[Ffmpeg-devel-irc] ffmpeg.log.20120705

[burek]

[00:58] <onexserie04> hi
[00:58] <onexserie04> any here Speak german and can help me ?
[01:01] <onexserie04> ok maybe any here how can me help ? i try to use ffmpeg with FME for rtmfp
[01:01] <onexserie04> need to connect with a special line but i cant get it work
[01:01] <onexserie04> Flash Media Encoder stream perfekt
[01:01] <onexserie04> Xsplit Perfekt
[01:01] <onexserie04> all other perfekt but not ffmpeg
[01:02] <onexserie04> rtmp://bla.testurl.com/multicast/streamname?fms.multicast.type=3&fms.multicast.groupspec=G%3A010121058d969eef6ecad3c29a3a629280e686cf0c3f5d5a86aff3ca12020c923adc6c92010c160e666d732e6d756c7469636173742e6578616d706c65210ead30f62ecaa35c20cd9adec44d1a95705b0bac051cd314765f126957aeb13377000715313233343536
[01:02] <onexserie04> thats does not work
[01:02] <onexserie04> maybe any can modify the line for me that work ?
[01:09] <onexserie04> hmm no guys here ?
[01:10] <lnd7> i can translate for you if you have some text
[01:11] <onexserie04> i have ask my question
[01:11] <onexserie04> i try to stream with ffmpeg but i need to use multicast.type and group
[01:11] <onexserie04> see link above
[01:12] <lnd7> sry no idea about that
[01:12] <onexserie04> :(
[01:22] <a|3x> is it possible to set up ffmpeg to retry rtmp connections instead of exiting?
[02:17] <catdude_> I've made a fair bit of progress, but still having issues with my settings. I've got ffmpeg sucking in a raw stream, transcoding it, and pushing it out to Wowza on another server. The details (command lines and results) are at http://pastebin.com/HgQUpXtR. Basically, ffmpeg reports circular_buffer: OVERRUN and PES packet size mismatch, and Wowza reports Zero sized function (client:68157671:unknown): type:20 size:0: and Data format incorrect, killing client
[02:17] <catdude_> connection. The ffmpeg command line is: ffmpeg -strict -2 -i udp://@:5000 -acodec mp2 -ac 2 -b:v 8000k -ar 48k -async 1 -bufsize 3M -s 720x480 -r 31.67 -vcodec libx264 -y -f h264 rtmp://192.168.200.247/live/camera1
[02:17] <catdude_> I'm experimenting with both RTMP and RTSP outputs but neither one seems to be especially effective. Any clues as to what parameters I'm messing up here?
[06:19] <niman> hi how i can fix this err when i try to convert file i see that Executable not found: "/usr/bin/x11/xterm".
[06:19] <niman> Press OK to ignore and risk data corruption.
[06:19] <niman> Press Cancel to kill the program.
[06:23] <niman> Executable not found: "/usr/bin/x11/xterm".
[06:23] <niman> Press OK to ignore and risk data corruption.
[06:23] <niman> Press Cancel to kill the program.
[06:23] <niman> how can i fix this err
[06:51] <sacarasc> niman: What are you using to get that error?
[08:48] <Alternity> haiiiii!
[08:48] <Alternity> any idea how good librtmp is these days?
[09:35] <brocatz> what's the best format to capture a live stream (dshow driver, webcam 720p) in audio and video
[09:35] <brocatz> i can transcode it to ffmpeg after the person is done
[09:35] <brocatz> er to x264
[09:35] <Alternity> some lossless codec probably?
[09:36] <Alternity> huffyuv
[09:36] <brocatz> does that do audio
[09:36] <brocatz> or do i just do that adjacent
[09:36] <brocatz> as aac
[09:36] <Alternity> its a video codec so no....
[09:36] <Alternity> for audio you want lossless as well
[09:36] <brocatz> or wav
[09:36] <Alternity> so use monkeys audio
[09:36] <Alternity> or flac
[09:36] <Alternity> or just wav ya
[09:36] <brocatz> kk
[09:36] <brocatz> let me try that
[09:41] <brocatz> not sure how to do that
[10:04] <brocatz> my video is getting spat out with some deformation
[10:04] <brocatz> i've set -s 1280x720 -aspect 16:9
[10:04] <brocatz> which is what the video signal should be
[10:04] <brocatz> but it's still 1/4 too short
[10:28] <brocatz> hum
[10:28] <brocatz> so if i set -s 1280x720 -aspect 16:10
[10:28] <brocatz> the output mp4 is at the right ratio but the wrong size
[10:29] <brocatz> i want a 16:9 outpiut video
[10:29] <brocatz> with the right ratio pixels
[10:29] <brocatz> any ideas what i'm doing wrong
[10:33] <Mavrik> what aspect ratio does your input video have?
