[Ffmpeg-devel-irc] ffmpeg.log.20131001

burek burek021 at gmail.com
Fri Oct 4 16:52:04 CEST 2013


[00:03] <elBradford> I'm trying to build ffmpeg with librtmp for Android, but I get a bunch of "undefined reference" errors for various RTMP functions.
[00:28] <peanuter> any tip/tricks to avoiding this?
[00:28] <peanuter> whenever i convert avi to ogg i am getting audio out of sync
[00:29] <peanuter> various avi's
[00:29] <peanuter> i edit them later to 'resync' the audio
[00:29] <peanuter> until to find after 20 or 30 minutes of play
[00:29] <peanuter> they lose sync again
[00:29] <peanuter> worse and worse as the movie progresses
[01:37] <smjd> peanuter: I would try using -af aresample=async=10000
[01:41] <Niatross> I am decoding a TrueHD stream to an AC-3 stream. It looks as if the TrueHD stream is 32-bit signed-integer. Am I seeing this correctly?
[01:41] <Niatross> Stream #0:2(eng): Audio: truehd, 48000 Hz, 5.1(side), s32
[01:43] <Niatross> The AC-3 stream looks like it is being encoded with 32-bit floating point. Am I seeing this correctly?
[01:43] <Niatross> Stream #0:0(eng): Audio: ac3, 48000 Hz, 5.1(side), fltp, 640 kb/s
[01:45] <Niatross> Am I right or am I not seeing this correctly? Just trying to understand the verbosity of ffmpeg
[01:51] <Niatross> I type  "ffmpeg -h decoder=truehd" and it doesn't give me what "ffmpeg" defaults to when decoding a TrueHD stream. It doesn't display a "Supported sample formats" when I type "ffmpeg -h decoder=truehd".
[01:52] <Niatross> Anyone?
[01:54] <Niatross> Everyone is in bed overseas. LOL!
[01:54] <anon__> Niatross: It's probably just 24bit. ffmpeg does that with flac and wavpack too
[01:56] <Niatross> what does the "s32" mean at the end of the second line (above)?
[01:57] <Niatross> signed-integer 32???
[01:59] <anon__> yes, any 24bit audiofile will default to pcm_s32le
[02:01] <Niatross> so it takes the TrueHD 24-bit file and turns it into 32 bit signed integer?
[02:02] <Niatross> Then the AC-3 encoder turns 32 bit signed integer into 32-bit floating-point
[02:02] <Niatross> and the AC-3 file is created?
[02:03] <anon__> no, it will just default to that if "-acodec pcm_s24le" isn't specified, and lossy audio doesn't have a bitdepth
[02:04] <Niatross> is it ok if i just let it default to 32 bit signed integer?
[02:05] <Niatross> that's why it turns into floating point because lossy doesn't have a bit depth, correct?
[02:10] <anon__> I don't know if ffmpeg pads with 0's or dithers when increasing the bitdepth, but you should keep it at 24bits to be safe; I'm not sure if ac3 defaults to 32f, but going to floating point would be preferable because you wouldn't add quantization noise
[02:14] <Niatross> is their a deeper verbose switch to show the detail of what ffmpeg is doing that will allow me to see the transcode process between TrueHD and
[02:14] <Niatross> a
[02:14] <Niatross> AC-3?
[02:15] <Niatross> Is there a deeper verbose switch to show the detail of what ffmpeg is doing that will allow me to see the transcode process between TrueHD and AC-3?
[02:15] <anon__> is your source a blu-ray?
[02:16] <Niatross> An MKV created from my Blu-ray disc
[02:17] <anon__> I was going to say that all THD blu-rays also have an AC-3 stream
[02:17] <anon__> why do you need the audio in AC3?
[02:19] <Niatross> I know&I'm just trying to understand the choices ffmpeg makes when using fixed-point and floating-point to transcode TrueHD to AC-3
[02:22] <Niatross> I'll have to look around the internet to find out if  "ffmpeg" defaults to 32-bit signed integer and 32-bit floating point to transcode a TrueHD track into an AC-3 track.
[02:27] <anon__> shouldn't make a difference quality wise
[02:27] <Niatross> i know&i'm just trying to understand this stuff
[02:32] <defaultro> ffmpeg rocks! I just made my first 4k sample clip - http://www.youtube.com/watch?v=uzNy9_Chx-g :)
[04:13] <WilliamWS> hi
[04:14] <WilliamWS> there is any asr_decoder on ffmpeg?
