[Ffmpeg-devel-irc] ffmpeg.log.20130507

burek burek021 at gmail.com
Wed May 8 02:05:01 CEST 2013


[04:09] <Giri> Hello
[04:09] <Giri> How are you
[04:10] <Giri> I want to cross compile the ffmpeg source code arm-linux
[04:10] <Giri> which is on imx35 and arm-none-linux-gnueabi complier
[04:11] <Giri> please tel me the ./configure command
[04:11] <Giri> i could able to compile it for the host
[04:11] <Giri> but i am want to compile it and use it for the targe
[04:11] <Giri> please help me
[06:09] <hunterp> Hey:  What does this mean: ld: error: cannot find -lvorbisenc
[08:24] <vdm> hi
[08:24] <vdm> I have video frame raw data in the PIX_FMT_YUYV422 format and need to save it to the JPEG file using LIBJPEG library. With swscale library i convert the PIX_FMT_YUYV422 frame to the PIX_FMT_RGB24 to save it as JPEG. I have analyzed YUYV422 (http://i.imgur.com/Oa0whCn.jpg) and RGB24 (http://i.imgur.com/kb1UHTX.jpg) buffers using YUVTools software - it shows both buffers correctly. The problem is that saved JPEG image is distorted (http://i.imgur.
[08:24] <vdm> com/bT5NR7B.jpg). Here is the function which save PIX_FMT_RGB24 format buffer to the JPEG file, using LIBJPEG library: (http://pastebin.com/K3NdJE8X). I saw that the same distorted effect happens if i choose YUV444 format for the RGB24 buffer in the YUVTools (http://i.imgur.com/ooMgrtH.jpg). Any ideas, why the buffer saved with LIBJPEG is distorted?
[08:29] <IanWizard-Cloud> I'm trying to compress an mp3 (by stripping to mono, dropping sample rate, etc), but each time, it comes out bigger than it went in.
[08:30] <IanWizard-Cloud> I compressed one with audacity, and got it down to ~300Kb, but with ffmpeg (avconv), it ends up larger.  Even when using the compressed version, it comes out as 1.7Mb
[08:30] <IanWizard-Cloud> My cmdline is some variation of:  avconv -i input.mp3 -ac 1 -ab 16000 output.mp3
[08:37] <Mavrik> avconv isn't really ffmpeg now is it? :)
[08:37] <Mavrik> and I very much doubt the mp3 encoder will want to encode your audio to 16kb/s
[08:37] <IanWizard-Cloud> Mavrik: that was my (possibly flawed) understanding.
[08:38] <IanWizard-Cloud> I wanted to use wav, but that ended up with a flat 8mb file.
[08:39] <IanWizard-Cloud> admittedly, I don't have the experience here that I should, and to a large degree, am winging it.
[08:43] <vdm> Mavrik, maybe you know, is it possible with ffmpeg to save RGB24 buffer or file to the JPEG?
[08:45] <Mavrik> vdm, should be
[08:45] <Mavrik> IanWizard-Cloud, well.
[08:45] <Mavrik> let's start at the beginning
[08:50] <IanWizard-Cloud> Mavrik: http://paste2.org/n7UFt3WJ
[08:50] <IanWizard-Cloud> The notice there is what led me believe that avconv was replacing ffmpeg
[08:52] <Mavrik> IanWizard-Cloud, yeah, that's outdated and misleading.
[08:52] <Mavrik> IanWizard-Cloud, do note that maybe some things we tell you here won't work on avconv
[08:52] <Mavrik> and we're not avconv support.
[08:53] <Mavrik> IanWizard-Cloud, um, you're trying to recompress that file to even less bitrate?
[08:53] <IanWizard-Cloud> Mavrik: The file I used there is one that I've already mucked about with in Audacity
[08:54] <IanWizard-Cloud> I should have used a virgin file probably.
