[Ffmpeg-devel-irc] ffmpeg.log.20180424

[burek]

[00:00:11 CEST] <JEEB> and then there's ffmpeg.c that has support for some of those hwenc APIs :P
[00:00:31 CEST] <tuna> I have only ever found decoding with hardware examples...or encoding with software frame input...but no hardware frame input encoding
[00:00:42 CEST] <alone-z> my bat file too long ;( may be ffmpeg can reading file with parameters (i hope this file will be not have lenght limit!) instead BAT file ?
[00:02:16 CEST] <JEEB> then use python or something?
[00:02:39 CEST] <alone-z> hm, thank you
[00:02:53 CEST] <BtbN> The Command Line Length limit is a Windows-Wide limit
[00:03:02 CEST] <BtbN> iirc longer than 8200 characters is just not possible
[00:03:06 CEST] <JEEB> lol
[00:03:29 CEST] <alone-z> BtbN, yes, my lines near this.
[00:03:44 CEST] <alone-z> BtbN, how can i avoding it
[00:03:45 CEST] <alone-z> ?
[00:04:08 CEST] <BtbN> Use Linux
[00:04:18 CEST] <alone-z> this is only solution?
[00:04:40 CEST] <BtbN> I'm not aware of being able to change that limit of Windows
[00:04:53 CEST] <BtbN> or fix your commandline, 8k chars is really quite long
[00:05:13 CEST] <alone-z> i have many drawtext with parameters...
[00:05:34 CEST] <JEEB> sounds like you might just want an ASS file
[00:05:37 CEST] <alone-z> well, may be ffmpeg can reading ommand line options file?
[00:05:39 CEST] <JEEB> with the subtitle lines
[00:05:45 CEST] <JEEB> no it does not
[00:06:26 CEST] <alone-z> i c
[00:06:27 CEST] <alone-z> thank you
[00:07:42 CEST] <furq> yeah it sounds like you just want subtitles
[00:08:25 CEST] <alone-z> furq. hello
[00:08:27 CEST] <alone-z> no!
[00:08:45 CEST] <BtbN> yes. Drawing text at various times in various locations is what subtitles do
[00:09:10 CEST] <alone-z> it's just current time printing 4*12
[00:09:18 CEST] <alone-z> times.
[00:09:25 CEST] <alone-z> with parameters..
[00:09:44 CEST] <BtbN> what?
[00:09:59 CEST] <furq> that doesn't seem complicated enough to take 8200 charactres
[00:10:00 CEST] <alone-z> i need to print TIME (localtime) many times.
[00:10:22 CEST] <alone-z> just a moment.
[00:10:24 CEST] <alone-z> i will show
[00:10:37 CEST] <furq> but yeah if you really need to do this with drawtext then you can use -filter_complex_file
[00:10:46 CEST] <furq> or uh
[00:10:49 CEST] <furq> -filter_complex_script
[00:12:01 CEST] <alone-z> https://pastebin.com/pzJzvp0s
[00:12:55 CEST] <alone-z> this file is working, but - if i will add some parameter to drawtext
[00:13:02 CEST] <alone-z> it will too long line ;(
[00:17:45 CEST] <alone-z> farq
[00:17:50 CEST] <alone-z> thank you again
[00:17:55 CEST] <alone-z> sorry - firq
[00:17:58 CEST] <alone-z> sorry - furq
[00:18:03 CEST] <alone-z> ;-((
[00:18:20 CEST] <alone-z> you example all i need.
[00:18:45 CEST] <alone-z> (and you monorhome filter!)
[00:48:14 CEST] <Stumbler_> so I have successfully setup and tested my media server. I can broadcast a h.264/AAC file from filesystem via ffmpeg -re -i FILENAME -c flv rtml://media.server.url
[00:48:52 CEST] <Stumbler_> I feel like I should be able to ceate a file stream and pipe my mic to it in AAC but I dont know enough of the keywords to take this to google yet
[00:49:17 CEST] <Stumbler_> thereby allowing me to broadcast my local mic through the media server into a remote browser.
[00:49:49 CEST] <JEEB> browsers suck for true live streams
[00:49:57 CEST] <Stumbler_> yeah, but thats what I have.
[00:50:10 CEST] <Stumbler_> Im okay with lowering the quality of the stream if I need to
[00:51:02 CEST] <furq> use icecast for audio-only livestreaming to browsers
[00:52:11 CEST] <Stumbler_> Im building my own solution. What Im looking for is a way to pipe my mic to a filesystem handle.
[00:52:12 CEST] <JEEB> yea I guess if you can just use an <audio> tag
[00:52:25 CEST] <JEEB> then if that format is supported it should JustWork
[00:52:43 CEST] <JEEB> and you can circuimvent the awfulness of MSE-based solutions
[00:56:15 CEST] <furq> you don't even need a video tag, the url should just play directly
[00:56:24 CEST] <furq> or audio tag rather
[00:57:10 CEST] <furq> anyway
[00:57:16 CEST] <furq> !indev alsa @Stumbler_
[00:57:16 CEST] <nfobot> Stumbler_: http://ffmpeg.org/ffmpeg-devices.html#alsa
[00:57:57 CEST] <Stumbler_> ooo, thanks!
[00:58:11 CEST] <furq> dshow on windows, avfoundation on osx
[04:14:39 CEST] <FunkyBob> morning, all... have inherited some code, and can't quite decipher what this is saying:  -filter_complex "[0:v:0]setpts=N+0[v0]"
[04:14:52 CEST] <FunkyBob> it seems to say to set the pts to the frame number from start of stream, but
[04:15:08 CEST] <FunkyBob> [0:v:0] <-- can't find clarity on this .. just [0:v],
[04:15:14 CEST] <FunkyBob> or why the +0
[04:15:19 CEST] <FunkyBob> or what the [v0] means
[04:15:44 CEST] <FunkyBob> the whole command is trying to encode some PNGs into a mp4
[04:19:54 CEST] <furq> FunkyBob: 0:v:0 is input #0, video stream #0
[04:20:02 CEST] <furq> usually files only have one video stream so 0:v is the same
[04:20:29 CEST] <furq> and [v0] is just whatever you decide to call the output
[04:21:02 CEST] <furq> i have no idea why you'd do N+0 for setpts though
[04:21:17 CEST] <furq> i also don't really know why you'd do that at all
[04:21:38 CEST] <FunkyBob> thanks, furq
[04:22:36 CEST] <FunkyBob> yeah, none of this is done how I'd like... and I expect it will take some time to unwind all the reasoning
[04:24:24 CEST] <furq> i wouldn't necessarily there was any reasoning
[04:24:30 CEST] <furq> i wouldn't necessarily assume there was any reasoning
[04:24:39 CEST] <FunkyBob> :)
[04:24:42 CEST] <FunkyBob> there is that
[04:25:10 CEST] <furq> based on this channel, about 98% of ffmpeg commands in existence were copied off stackoverflow
[04:26:20 CEST] <FunkyBob> currently this program forks ffmpeg and communicates with it via TCP... instead of just accessing libav* directly
[04:27:28 CEST] <stephen> youtube
[04:28:25 CEST] <FunkyBob> gesundheit
[04:28:44 CEST] <stephen> My apologies
[04:28:54 CEST] <stephen> Searching logs, unfruitfully even
[04:31:41 CEST] <FunkyBob> ok... so is it ever sane to have -f twice?
