[Ffmpeg-devel-irc] ffmpeg.log.20150912

Sun Sep 13 02:05:01 CEST 2015

[00:03:11 CEST] <manfromafar> ugh. Trying to to compile ffmpeg with libdcadec support any help?
[00:07:05 CEST] <c_14> I've done it before. Wasn't that hard.
[00:07:07 CEST] <c_14> What's the problem?
[00:08:59 CEST] <manfromafar> just trying to get it installed is all. do I need to compile it seperatly from ffmpeg
[00:09:04 CEST] <tlhiv_laptop> i've tried crossfading as BtbN's link suggested, and i'm getting the following error regard "black" --> http://pastebin.tlhiv.org/H9P5ZoY1
[00:09:52 CEST] <manfromafar> c_14, getting the ERROR: dcadec not found using pkg-config
[00:09:52 CEST] <manfromafar>  error on --enable-libdcadec
[00:09:54 CEST] <tlhiv_laptop> nevermind ... old version
[00:10:21 CEST] <c_14> manfromafar: Have you built dcadec and installed it somewhere?
[00:11:14 CEST] <manfromafar> ah not yet didn't know if ffmpeg would build it or not wish the git apge was a little better
[00:13:11 CEST] <manfromafar> hmmm now to configure dcadec to buildin the ffmpeg_build/bin fodler
[00:13:35 CEST] Action: manfromafar knows this hsould be simple. Is new to compiling 
[00:14:11 CEST] <c_14> make PREFIX=$HOME/ffmpeg_build
[00:16:08 CEST] <manfromafar> perfect! thanks I was just reading the make file just didnt know how the flag was denoted
[00:17:17 CEST] <manfromafar> thanks c_14 worked out great. compiling now
[00:29:58 CEST] <tlhiv_laptop> c_14: why is my crossfade not including the audio from my individual videos?
[00:30:12 CEST] <c_14> I won't know unless you show me your command.
[00:30:36 CEST] <tlhiv_laptop> previous pastbin
[00:30:46 CEST] <tlhiv_laptop> http://pastebin.tlhiv.org/H9P5ZoY1
[00:31:26 CEST] <c_14> Because you're not doing anything with the audio? In fact you're not mapping it.
[00:31:36 CEST] <c_14> I assume you want to crossfade it as well?
[00:32:05 CEST] <tlhiv_laptop> well i'd like to try it both ways to see :-)
[00:32:14 CEST] <tlhiv_laptop> both concatenating and crossfading
[00:32:29 CEST] <c_14> there's the concat filter and the acrossfade filter
[00:33:05 CEST] <tlhiv_laptop> i'm not familiar enough with ffmpeg's mapping to construct it on my own :-(
[00:33:55 CEST] <c_14> [0:a][1:a]acrossfade=values[a] -map [a] and [0:a][1:a]concat=n=2:v=0:a=1[a] -map [a]
[00:34:00 CEST] <c_14> check the docs for the acrossfade options you want
[00:34:22 CEST] <c_14> those are filterchains you add to your filter_complex filtergraph
[00:34:28 CEST] <c_14> and the -map goes next to the other map
[00:34:40 CEST] <tlhiv_laptop> ok let me try those ... thanks
[00:35:53 CEST] <tlhiv_laptop> no such filter acrossfade
[00:40:06 CEST] <llogan> once again, your ffmpeg is old.
[00:40:49 CEST] <llogan> see http://johnvansickle.com/ffmpeg/
[00:42:24 CEST] <tlhiv_laptop> llogan: 2.7.2 is too old?
[00:43:43 CEST] <llogan> your pastebin indicated that you were using ffmpeg version 1.0.8
[00:44:59 CEST] <tlhiv_laptop> llogan: yeah ... i have two versions installed ... i fixed my command a few minutes ago to use 2.7.2
[00:45:12 CEST] <tlhiv_laptop> the 2.7.2 is statically built into the current directory
[00:48:58 CEST] <tlhiv_laptop> [0:a][1:a]concat=n=2:v=0:a=1[a]; is generating a bunch of [Parsed_concat_8 @ 0x1c50c00] Buffer queue overflow, dropping.
