[Ffmpeg-devel-irc] ffmpeg.log.20130930

burek burek021 at gmail.com
Tue Oct 1 02:05:01 CEST 2013


[00:24] <megaTherion> Hello, I've a question - how does it work with subtitles which appear later in the stream, how am I able to copy them for example?
[00:24] <megaTherion> like 'New subtitle stream 0:4 at pos:450981902 and DTS:501.56s'
[00:42] <relaxed> megaTherion: pastebin the output of "ffmpeg -i input"
[00:42] <megaTherion> relaxed: they are not detected, if you mean that
[00:44] <relaxed> if not they're hardsubed into the video stream
[01:04] <tempus_fol> Hello, I've added --enable-libfdk_aac to my .configure, and now it doesn't seem to build anymore (ERROR: libfdk_aac not found); latest version from Git for ffmpeg. I've tried rebuilding libfdk_aac from src.rpm _and_ from git (as per https://trac.ffmpeg.org/wiki/CentosCompilationGuide )
[02:04] <peanuter> whenever i convert avi to ogg i am getting audio out of sync
[02:04] <peanuter> any tip/tricks to avoiding this?
[02:44] <lg_> 	ºô-‡
[03:02] <lg_> 'ffmpeg -i a.mp4 -vcodec copy b.mp4' command and the 'cat a.mp4 | ffmpeg -i pipe:0 -vcodec copy -f h264 pipe:1 | cat > c.mp4' command execution result What is the difference?
[03:03] <lg_> I have a video a.mp4, want it to play directly in chrome.
[03:04] <lg_> Command 'ffmpeg-i a.mp4-vcodec copy b.mp4' conversion play in chrome,
[03:04] <lg_> But the command 'cat a.mp4 | ffmpeg-i pipe: 0-vcodec copy-f h264 pipe: 1 | cat> c.mp4' can not be converted to play in chrome
[03:49] <lg_> Why use the pipe to convert video, no mp4 faststart head?
[03:51] <sacarasc> -f h264 will output raw H264, not the MP4 container.
[03:56] <lg_> Hi sacarasc . How to output mp4 container?
[03:56] <lg_> Do you have good idea sacarasc ?
[03:58] <sacarasc> -f mp4
[03:59] <lg_> use -f mp4 an error '[mp4 @ 0x12c8ce0] muxer does not support non seekable output'
[04:00] <lg_> Final conversion failed.
[04:01] <sacarasc> MP4 needs the whole file to create the file...
[04:01] <sacarasc> Not sure you can do what you want.
[04:02] <sacarasc> Wait, there's -movflags faststart will help.
[04:04] <lg_> Where ?
[04:05] <lg_> Did not find the 'man ffmpeg' manual
[04:06] <sacarasc> In the output options.
[04:13] <sacarasc> cat a.mp4 | ffmpeg-i pipe: 0 -vcodec copy -movflags faststart -f mp4 pipe: 1 | cat> c.mp4
[04:14] <sacarasc> Something like that.
[04:21] <lg_> error [mp4 muxer @ 0xff0ca0] [Eval @ 0x7fffdc54a1d0] Undefined constant or missing '(' in 'faststart'
[04:21] <lg_> [mp4 muxer @ 0xff0ca0] Unable to parse option value "faststart"
[04:21] <lg_> [mp4 muxer @ 0xff0ca0] Error setting option movflags to value faststart.
[04:25] <sacarasc> It seems the way ffmpeg does it is to encode the whole file, then do the faststart thing on the completed file.
[04:28] <lg_> that pipe is not support manner
[05:03] <defaultro> I'm really pissed off with youtube. Their compression settings have changed as compared to 2 years ago. Their encoding settings are way so bad. I'm looking at the video I posted last year and it's very sharp. Today, I created another video and it's very sharp. However, after uploading it to youtube and playing it at 1080p, the sharpness is gone. :(
[06:19] <EvanDotPro> defaultro: i'm not sure if it's still true, but at one point if you got your encoding settings just right, youtube wouldn't even transcode the HD version
[06:19] <EvanDotPro> they'd serve up exactly what you gave them
[06:26] <defaultro> I've tried testing so many different settings, they always reencode it
[10:34] <JY`> echo $$C
[15:21] <ayaka> I have know how to use libavformats' api to send a rtp stream, now I want to use rtsp, is there some example?