[08:54] <Mavrik> IanWizard-Cloud, well, that file is such a low qualith
[08:54] <Mavrik> that I'm surprised it still resembles anything at all
[08:54] <Mavrik> going lower than that with mp3....
[08:54] <Mavrik> not really recommended :)
[08:58] <IanWizard-Cloud> Mavrik: here's some progress, and a real file.  Just used the first song tab complete brought me to :P http://paste2.org/eHXbxh2e
[08:59] <IanWizard-Cloud> It does get it slightly smaller here, which is some progress, but I must be doing something wrong.  Is this the best I can expect?
[09:00] <IanWizard-Cloud> Audio is, admittedly, not my field, but with Audacity I got that amazing grace mp3 down to 300k, with out too much degradation.
[09:00] <Mavrik> huh?
[09:00] <Mavrik> you must be half deaf or have broken speakers if you don't hear degradation at 16kHz sampling rate O.o
[09:01] <Mavrik> most decent quality sound uses 44100Hz - 48000Hz and is compressed to 192kB/s+
[09:01] <Mavrik> IanWizard-Cloud, what exactly is your goal here?
[09:01] <IanWizard-Cloud> no, sorry, the first track, amazing_grace.mp3, sounds worse than the original, but it's still reasonable
[09:01] <Mavrik> why such needed compression? have you thought of better formats?
[09:02] <Mavrik> woult HE-AACv2 fit better to your needs?
[09:06] <IanWizard-Cloud> Mavrik: I'm trying to compress it on the fly, depending on a users connection speed.  And trying to get decent browser compatability, so sadly, it's a bit of a mess.
[09:07] <IanWizard-Cloud> Compounded by the fact that I don't know nearly enough about the pros and cons of the various codecs
[09:07] <Mavrik> IanWizard-Cloud, yeah, educating yourself will help :)
[09:07] <Mavrik> IanWizard-Cloud, for low-bitrate streaming the HE-AACv2 (also called HE-AAC+) would be the best
[09:07] <Mavrik> since you get practically free stereo
[09:07] <IanWizard-Cloud> Mavrik: heh, trying to, but also trying to get it done quickly to keep boss happy :(
[09:08] <Mavrik> and it sounds noticably better and very low bitrates
[09:08] <IanWizard-Cloud> I remember when i did this for nobble reasons, and not a paycheck :(
[09:08] <Mavrik> but please, keep samplerate high
[09:08] <Mavrik> or you'll destroy your sound
[09:08] <Mavrik> so keep it at least 22050
[09:08] <Mavrik> try something in terms of 22050, 32kb/s, mono with SBR on
[09:08] <Mavrik> encoded to AAC high profile
[09:09] <Mavrik> and go no lower
[09:09] <IanWizard-Cloud> ok.
[09:09] <Mavrik> it's really pointless to compress to 200kb audio
[09:09] <IanWizard-Cloud> I know -ar 22050
[09:09] <IanWizard-Cloud> the rest is lost :P
[09:09] <Mavrik> people have fast connections, don't rape their ears
[09:09] <IanWizard-Cloud> but RTFM I will
[09:09] <IanWizard-Cloud> :)
[09:09] <Mavrik> your clients won't be happy ;)
[09:09] <Mavrik> IanWizard-Cloud, libfdk_aac is the keyword
[09:09] <Mavrik> and updating from old avconv to something new is a must ;)
[09:10] <IanWizard-Cloud> IDK why it's outdated.  I may just compile it myself.
[09:10] <IanWizard-Cloud> Or find an oft-updated repo.
[09:10] <Mavrik> IanWizard-Cloud, or grab a static pre-compiled build with libfdk built :)
[09:13] <IanWizard-Cloud> Mavrik: downloading now.
[09:13] <IanWizard-Cloud> Thanks for taking the time.
[09:13] <IanWizard-Cloud> I'll let you know how I manage :)
[10:52] <Zeeflo> ive been noticing that ffmpeg doesnt consume all the force of the cpu.. It consumes all threads in the cores, but only at.. around 7-80% per thread..
