[Ffmpeg-devel-irc] ffmpeg.log.20191106

[burek]

[01:03:31 CET] <GenTooMan> so how do I set a subtitle track as the default in FFMPEG when using something like "-i ./subs/BC105.ass -metadata:s language=eng" what is the key to set that? I can't seem to find a list of metadata keys
[01:08:25 CET] <cehoyos> Should be the "disposition" command line option, not directly related to metadata
[01:10:49 CET] <GenTooMan> well I should have expected that... LOL ffmpeg is a bit complicated I guess.
[01:20:14 CET] <ncouloute> I'm trying to get ffmpeg to make the audio stream and the video steam the same length. I used the flag -shortest but the audio stream is still longer than the video stream. Whats the secret to get this flag to work?
[01:22:41 CET] <cehoyos> What do you mean with "longer"? Audio and video (may) have different properties that make "same length" depend on the definition or impossible.
[01:25:53 CET] <ncouloute> Duration of video stream is 4280.826550 and Duration of audiostream: 4283.823000 as reported by ffprobe. MediaInfo Video: 1 hour 11 mins 20 second 791 ms MediaInfo Audio: 1 hour 11 min 23 seconds 840 ms .
[01:27:12 CET] <ncouloute> I'm guessing you getting at that the fps must be the same or at least divisble by each other?
[01:28:24 CET] <cehoyos> ffprobe (and mediainfo which is not needed for any FFmpeg question or bug report) likely cannot know the actual duration (or do not necessarily know it)
[01:28:43 CET] <cehoyos> What does "ffmpeg -i file -f null -" show (possibly with -an and -vn)?
[01:29:55 CET] <cehoyos> And more important: Please paste the command line you used to create the file together with the complete, uncut console output and provide a link here, that may be enough to understand your issue.
[01:35:53 CET] <ncouloute> I'm guessing we are looking for the  "time:" and not the Duration that it comes back with instantly. Waiting for it to go through the file...
[01:45:55 CET] <ncouloute> Both time=01:11:23.77 when using -an and -vn. So I guess that means they are the same length ?
[01:47:29 CET] <cehoyos> That seems at least possible, yes
[01:48:06 CET] <cehoyos> But since we haven't seen any console output so far, nobody here can know. I primarily wanted to point out that ffprobe cannot easily see the duration of a stream
[01:50:21 CET] <ncouloute> I'm going to get a -report log here in a bit. Going to reencode both streams just to be sure and see if I can get the fps the same as well
[02:03:33 CET] <petecouture> Can the RTSP muxer be used as a server playing -i mp4s ? I would have thought so but Im not seeing anything in the spec and it seems to imply is a push only. The muxer can be used to send a stream using RTSP ANNOUNCE to a server supporting it (currently Darwin Streaming Server and Mischa Spiegelmocks RTSP server).
[02:04:15 CET] <petecouture> I can set it up by piping into CVLC but trying to keep it all in FFMPEG
[04:10:02 CET] <cpplearner> Guys, I'm trying to mux a raw H.264 stream into a FLV container programmatically. However, when I look though the FLV specification, it does not clearly state what a embedded H.264 stream looks like. It does mention for a F4V container, but it's a different container, right? Any thoughts?
[04:10:30 CET] <cpplearner> The specific document that I was referring to: https://www.adobe.com/content/dam/acom/en/devnet/flv/video_file_format_spec_v10_1.pdf
[08:54:07 CET] <pagios> hello, i need to run ffmpeg from my nodejs server, what is the best way to do it /
[09:15:40 CET] <TikityTik> pagios: why not ask nodejs?
[09:15:57 CET] <TikityTik> it's just a command, no?
[10:16:19 CET] <BeerLover> pagios install it and use it
[10:16:46 CET] <BeerLover> either via shell commands using child process or use a package with ffmpeg bindings
[10:17:18 CET] <BeerLover> ffmpeg man pages don't have -profile
[10:17:23 CET] <BeerLover> What does it do?
[10:40:29 CET] <pagios> How can i exit ffmpeg process if i stop receiving a livestream? by default it keeps waiting
[11:01:37 CET] <pagios> BeerLover any idea?
[11:03:40 CET] <BeerLover> you can try yourself
[11:08:58 CET] <BeerLover> if there is no stream, it should exit
[11:54:58 CET] <void09> anyone know if it's possible to detect video or audio (h264/aac) breaking errors in a file ?
[13:20:59 CET] <MarcWeber> I've been trying screen capture on Windows. The problem is that list devices (directx) has german special chars.
[13:21:19 CET] <MarcWeber> So the microfone gets listed, but no matter what I tried ffmpeg keeps telling me it doesn't find audio.
[13:21:50 CET] <MarcWeber> Eg I tried > x 2>&1 then copy paste from utf-8 file to .bat file (also setting utf-8) but didn't work.
[13:22:33 CET] <MarcWeber> Only the Mikrofon (Standard Audio Gerät) the ä is my problem.
[15:27:05 CET] <Freneticks> Is it possible to copy video file with ffmpeg without any remux ?
[15:27:53 CET] <DHE> umm. you mean "cp" ?
[15:43:02 CET] <Freneticks> I want to copy a video file (-map 0 -c copy) but since I use hls to cut it in piece I get a lot of DTS PTS problem and sound get desync
[15:43:19 CET] <Freneticks> I tried everything not sure how to solve this problem
[16:04:26 CET] <pagios> hi , can i  use -i -vf scale ...  -f hls ... -vf scale ... -f hls ... syntax?
[16:05:06 CET] <furq> no
[16:05:19 CET] <pagios> why?
