[Ffmpeg-devel-irc] ffmpeg.log.20161026

[burek]

[00:00:17 CEST] <llogan> i assumed it was a typo
[00:00:43 CEST] <NapoleonWils0n> just off to make some tea
[00:01:29 CEST] <Substring> i made the .asoundrc as suggested. I'm not familiar with alsa, do i need to reboot or restart sthg so that it's used ?
[00:11:17 CEST] <ATField> (test)
[00:18:43 CEST] <Phi> alright folks, can someone explain the 0s
[00:18:44 CEST] <Phi> http://pastie.org/private/adlzhhuinnmn3xwb8jl1q
[00:19:07 CEST] <Phi> I use avcodec_copy_context and avcodec_parameters_copy
[00:19:21 CEST] <Phi> but the RTSP vid and output mp4 vid are different
[00:20:17 CEST] <debianuser> Substring: .asoundrc is read when applications start, so you don't have to reboot, just restarting the apps playing sound should be enough. But reboot will work too, as it'd certainly restart all apps. :)
[00:21:01 CEST] <Substring> debianuser, yeah thanks for the tip ! I don't mind sparing myself from a reboot ;)
[00:21:01 CEST] <llogan> Phi: are you still using Libav's libav* or FFmpeg's libav*?
[00:21:11 CEST] <Phi> ffmpeg's
[00:21:24 CEST] <Phi> I did a make from git.ffmpeg.org yesterday
[00:21:51 CEST] <llogan> ok. well, you're at least in the right channel now.
[00:22:42 CEST] <Phi> maximum sass
[00:23:05 CEST] <Phi> but help would also be cool
[00:23:09 CEST] <Phi> I'm full already
[00:29:27 CEST] <Substring> thank you all for the help; time to sleep. Nite nite everyone
[00:32:21 CEST] <Phi> oyasumi
[00:32:38 CEST] <Phi> so llogan any idea?
[00:32:44 CEST] <Phi> do you want codes?
[00:43:12 CEST] <Phi> hmm
[00:43:29 CEST] <Phi> klaxa: what's with your example?
[00:43:40 CEST] <Phi> it allocates a CodecCtx, but never uses it
[00:44:13 CEST] <klaxa> it does
[00:44:25 CEST] <klaxa> in line 203, it uses codec_ctx->codec
[00:44:42 CEST] <klaxa> i leak it too, yeah i fixed that locally already haven't pushed it yet
[00:44:52 CEST] <Phi> yea, but you could just use the codec on its own
[00:45:08 CEST] <klaxa> from where?
[00:45:26 CEST] <klaxa> AVStream's codec is deprecated is it not?
[00:45:58 CEST] <klaxa> it's not even present in trunk anymore
[00:46:13 CEST] <Phi> avcodec_find_encoder(in_stream->codecpar->codec_id) ?
[00:46:36 CEST] <klaxa> oh that's how you use it
[00:46:40 CEST] <klaxa> thanks
[00:47:02 CEST] <Phi> [citation needed]
[00:47:08 CEST] <Phi> [many citations and tests needed]
[00:47:14 CEST] <Phi> [who knows what they keep in codec_id]
[00:48:24 CEST] <Phi> and I haven't got my code working so if your way works, roll with it :p
[00:59:51 CEST] <debkad> furq: mission completed, thanks all works!
[01:58:27 CEST] <Phi> welp they came for him
[02:17:19 CEST] <marcurling> Please remind me the other way to pass '-vcodec copy'
[02:18:48 CEST] <Phi> that doesn't make any sense
[02:19:09 CEST] <Phi> is av_log_set_level busted? I'm getting AV_LOG_TRACE messages when I've set it to AV_LOG_INFO
[02:21:17 CEST] <Phi> don't see how it can be busted
[02:22:50 CEST] <Phi> weird that custom callbacks are meant to filter the level themselves
[02:23:00 CEST] <Phi> seems like there's pointless verbosity in there
[02:27:04 CEST] <SchrodingersScat> marcurling: does -vc work?
[02:27:16 CEST] <furq> -c:v copy
[02:27:28 CEST] <SchrodingersScat> i see
[03:15:37 CEST] <marcurling> ty
[03:16:30 CEST] <marcurling> and can you -c:v:a copy ? furq
[03:16:57 CEST] <klaxa> and what would that do?