[10:52] <Zeeflo> is there an option to make ffmpeg utiliza all cpu power?
[10:52] <Zeeflo> utilize*
[11:21] <elkng> don't you think its strange ?
[11:24] <elkng> no
[12:02] <keyzs> little offtopic but, was planning to convert mov to avi, seen some converters super, format factory, media coder, whats the best way and tool to do this via win and mac? is there any video channel on freenode?
[12:52] <renevolution> Hi there, i'm trying to connect to an icecast/shoutcast server from ffmpeg. As far as i understand i "just" need to set the correct http headers for the connection. Has anyone tried to do this?
[13:20] <iMath>   I use the following python code to split a FLV video file into a set of parts ,when finished ,only the first part video can be played ,the other parts are corrupted.I wonder why and Is there some correct ways to split video files  http://stackoverflow.com/questions/16417598/use-python-to-split-a-video-file-into-a-set-of-parts
[15:13] <renevolution> Does anybody know if it is possible to stream to an icecast2 server using ffmpeg?
[15:17] <stf> where can i add the libdir to ./configure? I have installed libfaac instead of system folder into a seperated one.
[15:19] <Yulth> Hi everyone! I'm having troubles transcoding FLAC (even MP3) to 64kbps HE-AAC. Although the quality of the resultant HE-AAC file is high enough, some distortions in the basses are present, and I don't understand why this is happening....
[15:19] <Yulth> Does anyone have some experience?
[15:21] <klaxa> stf: LDFLAGS="-L/path/to/lib/dir"
[15:21] <klaxa> Yulth: what encoder are you using?
[15:21] <Yulth> sorry..., I'm using libfdk_aac
[15:21] <klaxa> if you are not using libfdk-aac, try it
[15:21] <klaxa> mmh... no idea then :/
[15:22] <Yulth> exactly: -acodec libfdk_aac -profile:a aac_he
[15:24] <Yulth> When transcoding from mp3 to he-aac, some distortions are supposed to will happen, but when transcoding from high-quality FLAC files..., conversion must be nearly "perfect"
[15:24] <Yulth> and these distortions in the basses still happening...
[15:25] <JEEB> uhh, HE-AAC is not magic
[15:25] <stf> klaxa: thx seems to work :-)
[15:25] <JEEB> there will be some distortions naturally, but you might want to try enabling afterburner first
[15:26] <JEEB> basically -afterburner 1
[15:26] <JEEB> with fdk-aac
[15:27] <Yulth> afterburner?
[15:27] <JEEB> it's a setting in fdk-aac
[15:27] <stf> klaxa: hm i was to uncaution
[15:27] <klaxa> hm?
[15:27] <JEEB> slower encoding, more quality is how the documentation notes it. Not that it should be *slow* .
[15:28] <JEEB> Anyways, basically HE-AAC's whole point is to start eliminating stuff that is harder to compress, and that does indeed lead to audible differences, whether or not you perceive that as "distortions" as you listen to the audio is a whole separate discussion.
[15:28] <Yulth> where Can I find more information about this option? In the ffmpeg manual?
[15:29] <Bor0> I'm trying to implement a "buffering" state to my player. so I added a new integer in the VideoState struct, and here's what I do: is->buffering = 1; av_read_frame(...); is->buffering = 0; but this always shows that is->buffering is 0, even when it's buffering. what am I doing wrong?
[15:29] <JEEB> Yulth, http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/libfdk-aacenc.c;h=e3992e1dbe4232b00507111931b1be0331f49916;hb=HEAD#l44
[15:30] <Yulth> Should this option be specified during compilation time?
[15:30] <Yulth> I'm on FreeBSD
[15:31] <Yulth> options "afterburner" is new in version 1.2?
[15:32] <JEEB> I think it's there as long as you have fdk-aac built in
[15:32] <stf> have i to link 'LDFLAGS="-L/path/to/dir"' to the lib dir or is it enought to give the dir above this folders?
