[Ffmpeg-devel-irc] ffmpeg.log.20150830

burek burek021 at gmail.com
Mon Aug 31 02:05:01 CEST 2015

[01:08:06 CEST] <fred1807> Could anyone kindle help find what is the difference between two small video files? I tried ffprobe... there is nothing more I can do to understand why 1 video freezes my raspberry pi and other dont
[03:22:29 CEST] <Abbott> Is there a support channel for ID3v2?
[04:57:57 CEST] <Evidlo> Are there any voice compression codecs for ffmpeg?
[05:37:00 CEST] <chungy> Like speex or opus?
[08:38:05 CEST] <Fyr> how to find out what version of ffmpeg is installed?
[08:38:22 CEST] <Fyr> it responds on --version in a weird way.
[08:39:40 CEST] <pzich> it's just -version
[08:40:16 CEST] <Fyr> =)
[08:40:23 CEST] <Fyr> why --help, but -version?
[08:40:49 CEST] <pzich> looks like -help also works
[08:41:07 CEST] <Fyr> for 2.7.2 - it doesn't.
[08:41:09 CEST] <pzich> -h, -?, -help, --help [arg]
[08:54:47 CEST] <chungy> yeah ffmpeg uses X-style args
[09:03:27 CEST] <frenda> What is the 420p in this: -pix_fmt yuv420p?
[09:03:56 CEST] <frenda> ffmpeg -video_size 1920x1080 -framerate 50 -f x11grab -i :0,0 -c:v libx265 -pix_fmt yuv420p output.mp4
[09:04:04 CEST] <pzich> https://en.wikipedia.org/wiki/YUV#Y.27UV420p_.28and_Y.27V12_or_YV12.29_to_RGB888_conversion
[09:06:45 CEST] <frenda> Ah
[09:10:57 CEST] <frenda> What's wrong w/ this command:
[09:11:06 CEST] <frenda> ffmpeg -video_size 1920x1080 -framerate 25 -f x11grab -i :0.0+0,0 -c:v libx265 -pix_fmt yuv420p output.mp4
[09:11:48 CEST] <frenda> 1. I does not stoped by pressing `q` key!
[09:11:59 CEST] <frenda> 2. Th captured video is damaged!
[10:44:49 CEST] <Mavrik> x265 for realtime capture?
[10:44:51 CEST] <Mavrik> Why.
[11:13:10 CEST] <Polochon_street> Hi! Does someone know with what flags I need to compile ffmpeg in order to debug programs with -fsanitize=address
[11:30:57 CEST] <DHE> you probably want --extra-flags=-fsanitize=address
[11:31:04 CEST] <DHE> (I usually use valgrind though)
[11:33:07 CEST] <DHE> *extra-cflags
[11:33:29 CEST] <DHE> do experiment a bit. compiler flags can have all sorts of side-effects
[12:09:51 CEST] <TikityTik> I'm having issues like, [h264 @ 0x7f5f7c0086c0] Missing reference picture
[12:09:56 CEST] <TikityTik> [h264 @ 0x7f5f7c0086c0] decode_slice_header error
[12:10:04 CEST] <TikityTik> from a video I encoded with libx264
[17:08:44 CEST] <Polochon_street> Hm, I'm trying to debug memory leaks, and I installed ffmpeg with debug symbols (finally!) but I all have is this: http://sprunge.us/PdCE
[17:09:00 CEST] <Polochon_street> is there a mean to see where the av_malloc() took place?
[17:09:34 CEST] <JEEB> is that valgrind?
[17:09:44 CEST] <JEEB> IIRC it should also output the stack trace
[17:10:02 CEST] <JEEB> if it's not valgrind, then I deeply recommend it
[17:10:16 CEST] <Polochon_street> it's the -fsanitize=address option of gcc
[17:10:25 CEST] <JEEB> ok, so probably less useful
[17:10:34 CEST] <JEEB> try running your code under valgrind next time
[17:10:37 CEST] <Polochon_street> what options should I pass to valgrind in order to have all the stack trace?
[17:10:47 CEST] <c_14> valgrind --leak-check=full
[17:10:56 CEST] <Polochon_street> thanks
[17:10:58 CEST] <JEEB> yeah, that was it
[17:11:30 CEST] <JEEB> I used to do valgrind for things that have children etc so I used to use
[17:11:31 CEST] <JEEB> --leak-check=full --track-origins=yes --trace-children=yes
[17:11:58 CEST] <Polochon_street> okay, thanks. Let's see!
[17:13:24 CEST] <Polochon_street> ah-ha!
[17:13:54 CEST] <Polochon_street> http://pastebin.archlinux.fr/1489347
[17:13:55 CEST] <Polochon_street> found it!
[17:17:35 CEST] <JEEB> Polochon_street: now wonder how fun it is to find that stuff /without/ the ability to use valgrind
[17:20:03 CEST] <Polochon_street> (I don't want to ._.)
[17:20:53 CEST] <JEEB> I do a lot of development on windows, no valgrind there
[17:20:55 CEST] <Polochon_street> okay, so after I avformat_open_input'd an avformatctx, apart from freeing it with avformat_close_input(), should I do something else?
[17:21:18 CEST] <Polochon_street> heh, that's why I really don't want to work on windows
[17:21:47 CEST] <JEEB> yeah, that's why most of my stuff can compile on lunix and windows
[17:22:01 CEST] <JEEB> so I can test it out on both
[17:22:07 CEST] <JEEB> (and use valgrind on the lunix side)
[17:22:36 CEST] <JEEB> Polochon_street: https://github.com/jeeb/matroska_thumbnails/blob/master/src/matroska_thumbnailer.cpp#L346
[17:22:43 CEST] <JEEB> looks like I also called avformat_free_context()
[17:25:37 CEST] <Polochon_street> hm, that didn't solve my problem, but it can't hurt either. Thanks!
[17:33:01 CEST] <Polochon_street> okay, I'm almost there, it seems that freeing an avformat is not so simple
[17:33:02 CEST] <Polochon_street> http://pastebin.archlinux.fr/1489348
[17:33:27 CEST] <Polochon_street> I'm calling avformat_close_input(&pformatCtx) and avformat_free_context(pFormatCtx)
[18:18:46 CEST] <Polochon_street> JEEB: You may want to edit your code, I checked, and avformat_close_input() already performs an avformat_free_context() (cf https://www.ffmpeg.org/doxygen/2.5/libavformat_2utils_8c_source.html)
[18:19:44 CEST] <JEEB> that might or might not have been true before :) but sure
[18:19:50 CEST] <JEEB> that code I haven't touched since ~2013 or so
[18:20:13 CEST] <JEEB> just to see how hard it would be to poke the Windows explorer thumbnails system
[18:30:21 CEST] <Polochon_street> yeah you're right, sorry. Now I'm trying to understand why, despite an av_freep(&s->internal) inside the avformat_free_context(), I'm still having this leak
[22:10:18 CEST] <AlexanderSh> Hello! I'm decoding udp multicast stream use ffmpeg. Source: Hardware encoder, 1080i, GOP structure is IBBP, I frame each 6 frames. But I have the lag about 30 frames. How to reduce latency?
[23:40:47 CEST] <sarkyniin> hey
[23:41:09 CEST] <sarkyniin> how do I replace an mkv file's subtitle track?
[23:55:15 CEST] <KarlFranz> Use mkvtools for that.
[00:00:00 CEST] --- Mon Aug 31 2015

More information about the Ffmpeg-devel-irc mailing list