[03:17:16 CEST] <klaxa> you can do -c:a copy -c:v copy or if you only have audio and video anyway just do -c copy
[03:17:34 CEST] <klaxa> -c copy will copy all streams
[03:17:57 CEST] <marcurling> Ha, nice, ty more
[03:18:24 CEST] <furq> -c copy will copy all streams but not all of them will necessarily end up in the output file
[03:18:38 CEST] <furq> if you have multiple streams of the same type then use -c copy -map 0
[03:19:55 CEST] <marcurling> Got it all!
[04:15:00 CEST] <k_sze[work]> Is there a way to show what x264 options (or preset) were used to encode a file, maybe using ffprobe?
[04:15:15 CEST] <k_sze[work]> (given that the video file was encoded through ffmpeg)
[04:18:04 CEST] <marcurling> Well, 'ffmpeg -i filename' gives you info but I'm not sure it will give all you want
[04:23:47 CEST] <Phi> k_sze[work]: ffprobe will tell you
[04:26:37 CEST] <k_sze[work]> Phi: I tried ffprobe -show_streams
[04:28:49 CEST] <k_sze[work]> but I couldn't see anything related to x264 options.
[04:29:05 CEST] <k_sze[work]> no qp, no crf, no preset
[04:30:42 CEST] <k_sze[work]> I mean, quality-wise, I see level=40, and that's about it.
[04:32:43 CEST] <furq> k_sze[work]: mediainfo
[04:33:38 CEST] <marcurling> hmmm, don't know about ffprobe (you could try encoding a file with param you want to see and then probe it) but I'm pretty sure MediaInfo will tell you (free but not related to ffmpeg)
[04:34:06 CEST] <furq> ffprobe didn't have it last time i checked
[05:07:59 CEST] <Phi> ffprobe did for me
[05:08:30 CEST] <Phi> but x264 options are metadata, so they may not be written
[05:12:04 CEST] <furq> http://vpaste.net/pX9cd
[05:12:06 CEST] <furq> it shows that?
[05:14:22 CEST] <furq> it's also not metadata, it's written into the SEI NAL in the video bitstream
[05:14:47 CEST] <furq> at least if you're using the cli tools you have to patch x264 to get rid of it
[05:16:01 CEST] <Phi> I keep getting messages about unrecognised SEI from my RTSP stream
[05:17:51 CEST] <furq> you can probably ignore it
[05:18:19 CEST] <Phi> probably
[05:18:31 CEST] <Phi> since they fill up my log I have to drop the messages
[05:18:42 CEST] <Phi> and RTSP is massively verbose in ffmpeg for some reason
[05:18:56 CEST] <furq> i assume it's some application-specific data that ffmpeg can't do anything with
[05:19:57 CEST] <Phi> mm
[05:20:04 CEST] <Phi> 1) receive frame from RTSP
[05:20:09 CEST] <Phi> 2) pass frame to decoder
[05:20:20 CEST] <Phi> 3) read frame from decoder
[05:20:29 CEST] <Phi> 4) pass frame to encoder
[05:20:35 CEST] <Phi> 5) read frame from encoder
[05:20:46 CEST] <Phi> 6) pass frame to file container
[05:20:50 CEST] <Phi> ._.
[05:22:37 CEST] <furq> http://vpaste.net/Bmgta
[05:22:41 CEST] <furq> i guess you can get it without mediainfo
[05:36:10 CEST] <Phi> furq, to clarify, to write a frame to an AVFormatContext I have to encode it manually?
[05:37:40 CEST] <k_sze[work]> crap
[05:37:50 CEST] <k_sze[work]> looks like MediaInfo doesn't understand .nut files.
[05:41:40 CEST] <Phi> neither do I
[06:01:35 CEST] <k_sze[work]> Is it expected that the x264 encoding preset also affects decoding performance?
[06:03:35 CEST] <k_sze[work]> (putting aside the fact that superfast would produce a smaller file than ultrafast, hence possibly requiring less I/O for decoding)
[06:16:42 CEST] <furq> k_sze[work]: to an extent, yes
[06:16:56 CEST] <furq> obviously cabac vs cavlc will make a big difference
[06:17:16 CEST] <k_sze[work]> I see.
[06:17:24 CEST] <furq> bframes and refs will make a difference as well
[06:19:22 CEST] <Phi> hmm
[14:19:44 CEST] <Polochon_street> Hi everyone! Just wanted to know, what's the difference between libavresample and libswresample?
[14:24:26 CEST] <nonex86> Polochon_street: http://ffmpeg.org/pipermail/ffmpeg-devel/2012-April/123746.html
[14:24:56 CEST] <Polochon_street> nonex86: I've found it, but 2012 is 4 years from now. Is it still valid?
