[Ffmpeg-devel-irc] ffmpeg.log.20180921

[burek]

[03:25:41 CEST] <rjeli> why is there both a stream time base and a packet time base?
[03:57:15 CEST] <rjeli> oh. they don't. that's just PyAV. facepalm
[03:58:16 CEST] <nicolas17> I think there *is* a timebase in the codec and another in the container, fwiw...
[05:15:43 CEST] <nicolas17> if I pass a live-stream HLS playlist as input to ffmpeg, will it automatically stop transcoding when the live stream ends?
[05:42:23 CEST] <nicolas17> when youtube-dl ran ffmpeg passing it a live HLS playlist, sometimes it would hang, and when doing Ctrl-C it wouldn't write the moov atom
[05:42:39 CEST] <nicolas17> (I'm doing -acodec copy -vcodec copy blah.mp4)
[05:42:49 CEST] <nicolas17> hopefully running ffmpeg directly I won't have that problem...
[06:49:51 CEST] <rjeli> fuck it
[06:50:17 CEST] <rjeli> ive spent too much time fighting nvenc and AVFormatContexts and nvidia libraries
[06:50:21 CEST] <rjeli> im using mjpeg
[11:33:16 CEST] <ilushka4> hey guys, I'm trying to run a rtsp stream from university's server. I always have an authentication problem, although I have the right credentials. Here is the log: https://paste.ee/p/CigFa
[11:53:52 CEST] <gzp> hello! I would like to select the first audio stream from an .mka container. Previously used -map 0:a but right now it doesn't works. mka have 2 streams: #0:0 dts and #0:1 flac
[11:54:44 CEST] <gzp> 0:a worked if I had a video stream too, but right now I have only audio streams
[11:56:30 CEST] <gzp> [mp3 @ 0xa96de00] Invalid audio stream. Exactly one MP3 audio stream is required.
[13:02:16 CEST] <rmbeer> hello...
[13:02:33 CEST] <durandal_1707> welcome
[13:02:45 CEST] <rmbeer> how to fuse audio and video where the result is with the length of the video?...
[13:19:35 CEST] <rmbeer> how to fuse audio and video with exact time in the video?...
[13:19:51 CEST] <rmbeer> or from the video...
[13:20:48 CEST] <w4t3r> Hi, I can't compile ffmpeg on macOS 10.13.6 with Nvidia GPU acceleration support. I'm using a GTX 1080ti on a hackintosh.
[13:20:48 CEST] <w4t3r> The error is: configure doesn't find ffnvcodec though it is installed and pkg-config sees it.
[13:20:48 CEST] <w4t3r> Build command: https://pastebin.com/hsSUrQF4
[13:20:48 CEST] <w4t3r> ffbuild/config.log: https://lfil.es/f/65f3a96b
[13:24:56 CEST] <w4t3r> @rmbeer if the video and audio files are equally long and only contain one stream each, which fit into your target container without reencoding, you can use ffmpeg -i videofile.mp4 -i audiofile.mp3 -c copy outfile.mkv for example
[13:26:20 CEST] <rmbeer> w4t3r, the problem is that the video is more shorter that the audio... i need a output cut off to the length of video...
[13:26:50 CEST] <rmbeer> if i use the -ss or -to not is very exact...
[13:26:56 CEST] <rmbeer> same with -t ...
[13:31:25 CEST] <w4t3r> Sure? I'd try to cut off the video with -ss too, never had any issues with accuracy. Just specify the length in seconds.subseconds, like 2591.124
[13:31:41 CEST] <w4t3r> I mean -ss to start point and then -t, sorry
[13:33:24 CEST] <rmbeer> i use from start (without parameter) to -t time that take from -i of a video time....
[13:33:51 CEST] <rmbeer> but is the best step for cut off to a exact video time?...
[13:37:20 CEST] <rmbeer> maybe with this take or put any extra frames in the result of video...
[14:20:46 CEST] <th3_v0ice> rmbeer: You can use -shortest
[14:22:23 CEST] <rmbeer> th3_v0ice, thanks!...
[15:01:37 CEST] <learningc> Can I play video file with ffmpeg command?
[16:15:02 CEST] <relaxed> learningc: sort of, I know there's opengl output. mpv would be a better choice
[16:15:17 CEST] <relaxed> there's also ffplay
[16:21:56 CEST] <Nacht> Anyone ever encountered MPEGTS files who didn't start with a Sync Byte ?
[16:22:15 CEST] <Mavrik> yp
[16:22:53 CEST] <Nacht> I'm looking at a few ones, and I'm really wondering what the garbage is that they put in front of the sync byte
[16:23:11 CEST] <Mavrik> Well it can be whatever.
[16:23:16 CEST] <Mavrik> MPEG-TS afterall is used for streaming
[16:23:23 CEST] <Mavrik> So if you do something like netcat or connect at random time
[16:23:31 CEST] <Mavrik> You'll jump into the stream before the sync byte
[16:23:53 CEST] <Mavrik> So if someone gave you files that were collected with netcat or something, there's absolutely no reason for them to start with a sync byte :P
[16:24:09 CEST] <Mavrik> After all, this is why sync byte exists ;)
[16:24:21 CEST] <Nacht> So how does FFMPEG check if it's a valid file then ? Check for sync byte, then check if the 188th byte from that is also a sync byte ? (if its 188-byte)
[16:25:26 CEST] <Mavrik> https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/mpegts.c#L565
[16:26:57 CEST] <Nacht> Cheers
[16:29:16 CEST] <abhilash_gunu> hii everyone
[16:29:30 CEST] <abhilash_gunu> has anyone used fluent-ffmpeg?
[16:33:20 CEST] <King_DuckZ> hi, my program is spamming this https://alarmpi.no-ip.org/kamokan/cl?colourless as it tries to read a .webm file
[16:33:49 CEST] <King_DuckZ> anyone's got a clue about what that means?
[16:34:54 CEST] <King_DuckZ> movie reading doesn't seem to be advancing to the next frame, I don't know if it's because of that message
[16:35:17 CEST] <kepstin> King_DuckZ: looks like a programming error in the application using ffmpeg rather than an issue with ffmpeg itself
[16:35:38 CEST] <kepstin> it seems to be expecting a specific pts value, but the file being read was rounded differently?
[16:36:19 CEST] <kepstin> no idea what's going on there, tbh, since it's not ffmpeg code.
[16:36:35 CEST] <King_DuckZ> looks like a programming error in the application <-- very likely :)
[16:37:39 CEST] <King_DuckZ> well, from what I can tell the code is stuck into this call: sws_scale(swscontext, frame->data, frame->linesize, 0, height(), dst_data, linesize);
[16:38:29 CEST] <kepstin> it looks like the application is reading to some point in the file, hitting a bug where it doesn't handle pts correctly, then looping back to the start of the file
[16:38:48 CEST] <kepstin> so it would be reading frames continuously, so if you're scaling that might show up as a hot spot, but it's not the issue.
[16:39:11 CEST] <kepstin> but that said, all guesses. I can't really say anything about code that I haven't seen
[16:40:20 CEST] <King_DuckZ> haha no worries, you've given me a clue, I'll try to see if there's anything that stands out *after* the sws_scale call then
[16:40:44 CEST] <King_DuckZ> in fact I see an if (frame->pts != needed_pts) { blah } right after
[00:00:00 CEST] --- Sat Sep 22 2018