[Ffmpeg-devel-irc] ffmpeg.log.20120823

burek burek021 at gmail.com
Fri Aug 24 02:05:01 CEST 2012


[00:01] <Ganymede> And ideally, it would be a script that makes the decision of whether to re-encode the video or do a stream-copy.
[00:04] <rud> use mediainfo, parse the result, and act accordingly
[00:06] <Ganymede> The main question being what "act accordingly" entails. Where can I go to discover which videos the device/OS will playback?
[00:06] <Ganymede> If the device claims to support H264, does that mean it supports any H264 as long as the resolution and bitrate are low enough?
[00:12] <JEEB> 1) device can be limited by profile 2) device can be limited by level 3) device can have custom limitations
[00:13] <JEEB> f.ex. the PSP needed main profile, level 3, no more than three reference frames, and b-pyramid was unsupported
[00:29] <jdjrjc> I need a code line to convert an MP4 file to an DivX  AVI file (for an Eclipse T2800 mp3/video player).  Can anyone provide me with one?
[01:10] <Ganymede> JEEB: Do you happen to know where I could find those limitations for an HTC Amaze 4G? Both the T-Mobile page here http://www.t-mobile.com/shop/Phones/cell-phone-detail.aspx?cell-phone=HTC-Amaze-4G-Black and the HTC page here http://www.htc.com/us/smartphones/htc-amaze-4g-tmobile/ only have very vague specs.
[01:59] <JAJMaster> So, if I use ffmpeg (current source build) to downsample an MP3 I get different results based on whether I output to a .mp3 directly or use "... -f mp3 ->song.mp3". The former gives me a valid, working MP3. The latter yields an error when I play it saying it can't find the MP3 header. Any suggestions?
[01:59] <JAJMaster> Regardless of the MP3 I select to downsample (seemingly) the one using STDOUT redirect is always 309 bytes smaller than writing the file directly.
[05:05] <TeknoJuce> Hi guys have a url to a ipcamera that spits out mjpeg via ffplay -f mjpeg http://user:pass@IP/video.cgi I would like to use this with ffserver but cant seem to get it to do anything with it I have 4 other cams working with it from a capture card but this one stumps me
[05:13] <TeknoJuce> ffmpeg -f mjpeg http://admin:@10.13.37.95/video.cgi -i /tmp/feed5.ffm
[05:14] <TeknoJuce> /tmp/feed5.ffm: Invalid data found when processing input
[05:36] <sacarasc> TeknoJuce: Move the -i to before the URL.
[05:49] <TeknoJuce> so remove the -f
[05:50] <TeknoJuce> or ffmpeg -f mjpeg -i http://admin:@10.13.37.95/video.cgi -i /tmp/feed5.ffm
[05:50] <TeknoJuce> oops
[05:50] <TeknoJuce> or ffmpeg -f mjpeg -i http://admin:@10.13.37.95/video.cgi /tmp/feed5.ffm
[10:53] <Loplin> Hello. Through googling, I have found out how to convert an mkv to an mp4, but the extra audio track, and subtitles were missing. What should I look into to make sure these get converted as well?
[10:54] <Loplin> For reference, the command I used was ffmpeg -i input.mkv -vcodec copy -acodec copy out.mp4
[10:56] <Loplin> Or, avi would be fine too, I guess
[10:56] <ubitux> try -map 0
[10:56] <ubitux> or map each track one by one
[10:57] <JEEB> also make sure with subtitles that the given subtitle format is actually supported in the output container
[10:57] <JEEB> and no, in most cases you do NOT want avi
[10:57] <ubitux> :))
[10:57] <JEEB> (mp4 officially only supports 3gp ttxt IIRC)
[10:58] <ubitux> we have a muxer for that, but i think it might need a few fixes
[10:58] <takis> hi all
[10:58] <takis> anyone used the tiny_psnr before?