[04:32:03 CEST] <furq> pastebin the command
[04:33:11 CEST] <FunkyBob> https://dpaste.de/wOYE
[04:36:15 CEST] <furq> yeah one is an input option and one is an output option
[04:36:23 CEST] <FunkyBob> suspected
[04:56:06 CEST] <stephen> Ah, there we are
[08:36:42 CEST] <Nethe> hey :) Guys one question. How can I specify the GPU for transcoding in ffmpeg? I tried ... -gpu N .... but is not working. Example:
[08:36:51 CEST] <Nethe>  ffmpeg -y -hwaccel cuvid -c:v h264_cuvid -i 'udp://227.10.20.58:1234?fifo_size=100000000' -map i:0x1022  -g 20 -vcodec  h264_nvenc -gpu 3 -map i:0x1023 -ab 128k -ar 48k -acodec aac -f mpegts udp://@239.0.0.100:1234?overrun_nonfatal_option=1
[10:12:28 CEST] <peak3d> Hi, we kodi devs have some issues on configure step windows using msys2 (both 32/64 bit), configure takes ~10-20 minutes
[10:12:45 CEST] <peak3d> we tracked down the issue to this place: https://github.com/FFmpeg/FFmpeg/blob/release/4.0/configure#L6722-L6724
[10:13:04 CEST] <BtbN> Using dash instead of bash helps a lot
[10:13:06 CEST] <JEEB> yes, that is a known issue partly because of how Windows tracks forks IIRC. and then Windows itself made it even slower.
[10:13:32 CEST] <peak3d> Any solution / trick what we can tell our devs?
[10:13:34 CEST] <JEEB> there were some configure improvements that got merged from Libav, and those implemented a stack machine or so in the configure script.
[10:13:45 CEST] <JEEB> BtbN's recommendation is the first thing
[10:14:01 CEST] <BtbN> It's also immensely faster in WSL, their way of emulating(?) fork is fast.
[10:14:09 CEST] <JEEB> second thing is to mostly utilize cross-compilation (from WSL or VM)
[10:14:11 CEST] <peak3d> ok, good to know
[10:14:25 CEST] <JEEB> yea, it definitely is not as slow on *nix, even on WSL
[10:14:27 CEST] <peak3d> so we'll first try the patch how it helps
[10:14:38 CEST] <BtbN> what patch?
[10:14:38 CEST] <JEEB> "the patch"?
[10:15:00 CEST] <peak3d> So most probably we'llfrom libav
[10:15:07 CEST] <BtbN> They were merged.
[10:15:10 CEST] <BtbN> And are what made it slow
[10:15:12 CEST] <JEEB> no, that improvement is what made it slow also
[10:15:18 CEST] <peak3d> ok, not sure if we already have it in
[10:15:19 CEST] <JEEB> and was merged, that is why it is slow on windows
[10:15:27 CEST] <peak3d> ah, right, ok
[10:15:29 CEST] <JEEB> if it is slow as molasses, you have it
[10:15:30 CEST] <peak3d> no I got it
[10:15:40 CEST] <peak3d> we have it :-)
[10:15:41 CEST] <BtbN> For some reason, the latest Windows 10 Feature-Update brought configure to a grinding halt as well
[10:15:45 CEST] <JEEB> yea
[10:15:50 CEST] <JEEB> so there were changes on windows' side as well
[10:15:58 CEST] <JEEB> it plays into some funny game
[10:16:09 CEST] <BtbN> The cost of creating/destroying processes must have increased quite a bit
[10:16:43 CEST] <peak3d> Ok, we'll provide a compiled binary to the folks, then ppl who just "use" ffmpeg are happy
[10:16:50 CEST] <BtbN> I'd just not build on Windows anymore, there even is a MSVC compatible cross compiler for Linux now.
[10:17:07 CEST] <peak3d> PPL who want to develop in ffmpeg should switch to linux :-)
[10:17:25 CEST] <BtbN> Or just accept that they need to wait a few minutes on configure
[10:18:05 CEST] <peak3d> yes, we'll accept, just wanted to be sure that we don't do anythng wrong
[10:18:13 CEST] <dragmore88> Is it possible for ffmpeg using a https url from Youtube as input for video ?
[10:19:06 CEST] <CoreX> dragmore88 to download the video ?
[10:19:14 CEST] <BtbN> dragmore88, no.
[10:19:18 CEST] <CoreX> if you want to download the video just use youtube-dl
[11:04:45 CEST] <dragmore88> live streaming ;)
[14:06:20 CEST] <GamleGaz> Do I need extra libraries to demux/decode video (ie mp4, vp8), or is that included in the ffmpeg internals?
[14:08:40 CEST] <Mavrik> H.264 decoder is in ffmpeg, VP8 requires libvpx compiled in IIRC
[14:09:25 CEST] <GamleGaz> okay thanks
[14:09:37 CEST] <the_gamer> hi again, yesterday i was told not to use -q:v 1 to have the same quality but -qp 0. now i need perfect audio. so i guess -q:a 1 is not good, either? what to use instead?
[14:10:23 CEST] <c_14> GamleGaz: there's an internal vp8 decoder too
[14:12:17 CEST] <BtbN> the_gamer, just use flac?
[14:12:53 CEST] <the_gamer> never used it, does giving the ending ensure good quality?
[14:13:06 CEST] <the_gamer> is it audio only?
[14:13:18 CEST] <the_gamer> i always used wav when i needed good audio
[14:14:34 CEST] <BtbN> It's a lossless codec, sooo...?
