[Ffmpeg-devel-irc] ffmpeg.log.20140602

Tue Jun 3 02:05:01 CEST 2014

[01:03] <jnvsor> Question about streaming: When I stream a premade video file to twitch it cuts off the end, and ffmpeg says it failed to update the header with the correct duration: http://pastebin.com/pWmznqXe
[01:03] <jnvsor> How do I fix this?
[08:30] <jnvsor> Hi, I'm uploading a video file to twitch, but twitch keeps cutting off the end of the file - how can I make it take the whole thing?
[09:00] <jnvsor> Hmm, is there any way to specify a length for the final video and just put out black when the input file runs out until the requested time?
[13:43] <user3> how can I test the integrity of a .mp4 file?
[14:11] <user3> answered (I googled out the answer)
[14:44] <solars> hi there, if I record from my webcam with  ffmpeg -f video4linux2 -i /dev/video0 out.avi how can I specify the maximum resolution and fps to increase quality?
[15:01] <another> -s 800x600
[15:18] <smo_> hi, i use ffmpeg to make transcoding of web tv what can i use to accelerate the beginning of the transcoding sometimes i have a big delay for the transcoding to get started ?
[15:19] <smo_> stuck on the Stream #0:1 -> #0:1 (mp3 -> libvorbis) for exemple then it start after xx seconds...
[15:26] <smo_> ps: i use "Spawning ffmpeg -i pipe:0 -f matroska -sn -c:v libx264 -preset fast -profile:v high -deinterlace -c:a libvorbis -b:a 256k -threads 0 pipe:1" to trancscode
[15:51] <smo_> no idears?
[15:55] <DelphiWorld> hey ;)
[15:55] <DelphiWorld> dvbsub... how to reencode?
[16:48] <lestrb> I am trying to extract a video from swf but the output file is empty. Extracting audio works. Any help?
[17:13] <shtirlic> Is it possible to split 1 big mxf file for 2 smaller files and transcode it in parallel in mp4(h264/aac) and after that concat in one mp4?
[17:13] <shtirlic> So eventually I get 2x speed of transcoding?
[17:33] <caius_> Hi all. I have Ubuntu 14.04 64-bit installed, and as many have noticed, ffmpeg has been substituted by avconv and no ffmpeg is package is provided. I then downloaded both version 2.2 and the latest Git source code. At configuration I seem to miss stereo3d in the list of filters printed on the screen, and after compilation ffmpeg complains that there is not such filter. I have searched for a solution without success. Any suggestions? Please
[17:33] <caius_> , let me know what extra information you may need.
[17:46] <iive> caius_: ffmpeg -filters | grep stereo3d
[18:25] <caius_> iive: this is the output
[18:25] <caius_> iive: ffmpeg version N-63694-ge91f27c Copyright (c) 2000-2014 the FFmpeg developers
[18:25] <caius_>   built on Jun  2 2014 17:23:49 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[18:25] <caius_>   configuration: --prefix=/opt/ffmpeg
[18:25] <caius_>   libavutil      52. 89.100 / 52. 89.100
[18:25] <caius_>   libavcodec     55. 66.100 / 55. 66.100
[18:25] <caius_>   libavformat    55. 42.100 / 55. 42.100
[18:25] <caius_>   libavdevice    55. 13.101 / 55. 13.101
[18:25] <caius_>   libavfilter     4.  5.100 /  4.  5.100
[18:25] <caius_>   libswscale      2.  6.100 /  2.  6.100
[18:25] <caius_>   libswresample   0. 19.100 /  0. 19.100
[18:26] <caius_> iive: I've just made a "fresh" compilation/installation
[18:27] <iive> i assume this is git, not 2.2.2
[18:28] <caius_> iive: yes, it is
[18:28] <iive> printing this info indicates it doesn't recognize -filters option.