[10:59] <Loplin> ubitux: thanks. there does appear to be a problem with the subtitles... Stream #0:2 -> #0:2 (ass -> ?)
[10:59] <ubitux> Loplin: try -c:s mov_text and see if it works
[10:59] <Loplin> ubitux: which might be fine, how could I just skip the subtitles then, (but still get the other audio track)
[10:59] <Loplin> I'll do that
[11:00] <ubitux> if you want to skip subtitles, maybe just remap a & v, so -map 0:a and -map 0:v (i guess)
[11:02] <Loplin> -c:s mov_text still gives the previous error
[11:03] <JEEB> most probably there is no conversion code to do ass->3gpp ttxt
[11:03] <JEEB> if mov text is that
[11:03] <ubitux> there is
[11:03] <JEEB> (´
[11:04] <ubitux> it's the internal representation we use so it should work
[11:32] <japjap> I am loosing RTPs when using RTSP via UDP. How to cope with this ?
[11:37] <EvanDotPro> is anyone familiar with a strange artifact issue when using x11grab + gnome-shell? i get some strange blue square areas flashing when a lot moves around, such as going to the activities/overview view.
[11:37] <EvanDotPro> i thought it may have been an x264 codec issue but even using flashsv, i get the same result.
[11:38] <barhom> I got a software that just won't accept any video that doesn't have an audio track as well (never mind what it is). Im trying to convert a .jpg/png to video with a silent audio track without any luck, anyone knows how to add a silent track that keeps looping?
[12:03] <saste> barhom: check lavfi and aevalsrc, you can generate silence with it
[12:06] <barhom> I managed to generate a .mpeg with silenced audio
[12:07] <barhom> ffmpeg -i title_noaudio.mpeg -vcodec copy -acodec copy -f rtp rtp://239.1.1.1:10000 < works good
[12:08] <barhom> but Im unable to "loop" the video with its silenced audio
[12:09] <EvanDotPro> well, assuming no one knows what i'm talking about with the gnome-shell thing, i have another question.. i have a live stream that i'm doing, and i'm taking audio in from a usb mic and using x11grab + a filter to overlay my webcam... problem is the x11grab and webcam are between 0.5 and 1.5 seconds behind the video... i haven't been able to find an easy way to get them in sync.
[12:11] <EvanDotPro> command i'm using is here, http://evan.pro/caps/76a166.png
[12:11] <EvanDotPro> (i'm using a separate program called x11graber for a mouse-tracking feature, but i get the same result with ffmpeg's built-in x11grab.
[12:13] <EvanDotPro> i have the stream going live right now too, if that would help debug it.
[12:14] <relaxed> EvanDotPro: remove -sameq
[12:14] <EvanDotPro> relaxed: ah, i had just added that yesterday trying to debug this actually. let me re-test to be sure the issue still exists though.
[12:15] <EvanDotPro> i have it going at http://evan.pro/screencast/
[12:15] <relaxed> try using ffmpeg directly, instead of piping ffmpeg -> ffmpeg
[12:15] <EvanDotPro> you mean x11grabr -> ffmpeg? it's the same result. i ruled that out first.
[12:16] <barhom> ffmpeg -re -i title_noaudio.mpeg -vcodec copy -acodec copy -f rtp rtp://239.1.1.1:10000 < How can I loop that input so that the process never quits
[12:16] <barhom> like if title_noaudio.mpeg was 1000hrs long
[12:16] <relaxed> -loop 1
[12:16] <barhom> doesn't work on video
[12:17] <barhom> it just says loop doesn't exist
[12:17] <barhom> Option loop not found. actually
[12:17] <EvanDotPro> tonight the lag is even worse than when i was testing it yesterday :-/.. it was ~0.5 sec, now it's 2 or 3 seconds behind the video.
[12:20] <relaxed> barhom: ffmpeg -re -i <(while :;do ffmpeg -i title_noaudio.mpeg -map 0 -c copy -y /dev/stdout 2>/dev/null;done) -vcodec copy -acodec copy -f rtp rtp://239.1.1.1:10000
[12:21] <relaxed> EvanDotPro: did you try lowering the frame rate?