[15:39] <monocle> Hello, I am using ffmpeg to download a stream, ffmpeg -i source -bsf:a aac_adtstoasc -c copy test.mp4 is working but I want x264 mkv, so I have tried several different options to no avail
[15:40] <monocle> ffmpeg -i source -c:v libx264 -bsf:a aac_adtstoasc -c copy test.mkv, ffmpeg -i source -c:v libx264 -c:a libfaac -bsf:a aac_adtstoasc -c copy test.mkv etc
[15:40] <monocle> how can I make this work?
[15:43] <freezway> can i get ffprobe to NOT spew out its version when run?
[15:43] <klaxa> monocle: if you specify -c copy it might overwrite the setting you have for -c:v and -c:a
[15:43] <klaxa> are there more streams than audio and video?
[15:43] <freezway> nvm got it
[15:44] <monocle> klaxa, no
[15:44] <klaxa> then throw away -c copy
[15:44] <monocle> source is h264 720p aac
[15:44] <klaxa> huh? wait...
[15:44] <monocle> ffmpeg -i source -c:v libx264 -bsf:a aac_adtstoasc test.mkv like this?
[15:44] <klaxa> you want to keep it unencoded? then throw out -c:v libx264 and -c:a libx264
[15:45] <klaxa> if you want to just save it in a matsoka container do this: ffmpeg -i source -c copy test.mkv
[15:45] <monocle> then it is unencoded right?
[15:45] <klaxa> yeah it will copy the audio and videostream and put it in a matroska container
[15:46] <monocle> I et error parsing aac extradata, unable to determine samplerate
[15:47] <monocle> *get
[15:47] <klaxa> hmm...
[15:47] <klaxa> well i guess you have -bsf:a aac_adtstoasc for a reason so try to put that back in
[15:47] <monocle> if I add -bsf:a aac_adtstoasc I get no error, but then I do not keep the original audio right?
[15:48] <klaxa> the documentation says the bitstream filters operate on the encoded bistream without doing any decoding
[15:48] <klaxa> it should be more or less the original
[15:48] <monocle> what if I want output with ac3?
[15:49] <klaxa> use -c:a ac3 then
[15:50] <klaxa> ffmpeg -i source -c:v copy -bsf:a aac_adtstoasc -c:a ac3 test.mkv
[15:52] <monocle> error parsing adts frame header
[15:52] <monocle> failed to open bitstream filter aac_adtstoasc for stream 0 with codec ac3
[15:53] <klaxa> huh? uh... hm...
[15:53] <monocle> I could just download it pure perhaps, and then demux it and convert with neroaacenc?
[15:53] <klaxa> you could certainly post-process it
[15:54] <monocle> so I use: ffmpeg -i  source -c:v copy -bsf:a aac_adtstoasc test.mkv ? with no -c copy on end?
[15:55] <klaxa> yes
[15:57] <monocle> what would be the difference? I mean if I put -c copy at the end? When I drop it I get information about profiles etc, where as if I have it with it skips it
[15:57] <monocle> also downloads faster with -c copy
[16:55] <ztane> is it possible to somehow store the input stream timestamps only in a text file or so?
[16:56] <ztane> on cmdline
[17:00] <saste> ztane, ffprobe?
[17:07] <ztane> saste: hmm, actually I am segmenting a flv feed, I was under the impression that an flv live feed would contain wallclock time (possibly), but I only get relative timestamps
[17:07] <ztane> where the first segment is at 0
[17:52] <anxt> Hi.  I am wondering if someone knows where I could look to find out how to convert s24 to s16.  can I just divide by 2**8 ?  What is the non naive way?  I am looking for the algorithm, yet many many googles turn up cli utilities to make the conversion, but not the algorithm.  It is LPCM that I am dealing with.