[00:49:16 CEST] <tlhiv_laptop> and [0:a][1:a]acrossfade=values[a] is an invalid filter
[00:49:20 CEST] <tlhiv_laptop> :-(
[00:57:49 CEST] <tlhiv_laptop> http://pastebin.tlhiv.org/rg65N8Me
[00:58:19 CEST] <llogan> reread the message from fflogger
[00:59:56 CEST] <tlhiv_laptop> http://pastebin.tlhiv.org/4Mx1gQOn
[01:00:25 CEST] <llogan> your ffmpeg is too old for acrossfade
[01:00:42 CEST] <tlhiv_laptop> what version is need?
[01:01:02 CEST] <llogan> users are recommended to use a build from current git master
[01:41:32 CEST] <votz> ffprobe -show_frames shows I-frames every ~2.5 seconds, but when seeking in the file (-ss before the input), it seeks with a granularity twice that, ~5s. Is ffmpeg seeking IDR frames instead of all I-frames? If so, how can ffprobe's output be parsed to determine just the frames ffmpeg will seek to?
[01:54:00 CEST] <pzich> it should be able to seek to any frame, are you using copy?
[01:54:06 CEST] <votz> pzich: Yes, stream copy.
[01:56:08 CEST] <votz> pzich: the gory details http://pastie.org/private/dggbdifojjrzgbwfyilq
[01:57:09 CEST] <votz> ffprobe (which generated the top data) found 8 I-frames, but ffmpeg's seeking skips the one at 3.136467s
[02:17:35 CEST] <votz> pzich: Any idea?
[02:26:57 CEST] <Roxis> Can anyone help-me?
[02:28:21 CEST] Action: Roxis slaps c_14 around a bit with a large fishbot
[02:30:10 CEST] <Roxis> hello?
[02:30:32 CEST] <votz> Roxis: In general on IRC, ask your question, don't ask to ask
[02:31:04 CEST] <Roxis> why the command ffmpeg -i input-file.mov -vcodec mpeg4 -s 320x180 -qscale:v 20 -brand 3gp6 \     -vf 'scale=iw*sar:ih,pad=max(iw\,ih*(16/9)):ow/(16/9):(ow-iw)/2:(oh-ih)/2' \     -acodec libfaac -ar 22050 -ac 1 -qscale:a 70 output-file.3gp not working?
[03:01:31 CEST] <MINIMAN10000> Alright so I was curious of a breakdown on this "Stream #0:0: Audio: opus (libopus) ([255][255][255][255] / 0xFFFFFFFF), 48000 Hz, mono, s16, 24 kb/s" Alright so it is a opus stream, 48000 hertz, mono, 24 kilobits a second but what is the s16?
[03:04:20 CEST] <MINIMAN10000> oh it's signed 16 bits per sample
[03:06:29 CEST] <MINIMAN10000> that's weird is it because it is a opus that it isn't s16le? because i see no such thing as s16 in the list of formats
[03:57:25 CEST] <sOlar_eClips-E> hi
[04:02:32 CEST] <sOlar_eClips-E> how to configure ffmpeg to get good quality but low file size?
[04:05:07 CEST] <sOlar_eClips-E> this is what I got http://paste.ubuntu.com/12356411/
[06:53:55 CEST] <hendry> what's the frame rate here? http://ix.io/kLm
[10:42:01 CEST] <fling> Hello. I just received a bunch of videos recorded from a webcamera with ffmpeg and _without_ an rtc. The video/sound is extremely out of sync. Is it possible to fix?
[11:49:04 CEST] <ClearSky> What prerequisite knowledge and skills are needed to be able to work on ffmpeg as a developer?
[11:54:42 CEST] <c_14> You need a decent understanding of C as well as knowledge on what you're adding/fixing/modifying. Both can be learned, so mainly just time.
