[Ffmpeg-devel-irc] ffmpeg.log.20200129

[burek]

[00:43:44 CET] <cehoyos> Feel free to test if any two streams are synchronized, the command line and console output look sane to me now
[00:44:32 CET] <cehoyos> At least there should be no increasing desync now
[01:05:59 CET] <nickname123> hi
[01:09:50 CET] <plitter> ok, ill do some more testing
[01:12:09 CET] <nickname123> i get a list of Fontconfig errors with the static build from https://johnvansickle.com/ffmpeg/ but not with ffmpeg from the Ubuntu repository
[01:12:34 CET] <nickname123> is it possible that the static builds don't work on Ubuntu 19.10/19.04?
[01:15:44 CET] <nickname123> fontconfig version is "2.13.1-2ubuntu2" is installed on my computer, if that helps
[02:19:16 CET] <Conjecture> Is there any codec better than flac for lossless audio compression?
[02:20:00 CET] <c3r1c3-Win> Are there any other lossless audio codecs out there?
[02:20:20 CET] <Conjecture> I believe there are
[02:21:45 CET] <pink_mist> some people like wavpack
[02:21:52 CET] <pink_mist> personally I prefer flac
[02:24:43 CET] <Conjecture> pink_mist, Didn't you used to be in #Slackware?
[02:24:52 CET] <Conjecture> Sorry, ##Slackware
[02:25:38 CET] <der_richter> there is also tak and ape
[02:25:39 CET] <pink_mist> yes, but I disagreed with how the ops discriminated against some users and did nothing against other users who did much worse things
[02:25:49 CET] <pink_mist> so I moved to ##slackware-help
[02:27:41 CET] <der_richter> there is also alac
[02:28:04 CET] <another> define "better"
[02:29:01 CET] <pink_mist> wavpack is fun in that you get two files, one much smaller, and if you only use that one, it's actually lossy (but still pretty good), and then the second which you can combine with the smaller one to get lossless
[02:29:16 CET] <Conjecture> Better compression
[02:29:18 CET] <Conjecture> Smaller files
[02:29:31 CET] <Conjecture> Though that would probably have to be an average and depend on the file type
[02:29:32 CET] <pink_mist> at least that was how I saw it used years ago
[02:35:52 CET] <furq> Conjecture: https://xiph.org/flac/comparison.html
[02:37:13 CET] <pink_mist> oh yeah, monkey's audio was also popular
[02:37:38 CET] <another> i think i saw ape compress better
[02:37:59 CET] <another> but support is much more limited
[02:38:51 CET] <furq> there's also the small matter of it taking 40x as much cpu to decode
[02:40:23 CET] <Conjecture> furq, Thanks
[02:41:36 CET] <another> all things considered, i think flac is the overall best lossless codec atm
[02:43:10 CET] <furq> https://clbin.com/M22pz
[02:43:14 CET] <furq> these are the ones that matter
[02:44:10 CET] <furq> but yeah anything other than flac is a waste of time
[02:46:03 CET] <pink_mist> lol, LEAD Screen Capture is included because L[EAD S]creen matches your regex :P
[02:47:12 CET] <furq> oh yeah
[02:47:24 CET] <furq> https://clbin.com/Ey8ow
[02:47:25 CET] <furq> there you go
[02:49:33 CET] <Conjecture> Thanks for the list
[08:34:20 CET] <th3_v0ice> Why would av_interleave_write_frame not write 1 last video frame to the file? I tried using the NULL packet to flush but that doesnt seem to help. FFmpeg CLI writes the last frame properly.
[08:39:07 CET] <th3_v0ice> I send 361 video AVPackets to the muxer and when I decode the file I get 360 video frames.
[11:11:34 CET] <Mavrik> th3_v0ice, did you flush ALL of the components? Video encoder and the muxer?
[11:18:27 CET] <th3_v0ice> Mavrik, yes. It seems that setting the AVPacket.duration solved the problem for me. Maybe decoders are not outputting the frame properly when duration of the AVPacket is set to 0.
[11:48:48 CET] <newbieG> How find the position of metadata in a mp4 video file like its in start or at the end?
