[Ffmpeg-devel-irc] ffmpeg.log.20140623

[burek]

[00:00] <iive> well, if it is just few lines , feel free to ask it right away.
[00:10] <elec29a> should i use cygdrive/d/ or d:/ for prefix cross prefix ?
[00:10] <elec29a> i guess there is the error
[00:12] <elec29a> sorry it try both
[00:20] <elec29a> noone here in teamview ? i couldnt pastebin all the possibilties
[00:38] <elec29a> https://www.youtube.com/watch?v=MlSkiHgA51A just have a happy day (wrong channel maybe)
[04:08] <cjoseph> I just created a ffserver setup that livestreams a webcam. The target client is the web browser. The latency even just on localhost is quite bad, though, ~3s
[04:08] <cjoseph> I saw the StreamingGuide post on latency, but a lot of it seems to target clients
[04:09] <cjoseph> I'm not sure how I could modify the cache configurations of a web client
[04:09] <cjoseph> this is the server conf: http://tny.cz/3b2a4924
[04:09] <cjoseph> if you have any suggestions on config changes to reduce latency, or advice on how to specify the cache for a browser client, please let me know
[04:49] <BtbN> cjoseph, i wouldn't recommend using ffserver. It's unmaintained since a long time, and likely to be removed from future ffmpeg versions.
[05:19] <ac_slater> hey all, I have an MPEG2 video. I want to know what ffmpeg demuxer to use for it. Any clues?
[05:19] <ac_slater> oh nevermind I found it
[05:58] <ac_slater> hmm I guess not. What muxers/demuxers are associated with mpeg2 video  ?
[07:52] <shape> i'm trying to convert mkv to mp4 and I get this message after running ffmpeg -i 720.mkv -vcodec libx264 -acodec libvo_aacenc test.mp4 ...i get "Error while opening encoder for output stream #0:1 - maybe incorrect parameters such as bit_rate, rate, width or height"
[08:05] <diesel420> ffmpeg -i 720.mkv -vcodec copy -acodec copy output.mp4
[08:05] <shape> diesel420: if i copy it, my tv won't play it for some reason
[08:06] <shape> diesel420: I only do this because my smartTV only wants mp4 and AAC
[08:11] <shape> diesel420: The error is actually from the audio Stream #0:1 -> #0:1 (dca -> libvo_aacenc)
[08:14] <shape> diesel420: it seems to be working if i put aac instead of libco_aacenc
[08:14] <shape> libvo_aacenc*
[08:39] <volmatrix> hello everyone, is it possible to encrypt a multicast udp live stream (with ffmpeg of course) ?
[09:09] <decembersoul> Hello, i have a question about the x264. If i chose "-crf 24 -preset fast" or "-crf 24 -preset veryfast" will the quality be the same (account of crf 24) and the result will be differ only in filesize? Or will the quality with -preset veryfast be more poor the with -preset fast?
[09:17] <relaxed> decembersoul: presets control compression
[09:19] <decembersoul> so the compression will not have any effect for the quality? I don't case much about filesize.
[10:18] <horus_> do anyone have  compiled ffmpeg for ios(armv7,armv7s,i386) without mp3 decoding bug, that add distortion to decoded sound?
[11:28] <iamtakingiteasy> hello, i have written a simple aalib video filter for ffmpeg. here is patch: http://paste.eientei.org/show/220/ and example of ascii-fied video: http://www.youtube.com/watch?v=Xd9dFuwVmD8
[11:29] <iamtakingiteasy> any hints how to proceed, what needs to be changed in order to upstream this patch?
[11:32] <brontosaurusrex> iamtakingiteasy, afaik just post that to devel mail-list
[11:34] <c_14> iamtakingiteasy: https://ffmpeg.org/developer.html#Submitting-patches-1
[11:40] <iamtakingiteasy> thanks
[16:22] <termos> can I use ffmpeg filter graph to detect interlacing and if it's interlaced I run the yadif filter?
[16:23] <termos> Without having to run two separate processes, one to detect and another to deinterlace
[16:34] <c_14> I'm not sure that's possible. But you can try using yadif with deint=1 so it'll only deinterlace frames marked as interlaced.
[16:41] <foonix> when encoding with cli, is there a way to set buffer size for output file or somehow force to flush in atleast 4MB chunks?
[16:41] <c_14> -bufsize ?
[16:42] <termos> okey I'll give it a try thanks! I will try to run it on some none-interlaced video and see what happens to the video and the cpu usage.
[16:43] <c_14> You should still require a good amount of cpu, since in order to use a filter you'll need to reencode the video track.
[16:46] <termos> that's, the filter is only running on the AVFrames that I get after decoding
[16:47] <termos> that's true, ...
[16:51] <foonix> c_14: i need that for writing the file to storage not for stream
[16:55] <foonix> but will look more on AVFormatContext AVOptions, few things looks promising
[17:09] <apoc> I just can't get this to work, I'm trying to record via x11grab/pulseaudio but the audio is always async (seems like it gets async over time): http://pastebin.com/hyhcYX3f
[17:10] <apoc> the start timecodes look strange to me, (start: 0.004973 for audio, start: 1403535582.786433) isn't any audio syncronization effort by ffmpeg based on this?
[17:11] <apoc> can I get the pulse audio input to use a unix timestamp as a timecode too?
[17:12] <foonix> chunk_size,packetsize,bufsize doesnt help me, just to mention i write to NFSv4 share
[17:38] <apoc> looks like an old known/ignored bug
[17:38] <apoc> @ my x11grab async issue
[19:20] <SrikanthJessu> Hi Team,
[19:20] <SrikanthJessu> Unable to resolve this error...