[10:34] <brocatz> i just captured it with the logitech software and looked at it with ffprobe
[10:35] <brocatz> it's just 1280x720 and a ratio of 1.7778 so 16:9
[10:35] <brocatz> but if i ffmpeg render from teh dshow device at 1280x720 with a ratio of 16:9 it's distorted
[10:35] <brocatz> i assume i don't understand dar/sar and how to specify things correctly
[10:36] <Mavrik> hmm
[10:36] <Mavrik> usually -s -aspect forces the right DAR
[10:36] <Mavrik> try forcing DAR to 1:1
[10:37] <brocatz> this is my current video line
[10:37] <brocatz> SET VIDEO=-i video="Logitech HD Pro Webcam C920":audio="Microphone (HD Pro Webcam C920)" -preset ultrafast -s 1280x720 -aspect 16:10 -r 25
[10:37] <brocatz> if i change it to 16:9 it becomes distorted
[10:38] <brocatz> er -vdecodec libx264 also
[10:38] <brocatz> -e
[10:39] <brocatz> yeah if i drop the -s and -aspec i get a perfect 4:3
[10:39] <brocatz> but it's not a 4:3 cam
[10:39] <brocatz> maybe something else is wrong
[10:41] <Mavrik> brocatz: what's your input?
[10:41] <Mavrik> can you please paste full ffmpeg output when you start the command?
[10:41] <brocatz> logitech 720p webcam
[10:41] <brocatz> sure
[10:41] <Alternity> in pastebin ;p
[10:43] <brocatz> http://dpaste.com/767603/plain/
[10:43] <Mavrik> brocatz: the output, not just command line :)
[10:43] <brocatz> kk
[10:46] <brocatz> http://dpaste.com/767604/plain/
[10:47] <brocatz> hum it's forcing the input stream to be 640:480 huh
[10:47] <Mavrik> brocatz: yeah, that's what I thought :)
[10:48] <brocatz> can i get around that
[10:48] <Mavrik> brocatz: do "-s 1280x720" BEFORE -i
[10:48] <brocatz> ok
[10:48] <Mavrik> (parameters before -i set INPUT)
[10:48] <Mavrik> so this should signal the webcam to send 720p video :)
[10:49] <brocatz> awesome
[10:49] <brocatz> yep
[10:49] <brocatz> sorry noob mistake
[10:49] <brocatz> next thing i don't want to use x264 in the capture phase
[10:50] <brocatz> i should just specify the file extension as .yuv?
[10:50] <brocatz> do i have 2 output files if i want to capture wav and yuv?
[10:50] <brocatz> is yuv only 420?
[10:50] <brocatz> is that its width or some other factor
[10:50] <brocatz> er height
[10:51] <Mavrik> brocatz: YUV is a pixel format, 420 is type of subsampling it uses
[10:51] <brocatz> ok
[10:51] <Mavrik> brocatz: your camera is grabbing input as BGR instead of YUV though
[10:51] <brocatz> so i should capture as bgr and say wav
[10:51] <brocatz> then convert to x264 later
[10:52] <Mavrik> brocatz: if you want to capture input without losing quality
[10:52] <Mavrik> I suggest you use huffyuv format
[10:52] <Mavrik> which will keep your video as BGR and also wont compress audio
[10:52] <brocatz> so that's just specifying .yuv? or also -vcodec
[10:52] <brocatz> ok
[10:52] <mxmx> z
[10:52] <Mavrik> brocatz: ffmpeg .... -i ... -vcodec huffyuv output.avi
[10:53] <Mavrik> brocatz: be warned: this will probably be supported only by ffmpeg and libav based players :)
[10:54] <brocatz> i am going to transcode after the capture to x264
[13:19] <virtus_> hi, is there a way to avoid errors with -map 0:a if there is no audio?
[13:21] <sacarasc> Run a script to see if there is audio and leave it out of the command if there isn't...
[13:26] <virtus_> I was afraid so, thanks
[13:28] <virtus_> another question, I want to remove the alpha channel from an RGBA file, is it sufficient to use "fifo, lutrgb=a=minval"? or is there a better way of doing that?
[13:29] <virtus_> or should I set it to 0 instead of minval?
[13:55] <saste> virtus_: yes should do
[13:55] <saste> o or minval is ok, minval is better although equivalent
[13:55] <saste> 0 or minval*
[14:56] <virtus_> did something change after 0.10 with the -ss option? I'm getting "Invalid duration specification for ss: 2.945833625"
[14:58] <Mavrik> virtus_: hmm, would 00:00:02.946 work?
[16:28] <onexserie04> hi any here can help me how i can stream to a rtmp server with this option ? rtmp://blabla.com/live/livestream?fms.multicast.type=1&fms.multicast.groupspec=G%3A01012105274123fdc56981c0201cc42d089effb4eb3484b218e7768d60ca801027015625010c160e666d732e6d756c7469636173742e6578616d706c65011b00070ae00000fe75300b1578636173646173646661&fms.multicast.address=224.0.0.254%3A30000
[16:28] <onexserie04> on fme i just insert this line on file name livestream?fms.multicast.type=1&fms.multicast.groupspec=G%3A01012105274123fdc56981c0201cc42d089effb4eb3484b218e7768d60ca801027015625010c160e666d732e6d756c7469636173742e6578616d706c65011b00070ae00000fe75300b1578636173646173646661&fms.multicast.address=224.0.0.254%3A30000
[20:21] <meekohi> Can you crop with libswscale, or only do rescaling?