[12:23] <EvanDotPro> relaxed: yeah i started at 15
[12:24] <EvanDotPro> relaxed: actually, that gets me back down to the ~0.5s delay (going back to 15fps).. good to know that was the variable that caused the change.
[12:25] <EvanDotPro> so i guess lower framerate == closer audio.. is there a way i could compensate with the higher framerates then?
[12:26] <japjap> I am loosing RTPs when using RTSP via UDP. How to cope with this ?
[12:28] <barhom> relaxed: I see what you're trying to do, doesn't seem to be working though, Ill keep trying thanks
[12:29] <relaxed> add -f dvd before -y /dev/stdout
[12:37] <barhom> nothing, damn, there must be a way to convert a still jpg with an empty audio track that never quits
[14:39] <gxk> problem to capture from usb video grabber by running - ffmpeg -f video4linux2 -s 720x480 -r ntsc -i /dev/video0 ./out.mpeg
[14:40] <gxk> my ffmpeg from latest git
[14:47] <JAJMaster> So, if I use ffmpeg (current source build) to downsample an MP3 I get different results based on whether I output to a .mp3 directly or use "... -f mp3 ->song.mp3". The former gives me a valid, working MP3. The latter yields an error when I play it saying it can't find the MP3 header. Any suggestions?
[14:47] <JAJMaster> Regardless of the MP3 I select to downsample (seemingly) the one using STDOUT redirect is always 309 bytes smaller than writing the file directly.
[15:42] <JAJMaster> I could file a bug report, but I have googled six ways from sunday and still am not sure if it is an actual bug or I'm just doing something wrong.
[15:42] <JAJMaster> Grr.
[16:05] <brontosaurusrex> JAJMaster, exact command line u use in latter case?
[16:05] <JAJMaster> copying, sec
[16:06] <JAJMaster> command: /var/subsonic/transcode/ffmpeg -i "/mp3/Country/George Jones - The Grand Tour/01 George Jones - The Grand Tour.mp3" -ab 96k -v 0 -f mp3 - >jj2.mp3
[16:09] <JAJMaster> ffmpeg version N-43799-gbc151ae
[16:09] <JAJMaster> configure: ./configure --enable-libx264 --enable-libtheora --enable-libvorbis --enable-libvpx --enable-pthreads --enable-gpl --enable-shared --enable-nonfree --enable-libmp3lame
[16:14] <brontosaurusrex> JAJMaster, meh, i havent compiled with lame
[16:14] <brontosaurusrex> perhaps try to redirect error to dev/null ?
[16:15] <JAJMaster> anything is worth a shot at this point, cool
[16:16] <JAJMaster> No dice. I have examined the binary files (converted to hex) and they are substantively different, not just "extra" stuff in one file vs. the other.
[16:18] <brontosaurusrex> JAJMaster, give me few minutes ....
[16:18] <JAJMaster> fantastic. Thanks for the help. Frustration definitely up on this one
[16:19] <brontosaurusrex> how about > cat - file.mp3
[16:19] <brontosaurusrex> at the end
[16:19] <brontosaurusrex> if thats nix box
[16:20] <JAJMaster> will try
[16:24] <JAJMaster> nope
[16:24] <JAJMaster> results in binary identical file
[16:26] <brontosaurusrex> you got it?
[16:26] <brontosaurusrex> looking at my scripts, iam using this for some reason:
[16:26] <brontosaurusrex>  -i "$files" -vn -y -f wav - 2> /dev/null| $lame --preset standard - "$out.mp3"
[16:26] <brontosaurusrex> +ffmpeg at the beginning
[16:26] <JAJMaster> Oh, no, it didn't work that way either. I meant that using the "> cat - >file.mp3" results in a file identical to the one with just ">file.mp3"
[16:26] <brontosaurusrex> ok
[16:29] <JAJMaster> Ahh. Well, for context, I am using this for transcoding in Subsonic music streamer. They fairly recently switched from lame to ffmpeg (not sure why). I can set the downsample command to use lame directly and it works fine.