[11:59:48 CEST] <c_14> If you want a place to start, you can look at open tickets/feature requests on track or places in the code marked with TODO or FIXME
[12:01:40 CEST] <ClearSky> c_14: Thanks, but I'm a complete newbie. I have just started learning programming.
[12:05:27 CEST] <c_14> Diving into the ffmpeg codebase without any prior programming experience is possible, but probably pushing it a bit. You might want to go through something like learningCthehardway first and then write your own program in C to learn the basics. Especially when it comes to pointers and memory allocation.
[12:09:01 CEST] <ClearSky> Besides C, what areas of math are needed for developers who work on ffmpeg encoding libraries? Is knowledge of DSP required?
[12:09:49 CEST] <fling> Does this look good for capturing? -> http://pastebin.calculate-linux.ru/ru/show/11250
[12:09:51 CEST] <c_14> For the most part no, unless you're working on implementing a new codec/filter or writing ASM for an existing codec/filter.
[12:09:55 CEST] <fling> c_14: Hello.
[12:10:32 CEST] <c_14> fling: assuming the -ss 5 works, sure
[12:10:45 CEST] <fling> c_14: why would not it?
[12:11:12 CEST] <c_14> Well, works for you anyway. Not sure what usecase would require not recording the first 5 seconds.
[12:12:30 CEST] <fling> c_14: 1. there is a framerate drop in the begginning of the video for some unknown reason (probably bad webcam hardware)
[12:12:41 CEST] <fling> 2. beginning of the video looks a lot brighter
[12:13:01 CEST] <c_14> mhm, looks fine to me then
[12:13:09 CEST] <fling> 3. I think it prevents the video from being too out of sync with audio otoh it is still a bit out of sync.
[12:13:30 CEST] <fling> also idk which input format to choose -> http://pastebin.calculate-linux.ru/ru/show/11251
[12:13:42 CEST] <fling> and should not I reencode? Is mjpeg good enough?
[12:14:41 CEST] <c_14> Good enough depends on you. It's compressed and the quality shouldn't be terrible.
[12:15:11 CEST] <fling> It is not terrible, yes. Also the camera gives better resolution*framerate for mjpeg than for uncompressed.
[12:15:57 CEST] <c_14> Probably bandwidth reasons.
[12:17:04 CEST] <fling> Sure.
[12:17:35 CEST] <fling> Also I figured the reason of a bunch of videos are terribly out of sync with audio
[12:18:19 CEST] <fling> When I capture audio from alsa which is usb_audio found on the webcam at 44100 the recorded audio then plays at 3x rate for some unknown reason.
[12:18:42 CEST] <fling> when I capture at 48000 everything is fine and getting only a single alsa Xrun at the beginning _before_ the ss
[12:19:08 CEST] <fling> this xrun probably causing the little sync problem but not really noticeable.
[12:19:16 CEST] <fling> Is not it possible to fix the files? :P
[12:19:30 CEST] <c_14> If the offset is always the same, you could use -itsoffset
[12:20:51 CEST] <fling> c_14: the problem is not in offset& playback speed is different. I only hear few clicks at the beginning of the video and the rest of it is silent.
[12:22:22 CEST] <c_14> That doesn't sound like playback speed, but rather like corrupt audio.
[12:23:42 CEST] <fling> Right, because of the wrong playback speed of the source it was muxed the wrong way and got damaged probably& I will try copying audio off the container&
[12:27:46 CEST] <fling> c_14: the audio sounds damaged. It is missing like 2/3 of the samples :P Really clicky.
[12:31:00 CEST] <c_14> I'm going to guess alsa didn't like trying to record at that samplerate.
[12:31:12 CEST] <c_14> And then started doing weird things.
[12:31:23 CEST] <fling> Right. I will slap the one recorded this :P
[12:31:44 CEST] <fling> c_14: also this caused the video to be captured at the wrong speed.
[12:32:14 CEST] <fling> c_14: is it possible to fix the video speed setting the hard fps?