[18:28] <iive> make sure you typed it correctly (single '-' , ends with 's' )
[18:29] <iive> and use pastebin site if the output is more than 1-2 lines
[18:30] <caius_> iive: it was correct. indeed, if I don't grep it, I have the full list of filters:
[18:30] <caius_> iive: ffmpeg version N-63694-ge91f27c Copyright (c) 2000-2014 the FFmpeg developers
[18:30] <caius_>   built on Jun  2 2014 17:23:49 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[18:30] <caius_>   configuration: --prefix=/opt/ffmpeg
[18:30] <caius_>   libavutil      52. 89.100 / 52. 89.100
[18:30] <caius_>   libavcodec     55. 66.100 / 55. 66.100
[18:30] <caius_>   libavformat    55. 42.100 / 55. 42.100
[18:30] <caius_>   libavdevice    55. 13.101 / 55. 13.101
[18:30] <caius_>   libavfilter     4.  5.100 /  4.  5.100
[18:30] <caius_>   libswscale      2.  6.100 /  2.  6.100
[18:30] <caius_>   libswresample   0. 19.100 /  0. 19.100
[18:31] <caius_> Filters:
[18:31] <caius_>   T.. = Timeline support
[18:31] <caius_>   .S. = Slice threading
[18:31] <caius_>   ..C = Commmand support
[18:31] <caius_>   A = Audio input/output
[18:31] <caius_>   V = Video input/output
[18:31] <caius_>   N = Dynamic number and/or type of input/output
[18:31] <caius_>   | = Source or sink filter
[18:38] <BtbN> Is the vbr option for fdkaacenc documented somewhere? It sets AACENC_BITRATEMODE to a valume in the range from 1 to 5, but i can't find any documentation in libfdk what exactly this does. It only documents the values 0 and 8.
[18:39] <caius_> iive: back
[18:39] <iive> welcome back :)
[18:39] <caius_> iive: http://pastebin.com/g441kbb9
[18:39] <iive> it seems that you need --enable-gpl in configure
[18:39] <caius_> iive: :)
[18:40] <caius_> iive: OK, let me try configure again and check the list of filters it prints out
[18:41] <caius_> iive: great, the filter is there.
[18:41] <iive> \o/
[18:42] <caius_> iive: thanks a lot, and sorry for the mess :)
[18:42] <iive> no problem. just be more careful.
[18:42] <caius_> iive: I will
[18:58] <aarobc> Good mroning!
[18:59] <aarobc> Anyone here messed with a chromecast?
[19:49] <a141982112> is there any way to capture webcam support in ffmpeg instead of v4l2-ctl
[19:59] <a141982112> just now i am using http://pastebin.com/fHwXZJVV
[19:59] <a141982112> but i am unable to find frame rate supported for each resolution
[20:01] <llogan> a141982112: can you show the output of "v4l2-ctl --list-formats-ext"?
[20:02] <a141982112> http://pastebin.com/KTEvYgdW
[20:06] <a141982112> what argument does ffprobe expect for option all, it says missing argument , here is whole command http://pastebin.com/KTEvYgdW
[20:07] <llogan> that's the same link
[20:07] <llogan> as before
[20:08] <a141982112> sry http://pastebin.com/QnPHv0MM
[20:09] <llogan> all is not a valid option.
[20:09] <llogan> "all" is a value for the -list_formats and -list_standards V4L2 input device private options
[20:09] <llogan> http://ffmpeg.org/ffmpeg-devices.html#video4linux2_002c-v4l2
[20:12] <a141982112> thanks, i did not go through document, i was looking in file libavdevice/v4l2.c and thought all is also option
[20:13] <aarobc> does the -map command just specify how many tracks there are? Trying to preserve the structure of an mkv file but just add another aac track in addition to the ac3 track
[20:14] <a141982112> do we have any option to get the supported frame rate for each supported resolution using ffmpeg.
[20:15] <a141982112> aarobc, you can use multiple map for two track
[20:17] <llogan> aarobc: ffmpeg -i input.mkv -i audio.m4a -map 0 -map 1:a -c copy -shortest output.mkv
[20:17] <llogan> http://ffmpeg.org/ffmpeg.html#Advanced-options
[20:20] <llogan> a141982112: not that i know of, but i believe default behavior is to select the largest supported frame size, and highest supported frame rate.
[20:22] <llogan> ...and to choose the "raw" input stream instead of compressed if available
[20:25] <a141982112> thanks llogan, as per looking in source code it looks there is also no option to do it even using libavdevice
[20:30] <a141982112> why do ffmpeg prefer raw instead of compressed, does it save time (ffmpeg encoder faster then webcam) or is it quality that we get from raw video
[20:35] <benlieb> I'm seeing (vie ffprobe) that one of my videos has a negative start time after trying to figure out why my concat is not working the way I'd like. Duration: 00:02:00.06, start: -9.045333, bitrate: 501 kb/s. This is having some strange consequences.
[20:35] <benlieb> First off, what does this mean?
[20:36] <llogan> a141982112: i'm guessing it will select raw by default. probably for quality.
[20:37] <benlieb> I'm the one who made that video using a command like this, which, at least in the eyes of the players I've used, is a good way to get a subsection of video from a longer one:     cmd = "ffmpeg -y -ss #{@import.lesson_times[0]} -i #{@dir}/raw/#{@import.raw_source} -c copy -to #{to} -map_chapters -1  #{@dir}/raw_lessons/#{@import.lesson_id}.mp4"
[20:44] <benlieb> c_14: to the rescue?