[14:29:30 CEST] <nonex86> not sure exactly, but if you open https://ffmpeg.org/documentation.html
[14:29:47 CEST] <nonex86> you dont find libavresample in libraries documentation
[14:30:19 CEST] <nonex86> so looks like nothing changes since that post
[14:30:35 CEST] <Polochon_street> okay, thanks :)
[14:30:47 CEST] <JEEB> it's there and changes that come from libav get merged into it. and you can build it if you enable-avresample in the config
[14:30:59 CEST] <JEEB> but the one that's actively developed in FFmpeg is swresample
[14:32:18 CEST] <Polochon_street> because I need to maintain some package that uses libswresample for both ubuntu 16.04 and arch, and since there has been a change in ffmpeg api not so long ago, it isn't in Ubuntu 16.04
[14:32:30 CEST] <Polochon_street> so I'm considering using avresample for ubuntu 16.04 instead
[14:35:03 CEST] <JEEB> the APIs should be quite similar
[14:35:37 CEST] <Polochon_street> but now I think I'm just going to do something like « if(LIBSWRESAMPLE_VERSION < 2) » and adapt the code accordingly
[14:35:40 CEST] <Polochon_street> thank you!
[15:37:21 CEST] <ATField> If the goal is to make a small audio-video fragment using the video-stream of INPUT.mkv and the audio-stream of INPUT.ogg  without re-encoding either stream, then is this a proper instruction formula for that task?   		
[15:37:23 CEST] <ATField> ffmpeg -ss [STARTPOINT] -i "INPUT.mkv" -t [DURATION] -ss [STARTPOINT] -i "INPUT.ogg" -t [DURATION] -map 0:0 -map 1:0 -c copy output.mkv
[15:41:12 CEST] <arin0v> how to enable frame when i capture the screen?
[16:33:58 CEST] <NapoleonWils0n> hi all
[16:34:07 CEST] <NapoleonWils0n> i see automatic bitstream filtering in the changelog for version 3
[16:34:37 CEST] <NapoleonWils0n> when i run fmpeg -bsfs i dont see auto bitstream filters in 3.1.5
[16:35:07 CEST] <NapoleonWils0n> is it a compile time option
[16:36:50 CEST] <NapoleonWils0n> its not an option listed in the config options for my build of ffmpeg
[16:40:04 CEST] <NapoleonWils0n> looked at git build or my distro and dont see any compile time flags for auto bitsream filters
[16:41:18 CEST] <DHE> ./configure --list-bsfs
[16:42:50 CEST] <NapoleonWils0n> DHE im using arch and dont see any options in the pkgbuild
[16:51:20 CEST] <NapoleonWils0n> so i could download ffmpeg git head and run ./configure --list-bsfs and see if anything shows up
[16:54:49 CEST] <DHE> you could, but to save time here's what I get: http://pastebin.com/rnrXn2xk
[16:56:51 CEST] <Mad7Scientist> How can I watch a downloaded youtube video with separate audio and video? Run two player instances at the same time? I do know that I can encode them with ffmpeg and probably use the -copy codec
[16:58:04 CEST] <DHE> depends on the player. mpv has an --audio-file option to select which file contains the audio track
[16:58:23 CEST] <bencoh> mplayer has something like that as well ... I'd suspect vlc has one too
[16:58:24 CEST] <Mad7Scientist> thanks
[16:58:34 CEST] <DHE> correction, mpv
[16:58:40 CEST] <DHE> sorry, old habit
[16:58:45 CEST] <Mad7Scientist> with ffmpeg I tried .mp4 .avi .mov and finally .vob would contain both streams
[16:59:17 CEST] <DHE> $ ffmpeg -i video.mp4 -i audio.mp3 -c copy combined.mkv
[16:59:54 CEST] <bencoh> you might need to play with -map 0 as well
[17:00:37 CEST] <Mad7Scientist> I did -vcodec copy -acodec copy
[17:01:48 CEST] <NapoleonWils0n> DHE i see the dump_header bsfs options in git version, but no dump_headers in my build
[17:02:05 CEST] <Mad7Scientist> the .vob file didn't work. mplayer said that the video dimensions were aut of range
[17:02:13 CEST] <NapoleonWils0n> wondering what auto bitstream filters would be listed as
[17:03:33 CEST] <Mad7Scientist> yeah I do need -map
[17:03:48 CEST] <Mad7Scientist> It did this and I have no audio:
[17:03:49 CEST] <Mad7Scientist> Stream mapping:
[17:03:49 CEST] <Mad7Scientist>   Stream #0:0 -> #0:0 (copy)
[17:03:49 CEST] <Mad7Scientist>   Stream #1:0 -> #0:1 (copy)
[17:05:22 CEST] <Mad7Scientist> I see videos on file download sites but most of them don't have surround sound or subtitles or multiple languages like the original DVD had
[17:06:34 CEST] <DHE> bencoh: by default the files genuinely contain only audio and only video, my command works. if that's not the case, then yes some use of -map may become necessary
[17:06:46 CEST] <Mad7Scientist> Oh the 48,000Hz audio that I downloaded can't be played. Cannot find codec 'libopus' in libavcodec...