[12:32:23 CEST] <fling> it should run exactly at 15 fps afaik
[12:33:21 CEST] <c_14> setpts=N/(15*TB)
[12:33:46 CEST] <fling> umm?
[12:34:14 CEST] <fling> ohh got it
[12:34:29 CEST] Action: fling discovered this in -filter:v
[12:35:43 CEST] <fling> Another question is about multiangle. Is it simple to mux together when I have all the angles in separate files?
[12:41:21 CEST] <c_14> What do you mean?
[12:43:03 CEST] <durandal_1707> ClearSky: depends on what you want to work
[12:44:01 CEST] <fling> c_14: I have multiple videos of the same thing shot from different angles. I want to mux the videos into a single multiangle video.
[12:45:05 CEST] <c_14> I can tell you how to mux a series of videos to create one file with multiple video tracks, but I don't know what your player software expects a multiangle video to look like.
[13:53:16 CEST] <fling> c_14: It is mpv, idk too, will try experimenting& Anyway I will probably able to remux it anytime.
[14:17:09 CEST] <fling> c_14: Filtergraph 'setpts=N/(15*TB)' was defined for video output stream 0:0 but codec copy was selected. Filtering and streamcopy cannot be used together.
[14:18:24 CEST] <c_14> yep
[14:18:59 CEST] <fling> The problem is video playing slower hmm hmmmm
[14:30:14 CEST] <tlhiv_laptop>  /join #perl
[14:30:26 CEST] <tlhiv_laptop> oops :-/
[14:47:44 CEST] <fling> Got you!
[15:23:14 CEST] <feliwir_> When is the meeting?
[15:23:38 CEST] <durandal_1707> next year
[15:23:40 CEST] <Loriker> today, 15:00 @ UTC
[15:24:14 CEST] <feliwir_> What is this in CET?
[15:24:30 CEST] <Loriker> next year
[15:24:38 CEST] <feliwir_> Wut
[15:24:42 CEST] <Loriker> ;)
[15:25:04 CEST] <feliwir_> I dont know UTC timezone :(
[15:25:22 CEST] <durandal_1707> not on this channel
[15:25:43 CEST] <durandal_1707> Cet is utc+2
[15:26:09 CEST] <feliwir_> So meeting starts in 90 minutes?
[15:26:37 CEST] <durandal_1707> in north Korea
[15:26:52 CEST] <feliwir_> -.-
[15:27:03 CEST] <feliwir_> Whats the starttime in CET?
[15:27:24 CEST] <durandal_1707> 17 I guess
[15:27:50 CEST] <feliwir_> Thanks
[15:32:43 CEST] <Loriker> yes, 17:00 @ CEST
[16:42:52 CEST] <fling> Trying to encode with libx264
[16:43:27 CEST] <fling> But http://pastebin.calculate-linux.ru/ru/show/11252
[16:43:57 CEST] <fling> But all the threads running on a single core -> & -c:v libx264 -threads 6 -crf 23 -tune grain -preset slow &
[16:48:09 CEST] <c_14> fling: can you pastebin the ouptut?
[17:10:11 CEST] <fling> c_14: script -> http://pastebin.calculate-linux.ru/ru/show/11253
[17:10:15 CEST] <fling> htop -> http://dpaste.com/09P5WCK
[17:10:21 CEST] <fling> output -> http://dpaste.com/040EP1Z
[17:12:42 CEST] Action: fling is trying to downgrade x264 and ffmpeg
[17:21:17 CEST] <fling> c_14: looks like a kernel issue btw.
[17:33:48 CEST] <fling> c_14: I'm experiencing this -> https://forums.gentoo.org/viewtopic-p-7774966.html
[17:34:00 CEST] Action: fling is rebuilding kernel with numa support enabled. What if it will help? :P
[18:10:02 CEST] <JanCantCompileWi> Hi folks, I'm compiling FFmpeg under Windows using Msys2, I get this error: gcc: error: unrecognized command line option -implib:libavutil/avutil.lib
[18:11:37 CEST] <JanCantCompileWi> I hope someone can point me in the right directions.   My setup is: I installed Msys2 32bit, grabbed tools manually using pacman -S, it builds but fails at avutil.lib. I'm using Visual Studio link.exe and lib.exe.