[14:15:20 CEST] <BtbN> but honestly, you're not going to get any audible imperfection when using aac at a relatively high bitrate
[14:15:25 CEST] <the_gamer> BtbN, i don't know, just asking for information about it. so ffmpeg needs no parameter for it?
[14:15:56 CEST] <BtbN> flac is only for archival of original material, or for audiophiles with questionable hearing-abilities
[14:16:26 CEST] <durandal_1707> i have golden ears
[14:16:34 CEST] <the_gamer> i need the audio only from a video file but in as good quality as possible because i want to edit it afterwards
[14:16:59 CEST] <BtbN> so, just use -c copy, and use the untouched original
[14:17:07 CEST] <BtbN> No point in re-encoding it
[14:17:29 CEST] <klaxa> only if the editing program used doesn't support the codec
[14:17:45 CEST] <the_gamer> Stream #0:1[0x1100]: Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 192 kb/s
[14:18:01 CEST] <the_gamer> the original ending would be .ac3? never seen that one
[14:18:45 CEST] <BtbN> I doubt ac3 has an original ending
[14:18:51 CEST] <BtbN> just put it in mp4 or mkv
[14:19:33 CEST] <the_gamer> audacity can import mkv directly?
[14:19:34 CEST] <the_gamer> wait
[14:19:36 CEST] <CoreX> no need to transcode ac3 just leave it like that and use -c copy or -acodec copy
[14:19:47 CEST] <BtbN> acodec is deprecated.
[14:20:11 CEST] <CoreX> ohh not useable anymore :(
[14:20:23 CEST] <BtbN> that's not what deprecated means
[14:20:40 CEST] <BtbN> it still works because a huge amount of scripts and guides keep using it
[14:20:44 CEST] <CoreX> against and such
[14:21:28 CEST] <klaxa> you use -c:a nowadays
[14:21:32 CEST] <klaxa> instead of -acodec
[14:22:03 CEST] <the_gamer> oh, audacity can, so there is no need for ffmpeg in this case
[14:23:08 CEST] <the_gamer> ,
[14:23:17 CEST] <the_gamer> sorry the dot was my cat
[14:26:25 CEST] <lieselotte> May I ask if there is any way to specify -to and -ss as number of frame instead of time?
[14:31:57 CEST] <the_gamer> lieselotte, frames are the last part of it
[14:32:06 CEST] <the_gamer> and it is frame accurate
[14:32:17 CEST] <the_gamer> but i don't know if you can give it frames only
[14:35:09 CEST] <lieselotte> the_gamer: That's the problem. I would like to use frames only. I don't want to deal with babylonian arithmetics in my scripts.
[14:36:25 CEST] <c_14> lieselotte: only if you use the trim/select filters
[14:37:11 CEST] <lieselotte> the_gamer: And, I am not sure how to format the parameter to force the last part will be accepted as a number of frame, because 00:00:01.2 means 30th frame.
[14:38:07 CEST] <lieselotte> c_14: Could they be used without reencoding?
[14:38:12 CEST] <c_14> no
[14:38:56 CEST] <lieselotte> c_14: That's what I thought.
[14:38:58 CEST] <c_14> timestamps are in the format HH:MM:SS.mmm
[14:39:10 CEST] <c_14> where mmm is milliseconds
[14:39:39 CEST] <lieselotte> c_14: Correct, I discovered the same. The . is decimal point.
[14:41:04 CEST] <c_14> you can write a script using ffprobe -show_frames -of json to get the timestamps for each frame
[14:41:09 CEST] <c_14> roundabout but it'll work
[14:41:35 CEST] <CoreX> do you have an example of that c_14
[14:42:23 CEST] <lieselotte> c_14: Thanks. That looks like a right way.
[14:43:00 CEST] <c_14> CoreX: the script itself no, but the ffprobe command should be something like `ffprobe -show_frames -select_streams v -show_entries frame=best_effort_timestamp_time -of json'
[14:43:58 CEST] <c_14> most scripting languages should be able to injest that as json (python json.load(sys.stdin) or so) and then access the frame number you want in the array
[14:45:32 CEST] <CoreX> thanks gonna take a look further into it
[14:45:44 CEST] <lieselotte> Is there any other format than json in -of option? TSV/CSV would be better for my usage
[14:47:10 CEST] <c_14> yeah, there's csv and flat afair
[14:48:02 CEST] <c_14> https://ffmpeg.org/ffprobe.html#Writers
[14:55:48 CEST] <lieselotte> Thanks u all, you helped me a lot.
[16:07:47 CEST] <ozette> is it possible to generate a video file of a specified size?
[16:08:32 CEST] <BtbN> generate in what sense?
[16:08:46 CEST] <BtbN> If you want to compress a given video to a specific size, that's what two pass encoding is for.
[16:09:21 CEST] <ozette> like lavfi but not just duration and resolution
[16:10:40 CEST] <ozette> so two pass encoding can increase the size of a video to a specific size?
[16:10:57 CEST] <BtbN> well, if you put the encoder into cbr mode, you can calculate the size from the duration
[16:11:09 CEST] <ozette> what is cbr mode?
[16:11:15 CEST] <BtbN> constant bitrate
[16:11:22 CEST] <ozette> oh
[16:12:03 CEST] <BtbN> increasing the size of an existing video is usually pointless
[16:12:05 CEST] <ozette> if it's not in cbr mode it's variable?
[16:12:07 CEST] <BtbN> quality will not get better
[16:12:23 CEST] <ozette> i don't want it to get better though
[16:12:45 CEST] <ozette> i'd like to generate videos of different sizes to test in my application
[16:35:41 CEST] <GamleGaz> is there a way to probe the pixel format of a video in an AVIO context?
[16:36:14 CEST] <garoto> ozette: if only the file size is relevant for your test purposes, what about trimming an existing large file to the desired size via -ss & -t
[16:37:01 CEST] <ozette> garoto: I don't really have any large video files
[16:37:20 CEST] <ozette> but that'd be an option yea
[16:37:36 CEST] <Mavrik> GamleGaz: well you need to get video stream info for the video format so you have to go through all the steps of parsing the container (AVFormatContext) and figuring out the codec (AVCodec / AVCodecContext)
[16:38:43 CEST] <GamleGaz> I've tried setting up a AVformatcontext, however avformat_find_stream_info doesn't catch the pixel_format
[16:40:01 CEST] <Mavrik> GamleGaz: yeah, you need to go one step further and get AVCodecContext based on codec of the stream
[16:40:03 CEST] <Mavrik> and that has pix_fmt
[16:50:24 CEST] <GamleGaz> thanks!