[20:22] <meekohi> and if not, what's the usual approach for cropping?
[20:34] <iive> meekohi: cropping is just pointer arithmetic.
[20:35] <meekohi> iive: sure, but sws_scale needs to know the left offset for each row and there is no way to pass that information in.
[20:35] <iive> meekohi: image have 4 major parameters, width and heigh are 2 of them, the other are starting pointer (aka where the first pixel starts) and linesize/stride (the width of image buffer in bytes)
[20:36] <iive> so, if you provide width that is smaller than the original, you are cropping the image, same for the height.
[20:36] <iive> you just have to recaclucate the starting point/pixel too.
[20:36] <meekohi> iive: Ahhh okay I understand. Let me try that out.
[20:41] <meekohi> Is it safe to change those pointers directly, i.e. frame->data[0] += 3*ox; ?
[20:43] <iive> well, for the most part. there must be crop filter in avfilter, you may take some ideas from there.
[20:43] <meekohi> Yeah I'm looking at vf_crop.c but it's a bit cryptic for me.
[20:43] <iive> e.g. 3*ox would be for rgb24/bgr24
[20:44] <iive> at least the height is easy oy*linesize[0]
[20:44] <meekohi> Yeah the math isn't the hard part, guessing what ffmpeg will do is the hard part ;)
[21:09] <andym_> I'm having some issues with avcodec_decode_video2 and whittling down the list of potential problems.  After calling avcodec_alloc_context3, my AVCodecContext codec_id does not match the codec I pass in - is that normal?
[21:14] <andym_> Sorry, I really meant after avcodec_open - the codec_id on the AVCodecContext isn't the codec I passed in
[21:17] <roxlu> hi, I want to create a video from my application and wondering what the easiest image format is that ffmpeg can convert into a video? (I hope targa)
[21:19] <llogan> roxlu: what do you mean by "easiest"?
[21:19] <roxlu> oh sorry, easiest to generate from withing my c++ application
[21:19] <roxlu> I've got a "unsigned char*" of pixels rgb values
[21:21] <andym_> That should work with PIX_FMT_RGB, you may have to tweak your line size, etc
[21:21] <roxlu> andym_: I want to export to images and then use the command line to import
[21:22] <roxlu> would that be just a raw file then?
[21:22] <llogan> roxlu: you can see what ffmpeg can decode with "ffmpeg -codecs"
[21:22] <andym_> Ah, sorry, I was referring to using the API/libraries
[21:22] <llogan> although it doesn't go into specifics
[21:24] <roxlu>  DEV D  targa           Truevision Targa image
[21:24] <roxlu> The 2nd D, "Supports direct rendering method 1", what does that mean?
[22:05] <marcosmlopes> hey, someone can ask my question on stackoverflow? http://stackoverflow.com/questions/11334568/how-to-stream-an-webm-video-through-ffmpeg-using-node-js
[22:07] <marcosmlopes> hey, someone can answer** my question on stackoverflow? http://stackoverflow.com/questions/11334568/how-to-stream-an-webm-video-through-ffmpeg-using-node-js
[22:48] <andym_> avcodec_alloc_context and avcodec_alloc_context3 both create a context with a codec_id of CODEC_ID_JPEGLS and I have no idea why.  Can anyone offer any suggestions on what to look for?  Forcing the codec type resulted in funny decoder behaviour.  This is H264 packets.
[23:04] <burek> andym_, did you read ffmpeg's source code to see how it is properly used?
[23:18] <andym_> burek: Yes - source, headers, and samples
[23:18] <burek> well, that should help you :)
[23:19] <burek> just see how ffmpeg does it and do it the same way :)
[23:20] <andym_> Hence the current scenario :)  I'm going it that way.  avcodec_alloc_context, avcodec_alloc_context3, and avcodec_get_context_defaults3 all set the codec_id incorrectly for me.  I'm using an iOS build script, so now digging around that.
[23:47] <meliss> hey everyone, is there a way to run ffmpeg without all that extra terminal output
[23:48] <brocatz> in linux?
[23:48] <meliss> yeah
[23:48] <brocatz> you could pipe std err to somewhere
[23:50] <brocatz> ffmpeg args 2>/dev/null
[23:52] <judget> Hi Im trying a new ffmpeg build on Ubuntu 11.10 and the configure is whining about needing libvpx encoder >=0.9.7
[23:53] <judget> I guess i have to get the new source and buld that library
[23:53] <judget> but i can only seem to locate 0.9.6-1
[23:54] <llogan> judget: https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide
[23:54] <llogan> has a libvpx section
[23:55] <judget> tx llogan
[00:00] --- Fri Jul  6 2012