why ogg, vorbis and opus all need separate tools to inspect it but never mind
wobbol has joined #ffmpeg
jemius has joined #ffmpeg
<olndrxyz>
JEEB-furq: I'm one of those AGPL guys xD and I think we are right to fight for free tech. Thanks for the conversation and the help anyway.
<JEEB>
well, thankfully the designed to hopefully be royalty free formats is what I prefer post-H.264
<JEEB>
and hopefully soon H.264 will be royalty free (it might even be that in your juristiction main profile already is?)
<JEEB>
that wiki article being an interesting artifact of people looking relevant documents up
<JEEB>
this patent royalty stuff is something that I understand corporations to care about, but as a private citizen not selling binaries it's something quite far from me :D
<JEEB>
also my thing was not about X being better than Y, mostly just trying to take the words proprietary or free as in libre away from the discussion, since all those ISO/ITU specified formats are as far as possible from actual proprietary stuff that you need to literally reverse engineer. and any implementations can be under whatever license, so they can be just as libre as x264 is GPL.
<furq>
there's a vp9 patent pool that's actively rentseeking if you want the situation to be even less clear
<furq>
although only for people shipping de/encoders
<JEEB>
yea, each of those designed to be hopefully royalty free formats has those
<furq>
i assume they could do the same for vp8 if there was enough profit motive
<furq>
does av1 have one yet
<furq>
i figured aom had bought enough people off
<JEEB>
they did have one, I think Google ended up striking a deal with them or whatever
<JEEB>
but I wouldn't be surprised if other orgs still have one
<JEEB>
yea
<olndrxyz>
JEEB: I think it's still a danger if you control the patent for such important techs like video and audio and just because nowadays they only charge corps doesn't mean they won't charge users in the future. I prefer to support other organizations that develop tech for the humanities.
<JEEB>
sisvel at least
<furq>
there isn't just one patent
<JEEB>
yea, and organizations not part of the standardization effort of course have had no reason to disclose
<furq>
it's difficult to make a codec that doesn't include some idea that someone can claim they own the rights to
<furq>
or practically many ideas
<JEEB>
for AV1 at least D and Qualcomm have specifically noted that they are *not* agreeing with AOM's royalty free licensing
Marth64 has quit [Quit: Leaving]
<furq>
there's an opus patent pool as well
<furq>
these are set up by bodies that have nothing to do with the development of the codec but believe it infringes some patent that they probably bought 15 years ago and have never used
<JEEB>
of course the starting point is completely different and indeed with AV1 the AOM members that have signed the AOM patent license will grant that
<furq>
i guess the moral is that someone could potentially try to extort you for doing literally anything
<furq>
so it's best to just hope it doesn't happen and not worry about it
<JEEB>
or if you are a corporation you talk to lawyers
<olndrxyz>
I changed author with artist and vlc now shows all the tags.
<JEEB>
regarding the AOM formats the most meh part is how the development is handled
<JEEB>
I can go all the way back to beginning of 2000s or so with H.264 development document archives
<JEEB>
seeing the fight between panasonic and another company with regards to how film grain/full HD stuff should be handled is both educational and fun.
<JEEB>
AOM archives on the other hand is closed to non-members just like the rest of the process :P
<furq>
are they following the vpx model of updating the spec as they write the code
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
Traneptora has quit [Quit: Quit]
JesseTG has joined #ffmpeg
foul_owl has quit [Quit: WeeChat 3.8]
foul_owl has joined #ffmpeg
alexherbo2 has quit [Remote host closed the connection]
<JesseTG>
Is it possible to list the possible frame sizes of a video camera? I'm implementing an API that allows clients to request webcam input of a specific size, but I want to pick something close enough if it's not supported
alexherbo2 has joined #ffmpeg
alexherbo2 has quit [Remote host closed the connection]
alexherbo2 has joined #ffmpeg
nadhor has joined #ffmpeg
<klaxa>
JesseTG: ffmpeg -f video4linux2 -list_formats all -i /dev/video0
<JesseTG>
Is there some cross-platform equivalent? I'm using ffmpeg to support a variety of platforms, not just Linux
<JesseTG>
That function is only intended to be used internally by ffmpeg
<klaxa>
hmm...
<JesseTG>
My gut instinct tells me I can somehow use the options defined here https://ffmpeg.org/ffmpeg-devices.html (via the C API equivalents), but I'm not sure how -- I don't see anywhere that info is surfaced
<klaxa>
well, you would still have to try them one by one, but i see what you mean
wziko has joined #ffmpeg
<JesseTG>
Actually, no. If a given video size fails, I'll try larger sizes of the same aspect ratio. Then smaller sizes of the same aspect ratio. Then look for a close-enough aspect ratio. Then if all else fails: use the default size, scale it, and hope the player doesn't mind awkwardly-stretched video input
<JesseTG>
(Assuming by "one by one" means "linear search from the beginning of the list")
<JesseTG>
I just need to know what video sizes are available, then I can do the rest