[16:54:21 CEST] <JEEB> it might only get it after you finish decoding something
[16:54:41 CEST] <JEEB> so the most accurate way to get some representation is to decode until you get your first AVFrame
[17:04:46 CEST] <thinkout> esc
[17:05:15 CEST] <thinkout> hello
[17:50:35 CEST] <GamleGaz> Is there a way to force the AVIO to probe the entire buffer in avformat_find_stream_info? If I set the avio_ctx_buffer value to less than than the file buffersize it reads around 90% of the buffer and can't find the info needed
[18:17:11 CEST] <GamleGaz> JEEB: how do I decode frames without setting the pixel format on the decoder?
[18:17:56 CEST] <JEEB> you don't need to?
[18:18:03 CEST] <JEEB> you just feed AVPackets to it
[18:21:44 CEST] <GamleGaz> So av_read_frame -> avcodec_decode_video2 should yield a frame even without pixel formats?
[18:33:37 CEST] <Filarius> I want to check how raw each of YUV color planes looks like in yuv420p video, I remember I found command for it long time ago, but now, when I need it, I can't find it in google
[18:35:29 CEST] <furq> !filter extractplanes @Filarius
[18:35:29 CEST] <nfobot> Filarius: http://ffmpeg.org/ffmpeg-filters.html#extractplanes
[18:35:54 CEST] <artao> AHA!
[18:36:07 CEST] <artao> this channel DOES exist ... how "alive" is it tho, that's the question
[18:36:26 CEST] <artao> so I have some footage from my tablet. Like a dumbass, I recorded it vertically, not "proper"
[18:36:36 CEST] <artao> I'm using Avanti as my frontend for ffmpeg
[18:36:58 CEST] <artao> i told it to rotate the footage 90 degrees, and it does indeed do that
[18:37:17 CEST] <artao> however, the video remains vertical, and as such the rotated footage is stretched
[18:37:21 CEST] <artao> what am I doing wrong?
[18:37:29 CEST] <c_14> depends on what the frontend is doing
[18:37:37 CEST] <artao> i
[18:37:44 CEST] <artao> i'm totallay new to using Avanti
[18:37:45 CEST] <furq> i have no idea how to use avanti, but typically you'd just set (or unset) the rotate metadata in the container
[18:37:50 CEST] <artao> and have barely used command line ffmpeg
[18:38:24 CEST] <artao> hmm, i wonder if #avanti exists
[18:38:30 CEST] <artao> nope
[18:38:44 CEST] <furq> you're generally better off just using ffmpeg
[18:39:40 CEST] <furq> check if the rotate metadata is set with ffprobe or mediainfo
[18:40:22 CEST] <furq> then ffmpeg -i foo.mp4 -c copy -metadata:s:v:0 rotate="XX" bar.mp4
[18:40:32 CEST] <furq> where XX is 0, 90, 180 or 270
[18:40:51 CEST] <Filarius> thanks furq, but I just found right words to google to find this command :D
[18:40:59 CEST] <artao> furq how?
[18:41:15 CEST] <artao> yeah. that's confusing as hell :\
[18:41:27 CEST] <artao> which is why i'm using Avanti as frontend to ffmpeg
[18:41:43 CEST] <artao> no need to figure out a proper CLI that actually works
[18:42:02 CEST] <artao> which can take quite awhile and be very very frustrating and irritating (for me at least)
[18:42:14 CEST] <artao> once you KNOW ffmpeg CLI, sure ...
[18:42:27 CEST] <artao> but I just need to quickly rotate this footage 90 degrees CW
[18:45:55 CEST] <durandal_1707> artao: this is ffmpeg help channel, if you do not want to use ffmpeg your are doing something wrong
[18:46:07 CEST] <artao> I AM using ffmpeg
[18:46:10 CEST] <artao> via Avanti
[18:46:28 CEST] <durandal_1707> then ask help on Avanti channel
[18:46:36 CEST] <artao> surely someone here uses a frontend rather than CLI
[18:46:41 CEST] <durandal_1707> nope
[18:46:43 CEST] <artao> durandal_1707 there is no avanti channel
[18:46:46 CEST] <artao> so STFU
[18:46:51 CEST] <artao> if you can't help, then don't
[18:46:53 CEST] <artao> >:(
[18:47:52 CEST] <durandal_1707> artao: do you want to use ffmpeg utility to transpose video or remove metadata?
[18:48:02 CEST] <artao> metadata? no
[18:48:07 CEST] <artao> i want to affect the video itself
[18:48:21 CEST] <artao> outputting a new video, using ffmpeg
[18:48:26 CEST] <durandal_1707> with ffmpeg command or avanti one?
[18:49:00 CEST] <artao> <sigh> go away if you're not going to help, cuz I know that when I say, "Avanti" you're going to say, "then go somewhere else"
[18:49:04 CEST] <artao> bye
[18:49:34 CEST] Action: artao ignores durandal_1707
[18:49:51 CEST] <artao> anyone else care to try, rather than tell me nothing useful?
[18:51:38 CEST] <durandal_1707> looks like this gui is no longer developed at all
[18:52:58 CEST] <sfan5> artao: your attitude is preventing you from getting any help
[18:53:07 CEST] <dl2s4> :)
[18:54:18 CEST] <BtbN> Yeah, like most UIs. The only thing that is well maintained seems to be Handbrake
[18:54:47 CEST] <artao> i've had issues with handbrake making files that Fusion can read
[18:55:00 CEST] <artao> sfan5 he was being an unhelpful dick
[18:55:04 CEST] <artao> i'm not in the mood for that
[18:55:05 CEST] <artao> :\
[18:55:07 CEST] <artao> sry
[18:55:19 CEST] <BtbN> He told you the truth though, no matter if you like it or not...
[18:55:21 CEST] <artao> so he's ignored now. fixed that problem
[18:55:37 CEST] <sfan5> you're as much of a dick as you believe him to be
[18:55:38 CEST] <artao> of the .. what? ... 100+ people here, no one uses Avanti
[18:55:40 CEST] <BtbN> Not going to get any help for third party UIs here other than "use ffmpeg"
[18:55:55 CEST] <sfan5> if you're going to act like this don't be surprised when you get no help
[18:56:03 CEST] <BtbN> I have never eben heard of Avanti, good luck finding any other users.