[19:20] <SrikanthJessu> >[libx264 @ 0x1639e30] broken ffmpeg default settings detected
[19:20] <SrikanthJessu> >[libx264 @ 0x1639e30] use an encoding preset (e.g. -vpre medium)
[19:20] <SrikanthJessu> >[libx264 @ 0x1639e30] preset usage: -vpre <speed> -vpre <profile>
[19:20] <SrikanthJessu> >[libx264 @ 0x1639e30] speed presets are listed in x264 --help
[19:20] <SrikanthJessu> >[libx264 @ 0x1639e30] profile is optional; x264 defaults to high
[19:20] <SrikanthJessu> >Cannot open video encoder for stream #0
[19:20] <SrikanthJessu> >Error occurred: Operation not permitted
[19:20] <SrikanthJessu> how to set -vpre option programmetically?
[19:20] <SrikanthJessu> options tried..
[19:20] <SrikanthJessu> //libx264-ipod320.ffpreset
[19:20] <SrikanthJessu> av_opt_set(enc_ctx->priv_data, "vprofile", "baseline", 0);
[19:20] <SrikanthJessu> av_opt_set(enc_ctx->priv_data, "level", "30", 0);
[19:20] <c_14> pastebin, please
[19:20] <SrikanthJessu> Sorry, okay.
[19:21] <sacarasc> SrikanthJessu: -vpre was removed a LONG time ago.
[19:21] <sacarasc> Your version is ancient.
[19:21] <sacarasc> Upgrade.
[19:21] <SrikanthJessu> http://pastebin.com/mgiGi0mM
[19:22] <SrikanthJessu> @sacarasc which one ffmpeg or x264?
[19:22] <sacarasc> ffmpeg.
[19:22] <SrikanthJessu> I compiled 2.2.3
[19:23] <sacarasc> The output you gave was not from 2.2.3...
[19:31] <SrikanthJessu> @sacarasc I updated with my ffmpeg configuration (at bottom) http://pastebin.com/WnXGLpWC on 2.2.3. did I miss anything?
[19:32] <sacarasc> If you don't need any of the other codecs, probably not.
[19:32] <SrikanthJessu> I need h264 and mp3 encoders.
[19:32] <SrikanthJessu> due to that I compiled x264 and mp3lame
[19:33] <SrikanthJessu> Used android cross compilation.
[19:36] <SrikanthJessu> @sacarasc added x264 configuration information http://pastebin.com/nLJqapu3 (at bottom)
[19:45] <SrikanthJessu> what could be the possible reasons for "broken ffmpeg default settings detected"?
[19:46] <sacarasc> An ancient ffmpeg.
[19:46] <sacarasc> The defaults work now.
[19:47] <SrikanthJessu> Okay @sacarasc, I will try recompiling ffmpeg once again. Thanks.
[19:48] <sacarasc> You could look at the output of just `ffmpeg` and it will tell you the version number.
[19:51] <SrikanthJessu> @sacarasc I used '--disable-programs' option in configure, so there is no `ffmpeg` executable available. I think so..
[23:36] <ogrgkyle> Hi!  http://pastebin.com/uJksFVfS
[23:36] <ogrgkyle> I have this OpenCV program which takes an image from the command line and outputs the detected faces' X,Y,width,height.
[23:37] <ogrgkyle> I'm trying to modify it so that it will anaylze every frame of a video
[23:37] <ogrgkyle> And I don't know the fastest, most efficient way to do that.  Right now, I have to extract all the frames of a video and put them in a folder, and then analyze those one by one through a .bat file.
[23:38] <ogrgkyle> Also, I read that finding an exact frame is slow.
[23:39] <ogrgkyle> So I wonder if you can help me with this.  Is there a way to feed the frames of a video through OpenCV in my simple script?  Thanks.
[23:42] <c_14> Maybe try looking at: https://code.dyne.org/frei0r/tree/src/filter/facedetect ?
[23:42] <c_14> Not entirely sure how this is FFmpeg related though since you're not using the library.
[23:43] <ogrgkyle> Yeah
[23:43] <ogrgkyle> Well, I think I'll need to learn how to use the library
[23:45] <ogrgkyle> So that code (in the link) is designed for video?
[23:48] <c_14> ye, if you compile ffmpeg with --enable-frei0r (iirc) you can use them directly with libav* as a video filter.
[23:50] <ogrgkyle> You mean if I enable frei0r, FFmpeg itself can do face detection?
[23:51] <c_14> Assuming the filter does what its name implies, yes.
[23:51] <c_14> I've never actually used it, YMMV.
[23:51] <ogrgkyle> very interseting
[23:51] <ogrgkyle> interesting
[23:52] <ogrgkyle> So, if you've never used it, how did you find it?  :)
[23:52] <ogrgkyle> I mean, is this the only one... or are there others?
[23:52] <c_14> I was browsing the FFmpeg configure options one lonely afternoon when I came across the --enable-frei0r option and decided to investigate.
[23:53] <c_14> There are probably others, but I don't know how well they interface with FFmpeg or where you could find them.
[23:53] <ogrgkyle> Very interesting indeed
[23:56] <ogrgkyle> If I build ffmpeg with frei0r, does that include the opencv auxiliary library?  http://www.dyne.org/software/frei0r/
[23:56] <ogrgkyle> That is, if I download the windows binary (with frei0r), does that include the opencv stuff?
[23:58] <c_14> Do you mean the zeranoe binary? If so, I think it does if you get the dev version, but don't quote me on that.
[00:00] --- Tue Jun 24 2014