[04:15] <WilliamWS> asf_deoder*
[05:00] <kriskropd> I could've sworn I saw an option to use x11grab to select a window to record from - was I just dreaming?
[05:12] <klaxa> kriskropd: probably was a script
[06:39] <Da_Coder> Hello everyone :D
[06:41] <Da_Coder> I am looking to do a livestream/recording java application, but i need to get Java to use FFmpeg. I have found a wrapper, but have never used a wrapper before, can anyone help me on how to use this?
[06:57] <Max-Powerz> i like 50 cent as a wrapper
[06:58] <Da_Coder> Sorry, i was not specific: FMJ for Java + FFmpeg, I am trying to have it capture the screen, and stream to a .mp4
[07:00] <Max-Powerz> best thing to do is wait for a java geek to show up as i would not know where to start on java plus its 6am
[07:02] <Da_Coder> Ah, well i have got what i wanted FFmpeg to do in Bash code, and i can link the source file too, but i want to make a GUI in java, and have FFmpeg capture the video and output to a stream .flv
[08:30] <skyroveRR> Hello, I'd like to convert a high quality 720p avi file into a low quality 480p avi file, what might be the appropriate command to do that?
[08:31] <skyroveRR> Uh sorry, I meant an mp4 file, not an avi file..
[08:44] <Da_Coder> I made a stack overflow query for my question.
[08:44] <Da_Coder> bye
[15:28] <matthias_arch> morning, my issue is that on my client in vlc media player it says cannot pre fill buffer. I'm unsing arch linux on my server and want to stream with this code: ffmpeg -f x11grab -s 1400x1050 -r 30 -i :0.0+0,0 -f mpegts udp://192.168.178.238:1234    ..238 is my server and on the client i type in vlc udp://192.168.178.238:1234
[15:29] <matthias_arch> Soulhunter:  morning, my issue is that on my client in vlc media player it says cannot pre fill buffer. I'm unsing arch linux on my server and want to stream with this code: ffmpeg -f x11grab -s 1400x1050 -r 30 -i :0.0+0,0 -f mpegts udp://192.168.178.238:1234    ..238 is my server and on the client i type in vlc udp://192.168.178.238:1234
[15:33] <durandal_1707> matthias_arch: that upscales from vga to 1400x
[15:33] <barhom> I have a udp multicast stream that has, _teletext_ subtitles, they work in VLC. However, I need to convert these to dvb_subtitles. ffmpeg -i udp://@239.201.3.74:3301 -c:v copy -c:s copy -acodec copy -f mpegts udp://239.201.3.74:3302?pkt_size=1316
[15:34] <barhom> this successfully takes the multicast input and copies the c:s to dvb_subtitle, but it seems to incorrectly recognise the teletext input because its taking it as dvb_subtitles
[15:34] <barhom> anyone have any ideas?
[15:36] <matthias_arch> durandal_1707: My screen is 1400x1050, what do you mean?
[15:41] <durandal_1707> matthias_arch: i mean that you record at vga and rescale
[15:44] <matthias_arch> durandal_1707: but i want to record my desktop how can i do that?
[15:48] <durandal_1707> read documentation
[15:50] <matthias_arch> durandal_1707: haven't you got a working example for me? i read the doc, but not getting it working
[16:01] <durandal_1707> matthias_arch: i'm answering when i have free time
[18:41] <tlhiv_work> i'm having some problem with A/V sync when i try to overlay two videos together ... i'm using the following command
[18:41] <tlhiv_work> ffmpeg -i foo.mp4 -i bar.mp4 -vf "movie=bar.mp4 [a]; movie='circle_240x240.png' [b]; [a][b] overlay [i];[in][i] overlay=1022:96" -y out.mp4
[18:42] <tlhiv_work> foo.mp4 has no audio ... bar.mp4 has the audio and bar.mp4 is perfectly synced A/V
[18:42] <tlhiv_work> however, out.mp4 gets really bad out of sync about 2 minutes in
[19:24] <bouba> hi folks, just to keep in touch with http://doom10.org/index.php?topic=304.msg5646#msg5646 h264/SVC , still not supported ?
