[Ffmpeg-devel-irc] ffmpeg.log.20140713

[burek]

[08:39] <drkshadow> So, even l-smash, I can't seek the generated MP4: l-smash/cli/muxer -i Island.x264 -i audio.eng.ac3 -o Island.mp4. "seek_frame_generic failed as this stream seems to contain no keyframes after the target timestamp, 1002 non keyframes found"
[09:42] <relaxed> drkshadow: I blame JEEB
[09:49] <drkshadow> wtf is jeeb?
[09:54] <drkshadow> That error shouldn't be right. I'm encoding with x264, keyint=24
[10:07] <drkshadow> trying keyint=250, keyint_min=23
[10:24] <Fjorgynn> :)
[10:25] <Fjorgynn> I also blame JEEB
[10:25] <Fjorgynn> lol :P
[14:02] <bencc> audio and video are out of sync when using the following command
[14:02] <bencc> ffmpeg f pulse -ac 1 -i alsa_output.pci-0000_00_05.0.analog-stereo.monitor -f x11grab -r 30 -s 1280x960 -i :0.0 -acodec pcm_s16le -vcodec libx264 -preset:v ultrafast -crf 0 -threads 0 output.mkv
[14:02] <bencc> how can I improve it?
[18:28] <Neppy> ohi :D
[18:29] <Neppy> trying to get ffmpeg to burn subtitles..  it compains about not finding fontconfig (even though it should Oo) and says it loaded the .ass file.. but the resulting video has no subtitles burned onto it
[18:34] <Neppy> well of course now it works, maybe because i restarted cmd
[20:03] <77CAAAB9Z> hi there. i have 8 frames in jpeg and i want them to create a animated gif is this possible with ffmpeg
[20:04] <c_14> https://trac.ffmpeg.org/wiki/Create%20a%20video%20slideshow%20from%20images
[20:04] <c_14> just use gif as the output format
[20:06] <sacarasc> IIRC, you have to choose the right pixel format for gif (or at least had to with MPlayer) or it will look horrid.
[20:09] <sine-> thank you.
[20:10] <sine-> it works anyhow that i bodged it
[20:10] <sine-> im just trying to make some animated buttons logos etc for a website using gif with 8 frames max
[20:45] <aditsu> hi, what lossless codec can I use for converting a series of png frames to a video?
[20:46] <aditsu> I tried huffyuv, but the result seems a bit dull, perhaps from the color space conversion?
[20:51] <sfan5> how about ffv1?
[20:55] <aditsu> sfan5: still looks the same
[20:55] <aditsu> anyway, I'll be off now, gotta watch the game :p
[20:55] <aditsu> back later, maybe :)
[21:11] <buu> So, um, why does this filter have 'too many inputs': -filter_complex '[0:0] [0:1] [1:0] [1:1] [2:0] [2:1] [3:0] [3:1] [4:0] [4:1] [5:0] [5:1] concat=n=5:v=1:a=1 [v] [a]'
[21:12] <buu> And is there a simple way to concat an a/v stream from 5 different files?
[21:12] <c_14> whole commandline please, also that would be n=6
[21:12] <c_14> 0-5 = 6
[21:13] <buu> oh
[21:13] <buu> haha
[21:13] <buu> I named my files 1-5 and then switched to 0-4 numbering
[21:14] <buu> Well, that made it slightly happier: ffmpeg -i 1.mpg -i 2.mpg -i 3.mpg -i 4.mpg -i 5.mpg -filter_complex '[0:0] [0:1] [1:0] [1:1] [2:0] [2:1] [3:0] [3:1] [4:0] [4:1] concat=n=5:v=1:a=1 [v] [a]' -map '[v]' -map '[a]' -c:v libx264 -c:a ac3 out.mkv
[21:14] <buu> Now I get specifier :0 matches no streams
[21:15] <buu> Each .mpg file has two streams, 0:0 video mpeg, 0:1 audio ac3
[21:16] <buu> c_14: You want the exact error message?
[21:17] <c_14> yep, +some more context
[21:17] <buu> c_14: http://paste.ubuntu.com/7790499/
[21:18] <buu> Hrm. Is it because the a/v stream is reversed in mpeg1?
[21:18] <c_14> it might be
[21:19] <c_14> I'd try reversing the specs for 2-5
[21:20] <buu> ok
[21:20] <buu> Well, that worked. So A) that error message is terrible and B) why doesn't reversing just the first st.. h
[21:20] <buu> oh
[21:21] <buu> Because I did [v] [a]
[21:21] <buu> I guess? Aren't those names arbitrary?
[21:21] <c_14> they are
[21:21] <buu> Why couldn't I do '[0:1] [0:0] [1:0] [1:1]' ?
[21:22] <c_14> I think concat just needs [video] [audio] [video] [audio]
[21:22] <buu> Well the example has [v] [a1] [a2]
[21:22] Action: buu shrgs
[21:22] <c_14> because that has two audio tracks
[21:22] <c_14> iirc
[21:22] <ChocolateArmpits> Is there any format that would allow to have audio and video track of different lenghts ?
[21:22] <buu> Right
[21:23] <c_14> ChocolateArmpits: every format (almost)
[21:23] <buu> c_14: But if I had two video and two audio, would I do [v] [v] [a] [a] ?
[21:23] <ChocolateArmpits> c_14, hmmm would could I mux mp4 in that way ?
[21:23] <c_14> yep, if you have v=2:a=2. At least that's how i'd do it.
[21:24] <c_14> ChocolateArmpits: ffmpeg -i video -i audio -codec copy out.mp4
[21:24] <ChocolateArmpits> c_14, but what if I'm converting both audio and video ?
[21:25] <c_14> Give me the command you tried and the output? (use a pastebin service)
[21:27] <ChocolateArmpits> Well I haven't yet tried, just thinking of how to go about splitting a file to multiple parts to render on several machines, then concanating the video part back while the level of the audio would be increased. I want to do this without creating an audio file
[21:27] <c_14> So you're trying to cut a single video file so that it has different video and audio lengths?
[21:28] <c_14> use the trim and atrim filters with different numbers
[21:30] <ChocolateArmpits> c_14, but isn't trim a slow seeking filter ?
[21:33] <c_14> ye, but I don't know of any other way to do that
[21:33] <c_14> cause -ss -t and -to will cut both
[21:35] <ChocolateArmpits> I was thinking of piping ffmpeg output to sox, adjust the level there, then pipe back into ffmpeg where each video part would be concanated, is the last part possible (concanating video while adding audio ?)
[21:36] <c_14> sure
[21:36] <c_14> Just concat the video and then also add the audio stream.
[21:37] <ChocolateArmpits> so it would be "ffmpeg -i concat:"video_a|video_b" -i audiopipe" right ? with mapping of course
[21:38] <c_14> ye, but don't use the concat protocol, use the muxer.
[21:38] <c_14> unless you're working with things in mpegts format and/or know it'll work
[21:38] <ChocolateArmpits> what exactly ?
[21:38] <ChocolateArmpits> "use the muxer" part I mean
[21:39] <c_14> https://trac.ffmpeg.org/wiki/How%20to%20concatenate%20(join,%20merge)%20media%20files
[21:39] <c_14> The concat demuxer part.
[21:41] <ChocolateArmpits> oh the video segments will share the same encoding settings, then concat seems to be it, no ?
[21:41] <c_14> concat yes, but use the demuxer not the protocol
[21:42] <c_14> The demuxer works at the stream level and the protocol works on the file level which means that the demuxer works more often than the protocol.
[00:00] --- Mon Jul 14 2014