[16:05:36 CET] <pagios> whats wrong furq
[16:06:14 CET] <furq> -lavfi "split[s1][s2];[s1]scale[o1];[s2]scale[o2]" -map "[o1]" -f hls out1 -map "[o2]" -f hls out2
[16:06:17 CET] <furq> something like that
[16:09:38 CET] <DHE> that's best if you're doing something else before it, like maybe deinterlacing. otherwise the original method doesn't seem any faster...
[16:21:39 CET] <pagios> DHE furq  my syntax seems to work :/
[16:21:52 CET] <pagios> i am only scaling down and setting a bitrate
[16:22:02 CET] <pagios> can i go with that?
[16:26:01 CET] <DHE> if you were doing deinterlacing, your method would run the de-interlacer 3 times. furq's method scales better since you would deinterlace it once and then scale the version to your 3 different variants
[16:28:38 CET] <pagios> got it thanks all
[16:41:15 CET] <Radioactive> hello guys
[16:41:31 CET] <Radioactive> more ac-3 questions
[16:41:48 CET] <Radioactive> does libavcodec have known issues reading ac3 audio?
[16:41:56 CET] <Radioactive> particularly seeking
[16:43:10 CET] <Radioactive> for instance, calling av_seek_frame with a specific sample i wish to read from does not seem to provide an avFrame that has a sample start/end containing that sample
[16:46:24 CET] <Radioactive> for instance, say i want to seek to sample 192000, calling av_seek_frame will return an avFrame with a best_effort_timestamp of 190080 and nb_samples 1536
[16:46:40 CET] <Radioactive> i.e. this frame does not contain 192000
[17:12:14 CET] <Radioactive> also ive been assuming so far that timebase for audio files/streams is just in samples
[17:12:28 CET] <Radioactive> i mean that certainly seems to be the case for pcm and aac audio
[17:12:35 CET] <Radioactive> but is it possible that's not the case for ac3
[17:12:57 CET] Action: Radioactive channels JEEB
[17:22:20 CET] <DHE> any luck?
[17:47:28 CET] <Radioactive> forgot to mention the most serious issue
[17:49:03 CET] <Radioactive> so if that frame doesnt contain my sample im after, i continue the loop and decode the next frame, the frame range for the next frame however could now be way ahead of my seek frame, as if a bunch of frames have just been skipped
[17:49:09 CET] <Radioactive> my seek sample*
[19:18:08 CET] <desgroup> Hey, we're trying to do atrac9 decoding with avcodec but we're having trouble.
[19:18:16 CET] <desgroup> Hold on...
[19:18:29 CET] <desgroup> ```
[19:18:39 CET] <desgroup>     AVCodec *atrac = avcodec_find_decoder(AV_CODEC_ID_ATRAC9);    AVCodecContext *context = avcodec_alloc_context3(atrac);    context->block_align = 0x1000;    context->extradata_size = 12;    int err = avcodec_open2(context, atrac, nullptr);
[19:18:48 CET] <desgroup> I hope you can read that better than I can.
[19:19:14 CET] <desgroup> We're getting bad access at 0x0 in avcodec_open2, any ideas?
[19:23:52 CET] <desgroup> Sorry, my internet disconnected. I'm not sure if my question went through.
[19:23:58 CET] <desgroup> But this time I have a pastebin link! https://pastebin.com/cPZKWwVa
[19:24:31 CET] <desgroup> avcodec_open2 gives me bad access at address 0, I'm not 100% sure why. Is there an obvious mistake I made or an atrac9 example? I haven't found any so far.
[19:25:22 CET] <desgroup> Do I have to specify a dictionary?
[19:27:49 CET] <desgroup> Eh?! I do? I thought I didn't need to...
[19:29:34 CET] <desgroup> It doesn't especially say in docs, but it works for other codecs like h264.
[19:29:47 CET] <desgroup> =L
[19:39:06 CET] <desgroup> Right, so my problem is still there... This is confusing.
[19:39:42 CET] <desgroup> avcodec_open2(context, atrac, (AVDictionary)0x100) will throw a bad access at 0x100, so clearly it is dereferencing the dictionary at some point when (I think) it should be checking for null.
[19:40:30 CET] <desgroup> However, if I pass it the reference to AVDictionary *dict = nullptr; (some existing dictionary), it throws a bad access at 0x0 anyway.
[19:40:41 CET] <desgroup> Even if I set a value with av_dict_set(&dict, "cheese", "bent", 0); it will still throw a bad access.
[19:40:49 CET] <desgroup> This is frustrating, am I doing anything wrong?
[19:43:26 CET] <desgroup36> eh?
[19:43:27 CET] <desgroup36> my bad, I'm back again
[22:50:13 CET] <Atlenohen> Hello
[22:50:44 CET] <Atlenohen> UTVideo codec's use cases?
[22:51:39 CET] <Atlenohen> I heard it can be alternative to ffv1?
[22:53:08 CET] <durandal_1707> nope
[22:53:25 CET] <durandal_1707> ffv1 and utvideo are very different
[22:53:54 CET] <Atlenohen> Oh I heard it a while back somewhere, or vs magicyuv
[23:18:53 CET] <furq> Atlenohen: ut video is very fast and correspondingly doesn't compress very well
[23:19:25 CET] <furq> it's better than huffyuv and there are vfw/dshow/quicktime components for it so it actually works with commercial NLEs
[23:25:16 CET] <Atlenohen> oh
[23:25:30 CET] <Atlenohen> tnx
[00:00:00 CET] --- Thu Nov  7 2019