[17:07:04 CEST] <NapoleonWils0n> Mad7Scientist have you looked at using Kodi which has lots of high quality videos
[17:07:12 CEST] <Mad7Scientist> I haven't
[17:07:33 CEST] <NapoleonWils0n> i have some ffmpeg scripts that can record pretty much any video from Kodi
[17:07:42 CEST] <NapoleonWils0n> https://github.com/NapoleonWils0n/kodi-playercorefactory
[17:07:50 CEST] <NapoleonWils0n> https://www.youtube.com/channel/UCriRR_CzOny-akXyk1R-oDQ
[17:08:16 CEST] <NapoleonWils0n> the scripts automatically restart recording if the stream cuts out
[17:08:44 CEST] <NapoleonWils0n> works on windows, mac and linux
[17:09:14 CEST] <bencoh> DHE: doesn't ffmpeg default to at most 1 audio and 1 video?
[17:09:51 CEST] <NapoleonWils0n> MadScientist7 have a look at Kodi and the Phoenix addon
[17:09:53 CEST] <NapoleonWils0n> https://seo-michael.co.uk/how-to-install-phoenix-for-xbmc/
[17:12:42 CEST] <Mad7Scientist> Is Kodi a media streaming Internet service?
[17:13:25 CEST] <NapoleonWils0n> Md7Scientist its a Media player that has thousands of addons to stream videos from loads of place on the internet
[17:13:31 CEST] <NapoleonWils0n> https://kodi.tv/
[17:13:57 CEST] <NapoleonWils0n> works on windows, mac, linux, android and mobile devices
[17:15:07 CEST] <NapoleonWils0n> you can find pretty much any film, tv or live tv streams you want
[17:42:07 CEST] <ATField> (reposting the question) If the goal is to make a small audio-video fragment using the video-stream of INPUT.mkv and the audio-stream of INPUT.ogg  without re-encoding either stream, then is this a proper instruction formula for that task?   		ffmpeg -ss [STARTPOINT] -i "INPUT.mkv" -t [DURATION] -ss [STARTPOINT] -i "INPUT.ogg" -t [DURATION] -map 0:0 -map 1:0 -c copy output.mkv
[17:42:27 CEST] <BtbN> It's not very likely you can remux mkv to ogg
[17:42:48 CEST] <BtbN> or vice versa
[17:43:00 CEST] <ATField> what do you mean? And you can replace .ogg with .ac3, because Ive tried with an .ac3 audio stream as well
[17:44:49 CEST] <ATField> The problem is that it works (cutting out a fragment without re-encoding that wont have any audio syncing problems) when I first make a new .mkv container with the video-stream from original video-file and the audio-stream from the .ac3 stream. But if I do all that in one go, it ends up with messed up audio-stream in one way or another.
[17:44:57 CEST] <ATField> So I thought maybe the command formula I was using was wrong.
[20:07:57 CEST] <pomaranc> hello, does anyone know if this has been solved: https://trac.ffmpeg.org/ticket/1679
[20:09:24 CEST] <DHE> the state of the ticket implies No...
[20:11:26 CEST] <pomaranc> I know
[20:12:05 CEST] <pomaranc> but here http://forum.videohelp.com/threads/352391-FFMPEG-Ability-to-identify-progressive-segmented-frame-material-in-h-264 someone says that it's been solved
[20:13:46 CEST] <pomaranc> and I don't have any sample to test with
[20:27:36 CEST] <pomaranc> I looked at the source code and it sets interlaced_frame according to ct_type, so in theory it should work
[20:28:09 CEST] <pomaranc> but my h264 knowledge is limited
[00:00:00 CEST] --- Thu Oct 27 2016