[17:53] <durandal_1707> just remove lowest 8 bits
[17:53] <durandal_1707> right shift
[17:53] <zap0> remove 7.5 bits and round up is *slightly* more accurate.
[17:54] <zap0> but mostly..  it's not rocket science...  your just removing the LS(Byte)
[17:57] <anxt> of course I will need to deal with sign bit, but thats no biggie
[17:57] <zap0> durandal_1707, right shift on signed integer isn't quite right.
[17:57] <anxt> but it seems like the LSB is where a lot of the audio lives in the 24 bit pcm i have
[17:58] <zap0> then apply some compression first.
[17:58] <durandal_1707> zap0: yes you just copy that...
[17:58] <zap0> or "normalize"
[17:58] <anxt> it certainly is not swinging full rang.
[17:59] <zap0> volume and dynamic range are different things
[17:59] <anxt> the rms comes out quite low
[18:02] <defaultro> is -profile:v high  different from -preset?
[18:03] <sacarasc> Yes.
[18:06] <defaultro> ok
[18:06] <defaultro> i just found out that profile isn't compatible with lossless. Another question
[18:07] <defaultro> on lossless h.264 present, ultrafast(fastest encoding) and veryslow(best compression). Am I right that the former will produce bigger file and better image quality than the latter?
[18:07] <defaultro> not sure why I wrote the word present. ignore it
[18:07] <sacarasc> If it is lossless, they would be the same quality.
[18:07] <sacarasc> That is what lossless does.
[18:08] <defaultro> ok, that means, I can always just use ultrafast
[18:15] <defaultro> how do I check if my ffmpeg has x264 support?
[18:15] <defaultro> it's because i'm on a different machine
[18:18] <anxt> hmm, i would do well to read about the sinc function, this is a new one to me.
[18:20] <anxt> ffmpeg -formats iirc defaultro
[18:21] <defaultro> cool
[18:21] <defaultro> trying it now
[18:21] <defaultro> it worked. Reviewing the output now:)
[18:23] <defaultro> which one is for libx264
[18:23] <defaultro> it's because I am getting Unknown encoder 'libx264'
[18:24] <defaultro> but I see this in -formats     DE h264            raw H.264 video
[18:37] <anxt> i think h264 may be a subset of x264
[18:38] <anxt> i have that entry too, and yet my ffmpeg is compiled without x264 support
[18:40] <sacarasc> x264 is an encoder of h264 streams.
[18:41] <sacarasc> And the format is for the container, not for the codec used.
[19:10] <defaultro> ok
[19:11] <defaultro> this command is failing with
[19:11] <defaultro>  ffmpeg -loop 1 -i stormatbigtimer.jpg -c:v libx264 -t 3 -pix_fmt yuv420p out.mp4
[19:12] <defaultro> Unknown encoder 'libx264'
[19:12] <defaultro> looks like I may have to recompile ffmpeg
[19:12] <defaultro>  i think i installed it on my mac os x via brew
[19:33] <defaultro> you guys familiar with this make error? http://pastebin.com/SZ29fPPw
[19:35] <defaultro> wow, i got it working when I run brew uninstall x264
[19:35] <defaultro> :)
[19:41] <defaultro> it's working now :)
[21:00] <chrisballinger> Hey I was wondering if anyone could help me out doing a direct stream copy with the libavformat API? My videos are being output at 3600 fps, more info here http://stackoverflow.com/questions/19060253/files-created-with-a-direct-stream-copy-using-ffmpegs-libavformat-api-play-back
[22:15] <smjd> defaultro: my guess is that slower preset only leads to lower bitrate with equal quality
[22:15] <smjd> with crf, at least
[22:57] <Niatross> I have a basic question: Can PCM be either IEEE 754 floating-point or fixed-point (AKA: integer) or is IEEE 754 floating-point and fixed-point (AKA: integer) separate entities altogether?
[23:21] <Niatross> fuggitaboutit&i think i found my answer
[23:31] <anon__> How do I keep ffmpeg from writing global tags to matroskas?
[00:00] --- Tue Oct  1 2013


More information about the Ffmpeg-devel-irc mailing list