[15:32] <JEEB> it should also be shown in fdk-aac's help
[15:32] <JEEB> and no, it's not that new, it just needs fdk-aac to be built in
[15:33] <JEEB> (and used, naturally)
[15:33] <Yulth> ok, I'm going to investigate it. Thank you!! :)
[16:10] <stf> how to use --extra-cflags right?
[16:11] <JEEB> --extra-cflags="-I/wat/is/this/i/dont/even/"
[16:11] <JEEB> for example
[16:11] <stf> hm okay thx
[16:15] <Yulth> JEEB: -afterburner 1 fixed the problem with the basses, but in general terms, the audio quality has been decreased
[16:16] <JEEB> Yulth, that's HE-AAC. It's made for relative low bandwidth audio and it is in no way trying to be transparent
[16:17] <Yulth> what pitty!
[16:18] <Yulth> however, at the same bandwidth OPUS is amazingly brilliant...
[16:20] <JEEB> could try the HE-AACv2 profile of course, I think you were using the normal he-aac profile
[16:20] <Yulth> yes
[16:20] <JEEB> but in general, HE-AAC was first officially publicized in '03, and HE-AACv2 was in '06
[16:20] <JEEB> you can compare that to Opus
[16:20] <Yulth> but generally, at 64kbps HE-AAC works better than HE-AAC,v2
[16:22] <JEEB> well, in any case use whatever you need. I generally use LC-AAC and just push out more rate if I need to get closer to transparency, and need something supported by certain hardware decoders.
[16:29] <Yulth> JEEB: How LC-AAC could be specified as profile in ffmpeg?
[16:31] <JEEB> the usual way? Although you could just raise rate and not specify a profile and it should use LC :P
[16:32] <JEEB> it should show which profile got used during encoding
[16:42] <Yulth> ams ok, its true :)
[17:32] <elkng> http://imagebin.org/index.php?mode=image&id=256816 why its ends on number 7 ? and why 7 is in red ?
[17:34] <smj> Why is -f alsa -i pulse always 48000Hz even when PulseAudio is running at 44100Hz?
[17:35] <smj> Can I fix that to save some CPU and audio quality?
[17:35] <Mavrik> smj, alsa probably doing some conversion
[17:35] <Mavrik> is audio speed good?
[17:35] <smj> yes
[17:36] <Mavrik> you can always try setting put samplerate (-ar before -i) or resample later (-ar after -i)
[17:36] <smj> could that work?
[18:41] <vdm> [swscaler @ 003638E0] Warning: data is not aligned! This can lead to a speedloss - i have this error using sws_scale method. How to fix it?
[18:42] <tmatth> vdm: allocate your buffer with av_malloc?
[18:42] <elkng> how to convert gif to video ?
[18:42] <elkng> can ffmpeg do it ?
[18:43] <ubitux> just like any other format
[18:43] <vdm> tmatth, i'll try
[18:46] <vdm> tmatth, thanks that was the solution.
[18:46] <tmatth> vdm: np
[18:51] <QT-BOSS> why does ffmpeg take so long to encode for mp4 adobe, x264 but for IPhone mp4, it encodes 1.5gb in 14 mints ?
[18:51] <QT-BOSS> is there any way to make IPhone mp4 work on adobe flash ? so i dont spend lots of time on encoding ?
[18:55] <QT-BOSS> ok
[18:57] <QT-BOSS> here for IPhone, which encodes really fast http://pastie.org/7814074
[18:57] <QT-BOSS> how to make that work on adobe flash player too >? but keep the encoding speed ?
[18:59] <durandal_1707> QT-BOSS: that is not full uncut output, without it i can't help, but just play lottery
[19:01] <QT-BOSS> i don't have any problems with the ffmpeg, both cli command lines work, im trying to figure how to make the pastie i just gave work on adobe flash player
[19:02] <QT-BOSS> ffmpeg version http://pastie.org/7814105
[19:17] <durandal_1707> QT-BOSS: that is still incomplete and thus useless output
[19:18] <durandal_1707> QT-BOSS: have you read faq?