[18:56:10 CEST] <artao> blah blah ignore sfan5 as well
[18:56:12 CEST] <artao> byeee
[18:56:14 CEST] <sfan5> nice
[18:56:20 CEST] <sfan5> i've always wanted to be put on ignore lists
[18:56:31 CEST] <another> achivement unlocked
[18:56:31 CEST] <artao> LOL
[18:57:06 CEST] <BtbN> Yes, everyone except you is being a dick, that's exactly how that works. Have fun with your Avanti.
[18:57:08 CEST] <another> artao: please put me on your ignore list, too.
[18:57:14 CEST] <artao> lol
[18:57:20 CEST] <another> because with that attitude i won't help you
[18:57:35 CEST] <artao> if you don't use Avanti, then you couldn't help anyhow
[18:57:52 CEST] <artao> i was QUITE clear when I asked that I'm using Avanti
[18:57:54 CEST] <BtbN> If you want help with Avanti, ask the Avanti people.
[18:58:00 CEST] <BtbN> This is #ffmpeg
[18:58:10 CEST] <artao> rather than so "no, i don't use avanti" i got the "answer" of "use the command line"
[18:58:14 CEST] Action: dl2s4 grabs popcorn
[18:58:16 CEST] <durandal_1707> avanti have ability to alter ffmpeg command line and add more parameters, but it is windows only app
[18:58:19 CEST] <artao> AVANTI USES FFMPEG FFS
[18:58:22 CEST] <artao> JESUS FUCK
[18:58:31 CEST] <artao> if you don't use avanti, then why say anything at all
[18:58:34 CEST] <artao> dafuq
[18:58:36 CEST] <sfan5> dl2s4: can i have some?
[18:58:49 CEST] <artao> avanti IS ffmpeg in that regard
[18:58:52 CEST] <dl2s4> sfan5, sure :)
[18:58:56 CEST] <artao> it just makes the gorram CLI for you
[18:59:04 CEST] <artao> good christ
[18:59:11 CEST] Action: another grabs some chocolate
[19:02:53 CEST] <tyll> Hi, I would like to dump the video of a webcam on a system, dump it to stdout and watch it with a second tool that reads from stdin, what could I use for this? I tried https://superuser.com/questions/652032/accessing-webcam-streams-through-shell that creates a stream to stdout but I cannot watch it with mplayer - or vlc -
[19:03:42 CEST] <durandal_1707> tyll: which OS?
[19:03:50 CEST] <tyll> durandal_1707: sorry, Linux
[19:04:00 CEST] <tyll> Fedora to be precise
[19:04:39 CEST] <sfan5> ffmpeg -f video4linux2 ... -i /dev/video0 -pix_fmt yuv420p -f yuv4mpegpipe - | mpv -
[19:04:54 CEST] <sfan5> not sure why you would do this though
[19:05:40 CEST] <durandal_1707> i think he can use avdevice dirrecly with mpv, never tried for v4l2
[19:06:15 CEST] <tyll> sfan5: I want to do ssh fmmpeg ... | mpv -
[19:06:21 CEST] <sfan5> thought so
[19:06:36 CEST] <YokoBR> hi there
[19:06:37 CEST] <sfan5> yuv4mpeg might be a bit too bandwidth intensive for that so you'll want to encode the video
[19:07:46 CEST] <YokoBR> please, I want to create HLS on my server, and maybe use a dynamic text on those videos
[19:07:50 CEST] <YokoBR> how can I do that?
[19:07:58 CEST] <tyll> sfan5: any suggestion about how to encode?
[19:09:17 CEST] <Rikkkardino> Ciao
[19:09:19 CEST] <Rikkkardino> HI
[19:09:57 CEST] <furq> tyll: if this is over lan then y4m should be fine
[19:11:19 CEST] <sfan5> otherwise just use h264
[19:11:21 CEST] <sfan5> $ ssh laptop.lan ffmpeg -loglevel quiet -i /dev/video0 -c:v h264 -f matroska - | mpv -
[19:11:52 CEST] <tuna> Ever gotten a Unhandled exception at 0x00007FFD12463F80 (msvcrt.dll) in RTSPStreamer.exe: 0xC0000006: In page error reading location 0x0000000C01C403F0 (status code 0xC0000022). while using nvenc with hardware input at the call avcodec_send_frame()??
[19:13:15 CEST] <tuna> Also, I make the call in a try/catch
[19:13:23 CEST] <tuna> but it doesnt catch it
[19:13:38 CEST] <Rikkkardino> Come posso installare fFFmpeg su kali linux
[19:13:40 CEST] <tuna> msvcrt.dll is the STD lib it sems
[19:13:40 CEST] <Rikkkardino> ?
[19:14:34 CEST] <furq> Rikkkardino: https://www.johnvansickle.com/ffmpeg/
[19:15:40 CEST] <Rikkkardino> Thx
[19:16:50 CEST] <furq> i should add that to the bot i guess
[19:25:00 CEST] <artao> solved it
[19:25:02 CEST] <artao> l8r
[19:27:59 CEST] <furq> what a nice young man
[19:38:10 CEST] <alexpigment> lol :)
[19:52:14 CEST] <tuna> Anyone know what would cause this using hardware input nvenc? Unhandled exception at 0x00007FFD12463F80 (msvcrt.dll)
[19:52:16 CEST] <dagb> good evening.
[19:52:24 CEST] <tuna> when I call avcodec_send_frame
[19:52:45 CEST] <tuna> (status code 0xC0000022).
[19:52:55 CEST] <tuna> seems maybe like a cuda memory issue??
[19:53:19 CEST] <tuna> https://devtalk.nvidia.com/default/topic/903039/managed-memory-crash/
[19:53:36 CEST] <tuna> Could this be an ffmpeg nvenc bug?
[19:54:01 CEST] <dagb> I am curious if the aptX support in ffmpeg can act as a drop-in replacement for the proprietary libraries.
[19:54:16 CEST] <dagb> So that bluez can make use of that.
[19:54:40 CEST] <JEEB> someone was very interested in implementing it so if you have more samples to test it with that'd be cool
[19:54:52 CEST] <JEEB> I think FATE has some tests
[19:55:41 CEST] <JEEB> ok, seems like not? :< that makes me a sad little person
[19:55:55 CEST] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=tree;f=tests/fate;h=287a98d8022fa78a75f1469f2912a32a945efb7f;hb=HEAD
[19:56:10 CEST] <JEEB> the tests are generally registered here in these makefiles
[20:31:39 CEST] <Fenrirthviti> Anyone familiar with the whole "nvenc active, memory clock decreases" issue?