[16:41] <brontosaurusrex> JAJMaster, perhaps explain what exactly you want
[16:51] <JAJMaster> My goal is to be able to use the defaults in Subsonic which is using ffmpeg to downsample a mp3 and output to STDOUT (which gets streamed to client). Since I can't do that from the command line directly with ffmpeg I figured I would investigate here instead of Subsonic (although I made a post in their forums there is no response).
[16:52] <JAJMaster> So I guess I would say my goal is to be able to downsample a mp3 with ffmpeg and output to STDOUT successfully
[16:52] <JAJMaster> I am quite perplexed as to why STDOUT redirection is giving me substantively different binary data than just writing the file directly
[16:57] <JAJMaster> Alright, gotta head to some meetings. If you (or anyone else) thinks of anything just toss it out and I'll hit it when I get back. Thanks again.
[17:05] <hi117> JAJMaster some programs detect if theyre putting stuff out to a term or a pipe, might be that
[17:34] <julius_> hi
[17:35] <julius_> building ffmpeg from source, configure fails immediately without output, just the error:    ERROR: Xext not found              what package is missing?
[17:41] <applegekko> check your disto package database
[17:43] <julius_> ah, youre right. i did search for xext in packages and not as a name
[17:46] <julius_> thx
[19:43] <JAJMaster> hi117: Thanks. Probably won't be able to fix it unless I can get the ear of someone on the ffmpeg team that is involved with file I/O, but good to know.
[19:43] <hi117> ?
[19:43] <JAJMaster> Comment you made to me, possibly some time ago (6 lines up maybe)
[19:45] <JAJMaster> referring to ffmpeg writing out an mp3 differently based on whether it was to STDOUT or directly
[19:51] <hi117> ah yeah
[19:52] <hi117> i dont think ffmpeg does though
[19:52] <JAJMaster> Yeah, just major weirdness. The fact that it seems to always be 309 bytes seems "telling" in some way, but I can't seem to connect the dots.
[20:29] <hexsprite> hello again... wondering about concatening .mov files ... i tried some of the xamples i found on google but couldn't get it to work... the files are in this format: Stream #0:0(und): Video: h264 (Baseline) (avc1 / 0x31637661), yuv420p, 1280x720, 10791 kb/s, 25.89 fps, 23.98 tbr, 600 tbn, 1200 tbc
[20:30] <hexsprite> ffmpeg -i concat:"a1.mov|a2.mov" a3.mov -- doesn't work
[20:30] <hexsprite> get the output of a single file only
[20:30] <hexsprite> trying to concat without re-encoding... is possible?
[20:59] <hexsprite> seems like the oinly way i have foudn so far is to convert to mpg, cat the files together and then apply a vflip since for some reason the out put is flippped!
[21:01] <ubitux> you also have -vf concat
[21:01] <ubitux> but it needs a reencode
[21:03] <hexsprite> ubitux: so the method you recommend is what?
[21:03] <hexsprite> basically what im doing ?
[21:04] <ubitux> no idea
[21:07] <hexsprite> gonna try something other than ffmpeg for now then... too many weird quirks
[21:30] <mparodi> Hello
[21:30] <mparodi> I'm trying to concatenate two videos, that's what I'm using:
[21:31] <mparodi> ffmpeg -i intro0101.webm -i class0101.webm -vcodec copy -acodec copy -vcodec copy -acodec copy test.webm
[21:31] <mparodi> here is the output v
[21:31] <mparodi> v
[21:31] <mparodi> http://paste.debian.net/185308/
[21:32] <mparodi> it's not working, test.webm is "empty"
[21:32] <mparodi> any idea?