[19:11:48 CEST] <votz> ffprobe -show_frames shows an I-frame every ~2.5 seconds (http://pastie.org/private/w1nqjvjqta57zaumnv708g), but when seeking in the file with ffmpeg (-ss before -i), ffmpeg seeks with a granularity twice that: ~5s. Does ffmpeg seek to IDR frames instead of all I-frames? If so, how can ffprobe's output be parsed to determine just the frames ffmpeg will seek to?
[20:46:01 CEST] <fred1807> any proven method to grab the m3u8 files form public open streaming? (hsl stream type)
[20:55:14 CEST] <fling> c_14: fixed the single core issue changing some kernel bits -> http://pastebin.calculate-linux.ru/ru/show/11258
[20:55:24 CEST] <fling> But this change broke alsa!
[20:55:44 CEST] <fling> this gives me high pitched clicky sound -> ffmpeg -f alsa -ac 1 -ar 48000 -i default:CARD=U0x46d0x809 -f wav - | aplay
[20:55:46 CEST] <fred1807> uh
[20:56:04 CEST] <fling> with a lot of these messages appearing on every click -> underrun!!! (at least 0,000 ms long)rate= 256.0kbits/s
[20:57:27 CEST] <votz> fred1807: He's not talking to you.
[20:58:05 CEST] <votz> And there won't be a one-size-fits-all 'method' to determine the m3u8 for a stream. The m3u8 could be hosted anywhere.
[20:58:40 CEST] <votz> So, if there's a solution, it will be in the form of some library/snippet that has handlers for the most common sites and/or streaming servers
[21:02:38 CEST] <klaxa> fred1807: you can just pass the link to the m3u8 file to ffmpeg and it will read it as an hls stream
[21:02:54 CEST] <klaxa> or do you mean "how do i find the url of the m3u8"?
[21:03:09 CEST] <votz> klaxa: I believe he's asking the latter.
[21:03:11 CEST] <klaxa> monitoring network requests in my browser works for me
[21:03:30 CEST] <klaxa> search for GET /[something]/something.m3u8
[21:11:25 CEST] <fred1807> votz helped me with manually finding the m3u8 based on the mp4 snipets url that I can find using chrome's network analyzer... ok   But I will like to create a script for this... so I can click ./play.sh and my raspberry will grab the m3u8 and start to play it (or send to ffmpeg it I want to record too).   This is the live event: http://www.worldsurfleague.com/events/2015/mct/1281/hurley-pro-at-trestles   ..  So I wonder if I could use wget or something l
[21:11:25 CEST] <fred1807> that
[21:13:36 CEST] <fred1807> who knows even create a plugin for https://github.com/chrippa/livestreamer
[21:14:21 CEST] <votz> fred1807: No. Again, the m3u8 could be hosted anywhere, so the logic to create the final URL could be anywhere: in the HTML source, in some JS file, constructed dynamically by JS, constructed dynamically by the flash player, etc
[21:14:26 CEST] <votz> There won't be an easy one-size-fits-all solution.
[21:14:47 CEST] <fling> How to shift audio input for 256kbit? Audio is captured slightly out of sync because of the single xrun at the beginning of each encoding. The desync is 256/768 seconds ;P
[21:15:00 CEST] <votz> Could run a headless browser and monitor network requests, ala klaxa's suggestion.
[21:15:25 CEST] <votz> http://phantomjs.org/
[21:41:11 CEST] <tlhiv_laptop> c_14 & llogan: thank you for all of your help ... i've been able to implement a pretty complex filter (no pun intended) of fade in/out among an arbitrary amount of input videos thanks to a Perl script I wrote
[22:02:13 CEST] <fred1807> votz: I am playing around with phantomjs... amazing tool
[22:06:01 CEST] <votz> fred1807: It's handy
[00:00:00 CEST] --- Sun Sep 13 2015