[21:17:37 CEST] <Reaper_man> hey, anyone here know how to encode with the new AOM codec?
[21:17:57 CEST] <Reaper_man> couldn't find any documentation on it at all
[21:20:01 CEST] <Reaper_man> even tried libaom-av1
[21:26:40 CEST] <alexpigment> reaper_man: have you tried ffmpeg -codecs to make sure the encoder is available?
[21:27:00 CEST] <Reaper_man> I did not know of this command
[21:27:48 CEST] <alexpigment> well, assuming that your copy is compiled with the encoder (it'll have an "E" in the list to the left), then you'll want to get the encoder ptions
[21:28:06 CEST] <alexpigment> which you can do with ffmpeg -h encoder=[encoder name]
[21:29:04 CEST] <Reaper_man> mine is he latest win build
[21:29:09 CEST] <alexpigment> for now, that's probably all the documentation you'll find short of browsing the source code
[21:29:17 CEST] <alexpigment> ok
[21:30:04 CEST] <alexpigment> nightly or stable?
[21:30:35 CEST] <Reaper_man> nightly
[21:30:53 CEST] <Reaper_man> looks preliminary from the -codecs command
[21:31:08 CEST] <alexpigment> yeah, it's certainly a work a progress
[21:31:12 CEST] <Reaper_man> no encode/decode suppord
[21:31:19 CEST] <Reaper_man> just "video" and "lossy"
[21:31:37 CEST] <Reaper_man> need encode and lossless
[21:31:50 CEST] <Reaper_man> I'm hoping it outperforms zmbv
[21:31:50 CEST] <alexpigment> nah, it's in there
[21:32:05 CEST] <alexpigment> try -c:v av1
[21:32:17 CEST] <alexpigment> anyway, "outperforms" is a vague word
[21:32:39 CEST] <klaxa> it's not one-dimensional :)
[21:32:42 CEST] <alexpigment> if outperforms has any consideration for your time, then it's probably not going to outperform many codecs
[21:33:13 CEST] <ChocolateArmpits> it'll outperform your nerves for sure
[21:33:17 CEST] <alexpigment> haha
[21:33:24 CEST] <alexpigment> i'll outperform your electric bill
[21:33:29 CEST] <alexpigment> *it'll
[21:33:32 CEST] <Reaper_man> Unknown encoder 'av1'
[21:33:47 CEST] <alexpigment> did you get this from zeranoe?
[21:34:05 CEST] <Reaper_man> Yes
[21:34:21 CEST] <alexpigment> today?
[21:34:26 CEST] <Reaper_man> Yes
[21:34:34 CEST] <alexpigment> win 64-bit, static?
[21:34:40 CEST] <Reaper_man> Yes
[21:34:48 CEST] <alexpigment> hmmm
[21:35:04 CEST] <alexpigment> ffmpeg.exe -h encoder=av1
[21:35:08 CEST] <alexpigment> did you try that?
[21:35:33 CEST] <Reaper_man> "Codec 'av1' is known to FFmpeg, but no encoders for it are available. FFmpeg might need to be recompiled with additional external libraries."
[21:35:58 CEST] <alexpigment> ok, just out of curiosity, are you running your command line and pointing to the file specifically in your /bin/ folder?
[21:36:35 CEST] <alexpigment> because i remember back in the day i didn't realize that you could have a copy of ffmpeg installed that would be called when you used ffmpeg [blah blah blah] on command line
[21:36:49 CEST] <Reaper_man> it's in my path variable, I'm running from my temp video processing folder
[21:37:08 CEST] <alexpigment> i don't know then
[21:37:11 CEST] <alexpigment> i just downloaded it
[21:37:14 CEST] <alexpigment> it works over here
[21:38:08 CEST] <Reaper_man> ...wait
[21:38:26 CEST] <alexpigment> pointing to the wrong copy still? :)
[21:38:32 CEST] <Reaper_man> it doesn't include that external library doesn't it?
[21:38:46 CEST] <alexpigment> if it's static, it's all built-in
[21:39:23 CEST] <Reaper_man> so the library is built into the exe, then
[21:39:31 CEST] <alexpigment> anyway, does it say --enable-libaom right after --enable-libxvid?
[21:39:40 CEST] <alexpigment> yeah, it should be
[21:40:45 CEST] <Reaper_man> I hav eto get going, but here:
[21:41:08 CEST] <Reaper_man>  built with gcc 7.2.0 (GCC)
[21:41:08 CEST] <Reaper_man>   configuration: --enable-gpl --enable-version3 --enable-sdl2 --enable-bzlib --enable-fontconfig --enable-gnutls --enable-iconv --enable-libass --enable-libbluray --enable-libfreetype --enable-libmp3lame --enable-libopenjpeg --enable-libopus --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libtheora --enable-libtwolame --enable-libvpx --enable-libwavpack --enable-libwebp
[21:41:08 CEST] <Reaper_man> --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libzimg --enable-lzma --enable-zlib --enable-gmp --enable-libvidstab --enable-libvorbis --enable-cuda --enable-cuvid --enable-d3d11va --enable-nvenc --enable-dxva2 --enable-avisynth --enable-libmfx
[21:41:24 CEST] <Reaper_man> bbl
[21:41:51 CEST] <alexpigment> that's not the same build i just downloaded
[21:41:53 CEST] <klaxa> obviously missing the important bit: the version
[21:42:06 CEST] <Fenrirthviti> also probably should use hastebin or something next time :x
[21:42:08 CEST] <alexpigment> still, it would have said built with gcc 7.3.0
[21:42:11 CEST] <alexpigment> haha
[21:42:29 CEST] <klaxa> so much for "latest" and "nightly" i guess? :P
[21:42:43 CEST] <alexpigment> yeah, it's clearly an older version
[21:42:55 CEST] <Reaper_man> well, that's odd
[21:43:07 CEST] <alexpigment> what does the copyright date say?
[21:43:12 CEST] <Reaper_man> and false alarm
[21:43:13 CEST] <alexpigment> and the version, while you're at it
[21:43:15 CEST] <Reaper_man> but this info helps
[21:43:41 CEST] <Reaper_man> ffmpeg version N-88555-g6ea7711532 Copyright (c) 2000-2017 the FFmpeg developers
[21:44:02 CEST] <klaxa> that's so last year
[21:44:02 CEST] <alexpigment> yeah, it should obviously be 2018 there
[21:44:20 CEST] <Reaper_man> ......
