[Ffmpeg-devel-irc] ffmpeg.log.20190713

burek burek021 at gmail.com
Sun Jul 14 03:05:02 EEST 2019


[00:25:10 CEST] <iive> linext, yep.
[01:22:03 CEST] <linext> iive, i was looking for the next episode but there appears to not be one
[01:22:57 CEST] <iive> just 12 episodes and incomplete story. I guess S3 would carry on.
[02:05:21 CEST] <nine_milli> blb_
[02:13:49 CEST] <bray90820> So I have an mkv that i converted with ffmpeg -i "video.mkv" -c copy "video.mp4" which plays VLC but won't play in quicktime
[02:18:20 CEST] <another> depends on the codecs i guess
[02:18:49 CEST] <bray90820> Also the same video that I coped with "video.mkv" -c:a copy "video.mp4" had audio issues the volume changes in Quicktime but not VLC
[02:19:29 CEST] <bray90820> another: Would mkv and mp4 be the codecs
[02:23:20 CEST] <bray90820> So no the codec is AC3 for both videos at least for the audio
[02:23:26 CEST] <bray90820> mpgv for video
[02:31:46 CEST] <another> mkv and mp4 are containers
[02:43:12 CEST] <bray90820> Yes that's why I corrected myself
[02:43:36 CEST] <bray90820> So why is the video weird in quicktime and fine in VLC
[02:49:44 CEST] <iive> ac3 have dymaic range compression. probably you get normal sound too low and then too strong explosions and stuff. in that case... they made it that way. vlc kind of "normalize" the sound level.
[02:50:39 CEST] <iive> as for the video... what is the video codec?
[03:11:19 CEST] <kepstin> bray90820: it just might be that the video codec settings aren't compatible with quicktime.
[03:11:51 CEST] <kepstin> bray90820: if i could see the output from ffmpeg when you did the conversion, i could probably say why it didn't work
[07:09:10 CEST] <SoItBegins> Im trying to use the tpad filter but ffmpeg says it doesnt exist. Huh?
[07:28:19 CEST] <furq> SoItBegins: that filter isn't in any release yet
[07:28:36 CEST] <SoItBegins> furq: Yeah, figured it out. Trying to install HEAD now.
[09:07:46 CEST] <nine_milli> blb_
[11:38:59 CEST] <MoziM> 23.976040
[17:58:21 CEST] <olavx200> I have a bunch of audio files which only outputs audio to one channel (right) and i would like to have the channel output audio to both channels (left and right). How can i modify the files to contain the same audio on the left as on the right.
[18:00:04 CEST] <furq> olavx200: -af "pan=mono|c0=FR"
[18:09:33 CEST] <olavx200> Thanks, that is what i am looking for.
[19:49:46 CEST] <Thomas_J> I don't understand. I am inputing a camera streaming mp4 at 720p and using a -c:v copy for output. the video is going out at 640x480, 4:3 aspect and not 720p.
[19:52:52 CEST] <Thomas_J> Could this be something that Facebook is insisting on during the handshake?
[20:13:33 CEST] <furq> Thomas_J: no
[20:13:50 CEST] <furq> either facebook is reencoding it or your camera isn't actually outputting 720p
[20:14:16 CEST] <furq> there's no mechanism for a protocol to change the encoding parameters like that
[20:14:54 CEST] <furq> also if this is a raspberry pi then i doubt it would be able to do realtime 480p x264 encoding
[20:39:21 CEST] <friendofafriend> Thomas_J: What is your command line?  You have to specify the image size when capturing from a camera by v4l.
[20:41:17 CEST] <Thomas_J> I just looked at the debug out and found this which doesn't make sense. I have the camera set to 720p. ;;;
[20:41:19 CEST] <Thomas_J> Stream #0:0, 0, 1/1000: Video: h264 (Main), 1 reference frame ([7][0][0][0] / 0x0007), yuvj420p(pc, bt709, progressive, left), 640x480 (0x0) [SAR 4:3 DAR 16:9], 0/1, q=2-31, 30 fps, 30 tbr, 1k tbn, 90k tbc
[20:41:41 CEST] <Thomas_J> Metadata from the camera?