[21:33] <Mavrik> that... probably doesn't do what you think it does.
[21:33] <Mavrik> what are you trying to do?
[21:34] <mparodi> Mavrik, I want to
[21:34] <mparodi> I want to concatenate two videos
[21:35] <Mavrik> mparodi, yeah, that's definately the wrong way to do that :P
[21:35] <mparodi> actually what I'm trying to do is to concatenate them and skip the first 16 seconds of the second video
[21:36] <mparodi> (the first video is a replacement of the first 16 seconds of the second video)
[21:36] <Mavrik> ffmpeg is probably the wrong tool to do that :\
[21:36] <Mavrik> http://ffmpeg.org/faq.html#How-can-I-join-video-files_003f
[21:37] <mparodi> "To put them one after the other: this is called to concatenate them (in short: concat) and is addressed in this very faq. "
[21:37] <mparodi> this ^
[21:37] <mparodi> apart from that I want to skip the first 16 seconds of the second video, that's it
[21:39] <mparodi> I suppose it's something like:
[21:39] <mparodi> ffmpeg -i input1.avi -qscale:v 1 intermediate1.mpg
[21:39] <mparodi> ffmpeg -ss 16 -i input2.avi -qscale:v 1 intermediate2.mpg
[21:39] <mparodi> ...
[21:42] <Mavrik> hrm, that'll force reencoding
[21:45] <mparodi> hmn, ok, probably. so.. what is your suggestion?
[21:45] <mparodi> Mavrik
[21:45] <mparodi> (it's actually reencoding the video :|)
[21:45] <mparodi> is there a way to void that?
[21:45] <Mavrik> as I said, ffmpeg is really bad at concatenating files
[21:46] <Mavrik> mparodi, I suggest you use mkvmerge
[21:46] <Mavrik> mparodi, maybe use ffmpeg to cut off 16 seconds from 2nd video, then join them with mkvmerge
[21:46] <mparodi> in other words what I want is to overwrite the first 16 seconds of input2 with input1
[21:47] <Mavrik> (I can't remember if mkvmerge supports cutting at exact second boundary)
[22:11] <Loplin> Is there a way that I can transcode an mp3 stream in mkv to aac? I am converting mkv to mp4. The video works fine, but apparantly my ps3 won't play mp3 audio streams from a video
[22:11] <JEEB> fdk-aac would be the best AAC encoder available, but you'd have to compile it yourself together with ffmpeg
[22:12] <JEEB> and then you would set c:a or acodec to libfdk-aacenc
[22:12] <JEEB> https://github.com/mstorsjo/fdk-aac
[22:13] <JEEB> (note: you need to run 'autoreconf -fiv' in the cloned fdk-aac folder to create a configure file
[22:19] <mparodi_> Mavrik, just in case you know.. I'm using the guide you gave me but I'm having problems... it's reencoding the video but besides that after the first video is shown there are like 3 seconds of black screen and then the second part (without the first 20 seconds) starts
[22:20] <mparodi_> err, s/20/16/
[22:20] <mparodi_> in other words:   first video | 3 seconds blank | second video without the first 16 seconds
[22:22] <mparodi_> I just noticed the problem is in this part: cat intermediate1.mpg intermediate2.mpg > intermediate_all.mpg
[22:22] <mparodi_> even though intermediate1.mpg and intermediate2.mpg are correct (there's no problem there), intermediate_all.mpg is wrong (as I mentioned earlier)
[22:23] <mparodi_> (just in case, this is the guide: http://ffmpeg.org/faq.html#Concatenating-at-the-file-level)
[22:23] <mparodi_> any idea guys?
[22:24] <mparodi_> I'll try with mkvmerge now...
[22:25] <Loplin> JEEB: I've compiled fdk-aac and am I am now compiling ffmpeg-git with --enable-libfdk-aac. Will the aac portion fail silently if I don't have the configure in the right place, or will it tell me?