[20:47] <c_14> If I had to guess, I'd say that the nearest iframe is 9.045333 seconds after the point you wanted to seek to and that gives it the negative start time, but I'm not sure.
[20:48] <benlieb> c_14: when I play this video in quicktime, it plays normally. When I play it in ffplay, it has 9 seconds of video BEFORE the -ss time in the "crop". When I use the concat filter, it includes these 9 seconds :(
[20:50] <c_14> ye, in that case I'm pretty sure it's what I said, try seeking those 9 seconds when you're concatenating the files.
[20:51] <benlieb> c_14: but that time to the nearest iframe will be different for every video. I'm going to be doing thousands of these...
[20:52] <c_14> Either don't use -codec copy in the first command, or try and programatically determine the start time and if it's negative seek that.
[20:52] <c_14> The 'problem' with codec copy is that it can only cut at iframes.
[20:53] <benlieb> c_14, ah
[20:53] <benlieb> so if I take that out it will not do that?
[20:53] <benlieb> but will it compress the video again and lose resolution
[20:53] <benlieb> ?
[20:53] <benlieb> c_14: Æ
[20:53] <c_14> Depends on the video codec you use.
[20:54] <c_14> You can shoot for something uncompressed or lossless and it won't losse quality in that step, but you'll be paying in filesize.
[20:54] <c_14> -s
[20:54] <benlieb> c_14: its going from mp4 to mp4
[20:54] <c_14> mp4 is a container, not a codec
[20:56] <benlieb> c_14 I've confirmed that this is what's happening by changing the the -ss time 2 seconds, and the amount the video was "off" equaled the difference.
[20:58] <benlieb> c_14: the original vid is :Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, smpte170m), 720x362 [SAR 8:9 DAR 320:181], 311 kb/s, 29.97 fps, 29.97 tbr, 90k tbn, 180k tbc (default)
[20:58] <benlieb> c_14: if I take out -c copy will it use this codec?
[20:59] <c_14> No, if you take out -c copy it'll default to whatever the container defaults to.
[21:00] <c_14> Which with mp4 is mpeg4 video.
[21:01] <llogan> if libx264 is not supported
[21:01] <benlieb> c_14: that reencoding will increase my processing time by 25%
[21:02] <c_14> benlieb: If you don't mind 'wasting' hdd space for a while you can use rawvideo and then convert to h264 in the last step.
[21:02] <benlieb> do have an idea as far as how to determine programatically the "start time"
[21:03] <benlieb> it seems I would have to parse the ffprobe output?
[21:03] <llogan> benlieb: see https://trac.ffmpeg.org/wiki/FFprobeTips
[21:16] <benlieb> llogan: tnx
[21:24] <benlieb> c_14: good news
[21:24] <benlieb> using -ss 0:0 is what is needed to set things right during concat, not the actual number of "offset"
[21:26] <c_14> hmm, that would make sense
[21:26] <benlieb> makes my life easier
[21:33] <myfree> question
[21:34] <myfree> I use "ffmpeg -v error -i 2.mp4 -f null -" to verify the integrity of file 2.mp4 and I'd like the exit status to be nonzero whenever there is an error, but that's not always so. for example, for file 2.mp4 the exit status is 0 in spite of these error messages: [mov,mp4,m4a,3gp,3g2,mj2 @ 0000000002b706a0] stream 0, offset 0x5bc67a1: partial file [h264 @ 0000000002b2f560] AVC: nal size 41195...
[21:34] <myfree> ...Last message repeated 1 times [h264 @ 0000000002b2f560] no frame! 2.mp4: Invalid data found when processing input
[21:36] <myfree> i.e. file 2.mp4 is somewhat corrupted but the exit status says it's not
[21:38] <myfree> on both Linux and Windows
[22:56] <ac_slater> hey all, I'm using the ffmpeg command line to learn how to invoke different pieces of the API. I see that ffserver can do RTP/RTSP stuff, but is it possible to stream via rtp just with the ffmpeg application?
[22:57] <ac_slater> and do I need to hand write an SDP ?
[23:03] <vl4kn0> Hi, is it possible to use cropdetect together with the crop filter? (in code, not in the ffmpeg commandline tool). I've looked at the cropdetect filter and found out that not only it prints the detected information in the log but also sets the frame metainformation like so: http://bit.ly/1pN2ZYg but I have no idea how to use that information in the code where loading the cropfilter.
[00:00] --- Tue Jun  3 2014