[19:18] <QT-BOSS> which one sorry
[19:18] <QT-BOSS> in the ffmpeg faq or the ffmpeg website ?
[19:19] <durandal_1707> faq on website
[19:20] <QT-BOSS> ok im reading @http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[19:20] <durandal_1707> that is not faq
[19:22] <durandal_1707> but last paragraph may be relevant for you
[20:34] <Jofironses> Hello, I was trying to execute this command: 'avconv -i video.mp4 -acodec copy -vcodec copy -map 0:0 -map 0:1 -map 0:2 -i English.ass -map 1:0 -scodec copy output.mkv' but it gives me this error ' Application provided invalid, non monotonically increasing dts to muxer in stream 0: 8500 >= 8500
[20:34] <Jofironses> av_interleaved_write_frame(): Invalid argument'. Does any one got a clue what the problem could be?
[20:38] <Jofironses> I should add that if I execute without the (.ass) it works just fine, that is 'avconv -i video.mp4 -acodec copy -vcodec copy -map 0:0 -map 0:1 -map 0:2 output.mkv' It works just fine
[21:48] <tonsofpcs> hi, how can I tell ffmpeg that an input stream is raw pcm 24bps 48kHz stereo?
[21:51] <tonsofpcs> pcms24le is the audio format but how do I tell it that?  trying to pass it to -f says 'unknown input or output format'
[21:51] <durandal_1707> -formats
[21:51] <tonsofpcs> oh, wait, it's just s24le!
[21:51] <tonsofpcs> I may have it working.  Thanks durandal_1707 :)
[21:51] <niemand> Can I say rtmpdump to don't save the received data? without /dev/null, because I want to use rtmpdump on Windows
[21:53] <durandal_1707> Jofironses: you are not using ffmpeg, this channel is for ffmpeg and not avconv/Libav fork
[21:55] <klaxa> niemand: http://stackoverflow.com/questions/313111/dev-null-in-windows
[21:56] <klaxa> also, i compiled ffmpeg and installed it, but pkg-config doesn't recognize it because there are no .pc files, do i generate them somehow or what?
[21:57] <niemand> klaxa, thanks!
[22:02] <klaxa> k nvm figured it out
[22:17] <benbro> I have RTMP stream
[22:17] <benbro> what encoding should I use to be able to stream it to adnroid and ios html5
[22:21] <Mavrik> HLS probably.
[22:30] <Jofironses> durandal_1707: Ok, but whenever I run ffmpeg directly it tells me: *** THIS PROGRAM IS DEPRECATED ***
[22:30] <Jofironses> This program is only provided for compatibility and will be removed in a future release. Please use avconv instead.
[22:31] <durandal11707> Jofironses: because you are using ffmpeg from Libav
[22:31] <JEEBsv> that's within the libav project that debian and gentoo now use by default, and the ffmpeg binary was later removed from libav
[22:32] <Jofironses> durandal_1707: I see, probably something about ubuntu apt-get. I will try compiling it then
[22:32] <JEEBsv> basically the binary to use with libav-using distros is 'avconv'
[22:32] <JEEBsv> but yes, building ffmpeg from source should do it as well :)
[22:32] <Jofironses> k :0
[22:32] <Jofironses> k:)
[23:15] <benbro> Mavrik: thanks. I'll check HLS
[23:22] <Jofironses> Just to report back with my problem: I had no problems using a similar command using compiled ffmpeg :)
[23:42] <IanWizard-Cloud> Mavrik: I don't think I ever told you last night, but thank you for the help.  Everything's running smoothly now :)
[00:00] --- Wed May  8 2013


More information about the Ffmpeg-devel-irc mailing list