[22:25] <mparodi_> ok, same problem if I use mkvmerge intermediate1.mpg intermediate2.mpg -o foo.mpg
[22:25] <JEEB> Loplin, set it after -i and it will be an encoder setting
[22:25] <JEEB> before -i = decoder setting
[22:26] <Loplin> Oh, I haven't gotten there yet
[22:26] <Loplin> I'm just trying to make sure I compile it with fdk correctly
[22:26] <JEEB> as it is a codec setting it shouldn't really error out unless you A) put it onto the wrong side of things or B) you don't have such a codec
[22:26] <JEEB> you need --enable-nonfree if I remember correctly
[22:27] <JEEB> maybe even --enable-gpl
[22:27] <JEEB> it should tell you
[22:27] <JEEB> IIRC
[22:27] <mparodi_> JEEB, do you know about my problem?
[22:27] <JEEB> nope
[22:30] <mparodi_> anyone?
[22:34] <cbsrobot> mparodi_: have you tried http://ffmpeg.org/ffmpeg.html#concat-1 ?
[22:35] <Mavrik> mparodi_, why are you trying to use MKVmerge, a program for working with MKV (and WebM, which is practically MKV) on mpg files?
[22:35] <Mavrik> use it on the webms.
[22:37] <mparodi_> Mavrik, err, ok.. that makes sense
[22:37] <mparodi_> let me see
[22:37] <Mavrik> :)
[22:39] <Mavrik> right command line is "mkvmerge -o full.mkv file1.mkv +file2.mkv" (note the plus) though
[22:40] <JEEB> or just use mmg :)
[22:40] <JEEB> probably the best muxer GUI I've seen)
[22:40] <mparodi> whoops, lost the connection
[22:41] <mparodi> it didn't work with webm either
[22:41] <mparodi> now it plays the first 16 seconds and then stop
[22:41] <mparodi> I'm missing something or this guide has a bug: http://ffmpeg.org/faq.html#Concatenating-at-the-file-level
[22:41] <mparodi> :P
[22:41] <mparodi> because "cat intermediate1.mpg intermediate2.mpg > intermediate_all.mpg" doesn't work well
[23:11] <Loplin> JEEB: does this command seem correct? ffmpeg -i file.mkv -vcodec copy -acodec libfdk_aac -map 0:a -map 0:v file.mp4
[23:11] <JEEB> <JEEB> and then you would set c:a or acodec to libfdk-aacenc
[23:14] <Loplin> libfdk-aacenc doesn't appear to exist, but libfdk_aac did
[23:16] <JEEB> interesting
[23:17] <Loplin> ffmpeg -codecs | grep aac gives encodesr: aac libfaac libfdk_aac
[23:17] <JEEB> then it's that
[23:29] <Loplin> Hmm, after compiling, I seem to have a problem copying the video portion
[23:30] <Loplin> Is it possible to separate the audio into another file, convert the mkv to mp4, then merg the aac to the mp4?
[23:30] <JEEB> hmm, I wonder if that's the problem I heard about having both libfaac and fdk there... unless you got that mapping wrong
[23:31] <JEEB> although I can't say much without knowing how it failed :P
[23:31] <Loplin> Mmm, it doesn't really fail
[23:31] <Loplin> It completes the command, but then the file refuses to play
[23:40] <Loplin> Oh, the problem might be that my install only produces .a files, and no .so
[23:44] <t355u5> there also seems to be a problem with converting mp4 (x264/aac) to mkv (x264/ac3). if you do this conversion with an ffmpeg release after 0.10.2, the resulting videos don't start to play in vlc. it takes up to 50 seconds before they finally start playing and then the time indicator is not at the beginning of the video, but at the end. very strange. I could not find anything anywhere why that is...
[23:47] <JEEB> Loplin, that's most probably not the problem :P
[23:47] <JEEB> static installs are just fine
[00:00] --- Fri Aug 24 2012


More information about the Ffmpeg-devel-irc mailing list