[11:52:03 CET] <Mavrik> MP4 should always have metadata at start right?
[11:52:07 CET] <Mavrik> My memory is a bit rusty
[11:52:47 CET] <BtbN> It commonly has it at the end of the file.
[11:52:58 CET] <BtbN> Since it can only be generated after the file is complete.
[11:53:08 CET] <BtbN> Moving it to the front is an extra post-processing step.
[11:57:36 CET] <newbieG> Mavrik, It should but I have some mp4 file which are not streaming its only playing after downloading
[11:58:04 CET] <newbieG> Thats why I thought maybe it has metadata at the end
[11:58:58 CET] <BtbN> Players will just request the end via range requests and stream the file just fine usually.
[11:59:26 CET] <BtbN> But if in doubt, just qt-faststart the file.
[11:59:26 CET] <newbieG> BtbN, Can we pinpoint the metadata position?
[11:59:42 CET] <BtbN> Open it in a hex editor and search for the MOOV atom
[12:00:29 CET] <JEEB> you can look at how your mp4/mov files look by pointing f.ex. L-SMASH's boxdumper at it with `boxdumper --box FILE |less`
[12:05:21 CET] <newbieG> BtbN, <bh:00><bh:00><bh:00><bh:18>ftypmp42<bh:00><bh:00><bh:00><bh:00>mp42mp41<bh:00><bh:02>a<bh:b7>moov<bh:00><bh:00><bh:00>lmvh
[12:05:32 CET] <newbieG> This is the starting of metadata
[12:06:07 CET] <newbieG> ?
[13:00:03 CET] <familiyaF> Hi, I am transcoding via VAAPI on AMD and Intel GPU, the same file I have found that AMD GPU 3x time taken by Intel GPU. Is there a way I can increase the performance observed using AMD GPU
[13:01:11 CET] <familiyaF> Intel GPU: Intel Corporation UHD Graphics 620 rev 07
[13:01:46 CET] <familiyaF> AMD GPU: Ellesmere Radeon RX 580 rev e7
[13:12:17 CET] <JaredCE> hi all.  i'm using a 3rd party app that makes use of ffmpeg and ffprobe... currently ffprobe seems to be outputting this error: In MFMediaID_IdentifyFile, got Error 20266 (1218268970) FFPROBE generated a zero-length output file. wonder if anyone has any advice
[13:13:14 CET] <JEEB> take out the wrapper and call ffprobe yourself with -v verbose or so on the file? see if it outputs anything?
[13:13:34 CET] <JEEB> ffprobe -v verbose -i BLAH
[13:13:36 CET] <JEEB> should do it just fine
[13:13:47 CET] <JaredCE> ok cool
[13:13:51 CET] <JaredCE> let me try that
[14:29:15 CET] <DHE> termos: the http code calls the tcp code, so take its defaults as well. options for tcp should also work
[14:57:23 CET] <termos> ah nice thanks! looks like it's 5s then (5,000,000 microseconds)
[16:53:03 CET] <matt1122> Wanted to ask again before I spend a lot of time on this ... does anyone know if it's possible using the FFMPEG cmd to change the output aspect ration on the fly, like when the user rotates their phone (The input is live RTMP) ... however right now the output aspect ratio stays stuck at whatever the original aspect ratio was at start
[16:53:39 CET] <matt1122> I don't want to spend a lot of time on this unless someone with more experience believes it's possible :)
[22:46:37 CET] <Etua> Hello, I wanted to calculate Chromaprint ID in my C program and being short on time I just popen fpcalc from libchromaprint-tools but now I wonder whether using one of ffmpeg libraries for that purpose could be prettier without adding a few hundred lines to the code. Is it possible to use library with it's demuxers the easy 'black box' way like with fpcalc so without caring about file format or specific parameters and only take advantage of chromaprint
[22:46:39 CET] <Etua> muxer?
[22:51:27 CET] <Etua> I understand it cannot be oneliner but at the same time ideally I would just need to specify the file path and get the result just like with fpcalc. How close to that ideal can I get?
[00:00:00 CET] --- Thu Jan 30 2020