[21:44:22 CEST] <Reaper_man> ah
[21:44:28 CEST] <Reaper_man> there's my problem
[21:45:14 CEST] <Reaper_man> I think at one point I thought it be wise to jut throw it in the windows directory while thinking I made a path pointing to it elsewhere
[21:45:17 CEST] <Reaper_man> I'm an idiot
[21:45:57 CEST] <klaxa> don't worry, this happens to the best people :P
[21:46:07 CEST] <alexpigment> no worries. was pretty sure that was what was happening, but yeah, i've been there before
[21:47:22 CEST] <alexpigment> anyway, now that you've got that all sorted, be prepared for blazing-fast sub-1fps encoding speeds! :)
[21:49:29 CEST] <Reaper_man> no lossless option?
[21:50:10 CEST] <Reaper_man> also, where exactly, would I be adding -strict -2?
[21:50:44 CEST] <alexpigment> just anywhere after -i [video file]
[21:53:26 CEST] <alexpigment> specifying a crf seems to work
[21:53:33 CEST] <alexpigment> so -crf 1 would presumably be lossless
[21:53:47 CEST] <alexpigment> er
[21:53:48 CEST] <alexpigment> sorry
[21:53:49 CEST] <alexpigment> -crf 0
[21:54:11 CEST] <Reaper_man> I did "lossless 1"
[21:54:15 CEST] <Reaper_man> which I think works
[21:54:31 CEST] <alexpigment> fair enough
[21:54:42 CEST] <alexpigment> it probably just maps back to crf or vice versa
[21:55:52 CEST] <alexpigment> oh duh; i just did those tests quickly and forgot to specify an encoder. CRF may or may not work with AV1 - i was doing a bad test
[21:55:53 CEST] <Reaper_man>  cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: -1
[21:56:04 CEST] Action: Reaper_man shrugs
[21:56:11 CEST] <Reaper_man> I'm no encoding expert
[21:56:25 CEST] <Reaper_man> I just like lossless video for my videos
[21:56:46 CEST] <klaxa> re-encodes will most likely blow up the filesize
[21:56:50 CEST] <alexpigment> that's noble, but it's not too useful unless you're coming from a lossless source
[21:56:53 CEST] <klaxa> while making it harder to decode
[21:57:05 CEST] <klaxa> so you get the disadvantages of slow decoding
[21:57:07 CEST] <Reaper_man> the source is lossless
[21:57:13 CEST] <alexpigment> like if you're going from lossy > lossless, the file sizes will get stupid
[21:57:15 CEST] <klaxa> with the disadvantages of takes up a lot of disk
[21:57:20 CEST] <klaxa> in that case, go ahead
[21:57:25 CEST] <alexpigment> yeah, fair enough
[21:58:18 CEST] <alexpigment> i do kinda wonder if codecs like this really do much for *true* lossless video
[21:58:31 CEST] <Reaper_man> no idea
[21:58:40 CEST] <Reaper_man> zmbv was designed for it
[21:58:46 CEST] <Reaper_man> and it works pretty damn well
[21:58:49 CEST] <alexpigment> i kinda suspect that most of the optimization is just making low bitrate video perceptually better
[21:59:49 CEST] <Reaper_man> shoudn't a codec like this use up all 8 cores, or am I missing something here?
[22:00:00 CEST] <alexpigment> well, it's an early encoder
[22:00:07 CEST] <alexpigment> so it's going to be using 1
[22:00:15 CEST] <alexpigment> maybe it takes a -threads [number] argument?
[22:00:16 CEST] <alexpigment> not sure
[22:00:35 CEST] <kepstin> iirc, even libvpx (vp9) doesn't support thread autodetection?
[22:00:43 CEST] <kepstin> so you have to manually specify
[22:01:16 CEST] <YokoBR> Guys, I'm thinking about using this script https://gist.githubusercontent.com/mrbar42/ae111731906f958b396f30906004b3fa/raw/85d14d36559e0010346993e779ab73b18d856d49/create-vod-hls.sh to generate hls
[22:01:31 CEST] <YokoBR> How can I use S3 (AWS) to store it?
[22:02:26 CEST] <Reaper_man> no still 1
[22:02:28 CEST] <Reaper_man> boo
[22:02:33 CEST] <alexpigment> yeah, confirmed over here
[22:03:20 CEST] <alexpigment> anyway, i'd say that AVC/H.264 in lossless mode is probably going to be a hell of a lot better at the moment
[22:03:44 CEST] <Reaper_man> I tried that once, loooong long ago
[22:03:45 CEST] <alexpigment> well supported by everything, highly multi-threaded, etc
[22:03:51 CEST] <alexpigment> well, what happened?
[22:04:06 CEST] <Reaper_man> it wasn't truly lossless from my tests
[22:04:13 CEST] <alexpigment> did you do -crf 0?
[22:04:26 CEST] <klaxa> how did you test?
[22:04:27 CEST] <alexpigment> or -q:v 0
[22:04:43 CEST] <klaxa> "it looks like shit" is also acceptable
[22:04:44 CEST] <Reaper_man> color analysis
[22:05:03 CEST] <klaxa> fair enough
[22:05:19 CEST] <alexpigment> is it possible you were encoding in yuv420p?
[22:05:23 CEST] <klaxa> maybe there was a color space conversion?
[22:05:51 CEST] <Reaper_man> shouldn't have produced more colors than what the NES could handle
[22:05:52 CEST] <alexpigment> anyway, i think in 2018, you should be able to encode without any of the downsides you experienced back then
[22:05:53 CEST] <Reaper_man> :P
[22:06:09 CEST] <alexpigment> man, i'd just try again
[22:06:14 CEST] <klaxa> colors will still be different
[22:06:26 CEST] <klaxa> rgb -> yuv444 also is different iirc
[22:06:33 CEST] <Reaper_man> sligh differences in colors isn't a huge deal
[22:06:37 CEST] <klaxa> also because yuv is 16-240 or something?
[22:06:47 CEST] <Reaper_man> shit NES colors aren't exactly "well defined," anyway
[22:06:49 CEST] <alexpigment> i think that depends on the range you specify though, right?