[19:24] <bouba> please
[19:24] <durandal_1707> bouba: what is that?
[19:25] <bouba> an adaptive streaming standard
[19:26] <durandal_1707> is there bug open for that?
[19:27] <bouba> no, but the topic is 2010 old
[19:27] <bouba> so i tried to keep in touch
[19:32] <durandal_1707> well user are just users, they do not code
[19:37] <tlhiv_work> any idea about my a/v sync issue?
[19:42] <tlhiv_work> ok ... i'll paste the exact command i placed in here in a pastebin if that will help
[19:42] <durandal_1707> not just exact command, but whole output
[19:42] <tlhiv_work> ffmpeg -i foo.mp4 -i bar.mp4 -vf "movie=bar.mp4 [a]; movie='circle_240x240.png' [b]; [a][b] overlay [i];[in][i] overlay=1022:96" -y out.mp4
[19:43] <tlhiv_work> foo.mp4 and bar.mp4 have the EXACT number of frames ... foo.mp4 has no audio ... bar.mp4 is perfectly synced ... i can't figure out why out.mp4 would have A/V sync issues
[19:44] <tlhiv_work> i'm re-encoding now to get complete output of the encoding process
[19:45] <tlhiv_work> http://pastebin.tlhiv.org/caxLKGia
[19:46] <durandal_1707> tlhiv_work: you are using very old version
[19:47] <tlhiv_work> i would definitely upgrade if i thought that was the problem, but i would like to be fairly certain before
[19:48] <tlhiv_work> if there were a statically (self-contained) ffmpeg for linux with all of the codecs "embedded" without having to do an install over my existing version, i would try that, but i'm not sure such a thing exists
[19:50] <durandal_1707> it sure it is
[19:50] <durandal_1707> and you can compile one yourself
[19:51] <tlhiv_work> the only reason i say this is that i have a custom version that i have built that accepts spacebar inputs while encoding to store timestamps into a file for something else i'm doing
[19:52] <tlhiv_work> ok ... using the newest version now
[19:52] <tlhiv_work> we'll see if it makes a difference
[19:54] <tlhiv_work> durandal_1707: wow ... you were correct
[19:54] <tlhiv_work> that was the issue
[19:54] <tlhiv_work> grrr
[21:32] <xlinkz0_> can force_key_frames be accurate?
[21:32] <xlinkz0_> i asked for a keyframe at 0.500 seconds and it put one on 0.6
[21:33] <xlinkz0_> if I ask at 0.4 it puts one at 0.48
[21:33] <xlinkz0_> my framerate is 25 so it's not like there are not any frams in between...
[22:24] <xlinkz0> did anyone answer by any chance to the force_key_frames question?
[23:13] <llogan> burek: is there a difference betwee https://trac.ffmpeg.org/wiki/Limiting%20the%20output%20bitrate and https://trac.ffmpeg.org/wiki/Limiting%20The%20Output%20Bitrate
[23:28] <yajiv> hi, i'm trying to lower the size of my ffmpeg libraries (in particular libavcodec.so)
[23:29] <yajiv> i ./configure like so: http://pastebin.com/3n3JyXC2
[23:29] <yajiv> all the libraries are significantly smaller, however libavcodec is still 900k
[23:29] <yajiv> it's built for android as the configure command cross-prefix indicates.
[23:29] <yajiv> is this about as small as it's going to get? Does anyone have tips to get it smaller?
[23:33] <durandal_1707> yajiv: no audio?
[23:34] <yajiv> no audio
[23:34] <yajiv> (well for now) i'm just hoping to get this as small as possible.
[23:35] <jedir0x> Howdy - i'm using attempting to use ffmpeg with android via javacv (java->jni bridge to ffmpeg libraries) - I'm very new to this stuff.  I'm trying to add audio to a video that I'm generating from an in memory canvas (bitmap).   The video is working fine, and the audio works - unless i try to offset the start point of the audio.  I'm using avformat_seek_file to seek to a specific time and begin decoding.
[23:36] <jedir0x> I'm not sure how to ask my question
[23:36] <jedir0x> :/
[23:36] <jedir0x> I'll sit in the corner until i figure it out
[00:00] --- Wed Oct  2 2013


More information about the Ffmpeg-devel-irc mailing list