[20:41:55 CEST] <friendofafriend> That's saying the camera is not set to 720p.
[20:43:14 CEST] <friendofafriend> Use the video_size flag to explicitly set the size, like:  ffmpeg -f v4l2 -framerate 25 -video_size 640x480 -i /dev/video0 output.mkv
[20:43:19 CEST] <Thomas_J> Exactly, but I go to it's GUI and set it to 720p and it acknowledges this when set.
[20:43:22 CEST] <friendofafriend> Check out https://trac.ffmpeg.org/wiki/Capture/Webcam
[20:43:59 CEST] <Thomas_J> My input is an rtsp stream.
[20:44:40 CEST] <friendofafriend> OK, then verify its output with: ffprobe "rtsp://whatever_the_uri"
[20:44:52 CEST] <Thomas_J> Thanks
[20:44:58 CEST] <friendofafriend> Always welcome.
[20:48:13 CEST] <Thomas_J> Interesting. The camera's primary output is as I set it, 1080p but the secondary output has gone rogue.
[20:48:52 CEST] <Thomas_J> This is no cheap camera but it is a POS.
[20:49:21 CEST] <Thomas_J> It's an Amazon speciall.
[20:51:48 CEST] <Thomas_J> If the primary is working properly then to heck with it. I intend to crop and resize the display port on the input anyway.
[20:53:53 CEST] <Thomas_J> Irony is that I may just stream to Facebook with a final 640x480 anyway. It's just that the camera is scanning a 16:9 aspect and sending it as 4:3.
[20:56:32 CEST] <Mysterytrain> I Have a buncha videos for which I want to rip out the audio. is there a way to tell ffmpeg to just grab the first audio stream? Normally I've been going: for i in *.mkv;do ffmpeg -i "$i" -map 0:1 -c:a mp3 "${i%mkv}mp3";done is there a shortcut to just do the first audiostream?
[20:58:07 CEST] <Lyphe0> -map a:0
[21:01:11 CEST] <Mysterytrain> ah nifty
[21:01:32 CEST] <DHE> I think you mean 0:a
[21:02:16 CEST] <DHE> or maybe even 0:a:0 since 0:a has some selection logic, meaning what's 'first' is less consistent
[21:06:57 CEST] <Mysterytrain> yeah my method doesn't always work, because sometimes 0:1 is subtitle or video and i have to ffprobe each video to double check that the audio streams are consistent throughout the batch
[21:07:14 CEST] <Mysterytrain> so, that will help immensely :)
[21:36:32 CEST] <Lyphe0> DHE: whats wrong with a:0? if theres only one input, its equivalent to 0:a:0, no?
[22:03:39 CEST] <DHE> Lyphe0: I am unable to find anything in the ffmpeg-all.html indicating this is acceptabe to -map
[22:06:30 CEST] <Lyphe0> read https://www.ffmpeg.org/ffmpeg-all.html#Stream-specifiers-1
[22:07:49 CEST] <DHE> that is specific to things like -c which might be annotated with -c:a:0, but not for -map
[22:10:52 CEST] <Lyphe0> huh yeah, the according to the doc, and input file id is in fact mandatory
[22:11:17 CEST] <Lyphe0> but hey, it works anyway, so ¯\_(Ä)_/¯
[22:11:36 CEST] <Lyphe0> -the
[22:11:59 CEST] <furq> looks like a:0 just autoselects input file 0
[22:12:28 CEST] <furq> $ ffmpeg -f lavfi -i sine -f lavfi -i testsrc -map v:0 -f null -
[22:12:31 CEST] <furq> so that throws an error
[23:01:55 CEST] <SortaCore> Hey fellas... just a quick one, I'm running a file through null muxer to get the duration from output... is it okay to use codec copy or will that mess with the timing? The official example doesn't use copy so I'm suspicious
[23:02:25 CEST] <furq> it might mess with the timing
[23:02:36 CEST] <furq> the default for the null muxer is to just decode and discard, so it won't take very long anyway
[23:03:10 CEST] <SortaCore> heh, compared to codec copy it's very slow, but yea, I get ya
[23:03:16 CEST] <SortaCore> thanks furq
[23:03:22 CEST] <SortaCore> ciaooo
[23:11:53 CEST] <nine_milli> blb_
[00:00:00 CEST] --- Sun Jul 14 2019


More information about the Ffmpeg-devel-irc mailing list