[22:07:25 CEST] <klaxa> you can use yuvj444 i think which uses the full range, but no consumer device will display that "properly" i think
[22:07:30 CEST] <klaxa> or rather yuvj420 even
[22:07:35 CEST] <alexpigment> ffmpeg.exe -i [samplevideo] -c:v libx264 -crf 0 output.mp4
[22:07:53 CEST] <alexpigment> assuming that you are serious about "slight differences in colors isn't a huge deal"
[22:08:03 CEST] <alexpigment> because the differences should be slight imo
[22:09:11 CEST] <TD-Linux> alexpigment, AV1's lossless mode isn't terribly good. it exists mostly because it's easy to add
[22:09:30 CEST] <YokoBR> how can I create a gif from random parts of the video?
[22:09:55 CEST] <YokoBR> Like 4 or 5 parts,  0%, 25%, 50%, 75%, 100%
[22:09:58 CEST] <TD-Linux> (not worse than h.264, but not as good as a specialized codec could be)
[22:11:15 CEST] <alexpigment> TD-Linux: yeah, i kinda expected that. thanks for the confirmation
[22:12:59 CEST] <klaxa> hrm, was about to post a comparison but it's fro 2007
[22:14:17 CEST] <alexpigment> between avc and av1?
[22:15:12 CEST] <klaxa> between all kinds of lossless codecs
[22:15:18 CEST] <klaxa> http://www.compression.ru/video/codec_comparison/lossless_codecs_2007_en.html
[22:15:49 CEST] <klaxa> it's older than 10 years by now, so i wouldn't really say it still applies today?
[22:15:54 CEST] <klaxa> also lacking newer codecs
[22:16:23 CEST] <klaxa> then again, i don't think vp8 and vp9 support lossless and x265 lossless ended up larger in my tests a year or so ago
[22:16:47 CEST] <TD-Linux> vp9 has lossless
[22:16:53 CEST] <TD-Linux> it is more or less the same as av1 lossless
[22:16:53 CEST] <klaxa> ah
[22:22:12 CEST] <Reaper_man> -crf 0 almost does it, but makes barely noticable changes.
[22:23:14 CEST] <klaxa> pretty sure that's the color space conversion
[22:23:26 CEST] <klaxa> if your source is rgb
[22:23:30 CEST] <klaxa> and you are encoding to yuv
[22:23:31 CEST] <iive> would crf use qp=0 ?
[22:25:15 CEST] <alexpigment> pretty sure crf 0 and qp 0 are the same
[22:25:18 CEST] <Reaper_man> if that's the case it's doing some weird rounding to get diferent results on pixes that should be the exact same color
[22:25:20 CEST] <alexpigment> i could be wrong though
[22:25:32 CEST] <Reaper_man> *pixels
[22:25:45 CEST] <alexpigment> hmmm
[22:25:57 CEST] <alexpigment> i have a feeling there's a completely different explanation for this
[22:26:21 CEST] <klaxa> yeah maybe pastebin input and output of your ffmpeg calls
[22:26:23 CEST] <alexpigment> for example, if you have an nvidia card, don't use VLC
[22:26:42 CEST] <Reaper_man> I'm using media player classic
[22:26:49 CEST] <alexpigment> because it'll honor the limited range that's set by default in the nvidia control panel (whereas Windows Media Player will work fine)
[22:27:19 CEST] <alexpigment> yeah, just try good old Windows Media Player. it's usually right because there are many paid professionals ensuring that it's doing the right thing
[22:27:41 CEST] <alexpigment> it also hooks into dxva2 automatically
[22:44:38 CEST] <oliverdain> anyone around to answer a simple (I think) question about setting profiles on an encoder?
[22:44:46 CEST] <oliverdain> (in C, not command line)
[22:50:56 CEST] <klaxa> just ask
[22:51:15 CEST] <klaxa> best with code samples of what you are doing on pastebin
[22:52:32 CEST] <oliverdain> I'm trying to (1) find the set of profiles available on a codec, (2) select from that list using some heuristics and then (3) tell the codec to use my selection.
[22:53:54 CEST] <oliverdain> The problem is that AVCodec::profiles is null until you call avcodec_open2 but the only way I know of to set the profile is to put it in an AVDict and pass that to avcode_open2. So I can make it work by opening twice -- open, find codec and then re-open, but thinking there's probably a way to change the profile on an open codec.
[22:54:20 CEST] <alone-y> hello, is it possible to monitor some pixel with constant coordinate and if it will be some color then do something?
[22:54:45 CEST] <oliverdain> Looks like there's a profile member on the AVCodecContext so thinking maybe I can set it there. Coding that up now but if memory serves it won't switch profiles once opened (I think I tried it once before but not sure)
[22:55:31 CEST] <garoto> YokoBR: http://trac.ffmpeg.org/wiki/Create%20a%20thumbnail%20image%20every%20X%20seconds%20of%20the%20video
[22:55:50 CEST] <jkqxz> Why do you want to pick something when you don't know what you want in advance?  Most codecs with profiles will pick the right thing for whatever your input is automatically, the only reason to set it explicitly is because you have some hard constraint (decoder which only supports a specific profile, say).
[22:56:13 CEST] <klaxa> oliverdain: it's an int, the documentation says you can just set it
[22:56:16 CEST] <klaxa> https://ffmpeg.org/doxygen/trunk/structAVCodecContext.html#a7abe7095de73df98df4895bf9e25fc6b
[22:56:40 CEST] <oliverdain> Thanks klaxa. Will give it a shot.
[22:57:53 CEST] <klaxa> i'm not sure if that's the same profile you set with the options dict though
[22:58:29 CEST] <oliverdain> yeah - that's my memory. Will let you know if it doesn't work.
[22:58:32 CEST] <klaxa> wait... hmm... i haven't used the avcodec api yet really
[22:59:55 CEST] <klaxa> hmm yeah probably is what you want though, the AVCodecContext says what profile the AVCodec should be used with
[23:15:10 CEST] <oliverdain> apparently even after call avcodec_open2 `condec_context->codec->profiles == nullptr`
[23:31:34 CEST] <alone-y> furq u here?
[23:49:51 CEST] <oliverdain> OK. I found the list of profiles. You can get it from AVCodecContext->codec_descriptor->profiles (but not from AVCodec->profiles) but only *after* avcodec_open2. Unfortunately, it appears that you can no longer change the profile after that. Setting AVCodecContext.profile or using av_set_opt on the context's priv_data after the avcode_open2 call doesn't result in an error, but also doesn't change the profile that was used to encode.
[00:00:00 CEST] --- Wed Apr 25 2018