[Ffmpeg-devel-irc] ffmpeg.log.20200626

[burek]

[00:01:10 CEST] <another> could also try different tags with mpeg4
[00:21:14 CEST] <saurabh702> `This channel is publicly logged`, what does this indicate? (irc noob here)
[00:28:44 CEST] <another> that the channel is publicly logged https://lists.ffmpeg.org/pipermail/ffmpeg-devel-irc/
[00:30:18 CEST] <another> huh.. are there no logs since february?
[00:31:02 CEST] <saurabh702> another: but that is the ffmpeg-devel channel, is this channel also logged similarly?
[00:31:26 CEST] <another> look closely
[00:34:57 CEST] <saurabh702> ohk, but not logged since feb?
[00:37:53 CEST] <jerome-> I have receive a stream of jpeg. those jpeg are taken live. but the frequency of those jpeg is not accurate. Is it possible to create a mjpeg using the time when those jpeg was taken (based on the timestamp of the file, or encode the time in the filename or something else)?
[00:41:01 CEST] <saurabh702> michaelni: is the channel still being logged publicly? (march 2020 onwards)
[00:48:20 CEST] <reepca> is there a way to force mpeg to use lossless (or minimal-loss) compression with lossy codecs? mjpeg is the only format that this player seems to accept, but I have plenty of space to spare and the quality is poor
[00:48:28 CEST] <reepca> s/mpeg/ffmpeg
[00:50:50 CEST] <reepca> also, is there a way to read the documentation offline?
[00:51:03 CEST] <another> man ffmpeg-all
[00:52:16 CEST] <another> see if the device supports a lossless codec
[00:52:24 CEST] <another> i highly doubt it does
[00:54:22 CEST] <cehoyos> Use -qscale 2 with the jpeg encoder to get higher quality
[00:54:29 CEST] <cehoyos> (possibly -qscale 2 -mbd 2)
[00:55:31 CEST] <cehoyos> Note that most quality loss comes through subsampling: If your input is yuv(j)420p, there is nothing you can do, but if the input uses a different colour space, you may be able to improve output quality
[00:55:57 CEST] <cehoyos> (You cannot see a visual difference between input and output with -qscale 2)
[01:01:02 CEST] <reepca> thanks, that vastly improved the quality
[01:01:15 CEST] <reepca> -qscale 2 that is
[01:02:46 CEST] <JustLandedOnMars> so can I use both options: "-crf 20 -vb 2000k" or they are mutually exclusive ?
[01:03:47 CEST] <another> depends on the encoder
[01:10:29 CEST] <JustLandedOnMars> h264
[01:10:45 CEST] <JustLandedOnMars> I use: -c:v h264 -preset fast -crf 20 -vb 2000k
[01:10:50 CEST] <another> that's a codec not an encoder
[01:10:58 CEST] <JustLandedOnMars> should I remove "-vb 2000k" or "-crf 20" ?
[01:11:08 CEST] <JustLandedOnMars> what is encoder /
[01:12:59 CEST] <another> probably libx264
[01:16:04 CEST] <JustLandedOnMars> so should I remove "-vb 2000k" or "-crf 20" ?
[01:16:48 CEST] <c_14> https://trac.ffmpeg.org/wiki/Encode/H.264
[01:16:54 CEST] <c_14> ^read that
[01:17:56 CEST] <JustLandedOnMars> yes, I read it, and this is where I got "-crf 20" from, and I added into my command, now I'm thinking should I remove "-vb 2000k" if it has "-crf 20" in it
[01:18:25 CEST] <JustLandedOnMars> I will try with and without and see check sum of output files if they differs
[01:18:26 CEST] <c_14> yes, with x264 it's one or the other
[01:18:44 CEST] <JustLandedOnMars> and which one has preference if I use both ?
[01:19:10 CEST] <c_14> that I don't know
[01:27:18 CEST] <JustLandedOnMars> if I use both "-crf 20 -vb 2000k" and if I use one "-crf 20" the output video files checksums are the same, so "-crf 2" has preference that means
[01:33:53 CEST] <cehoyos> reepca: I believe you should also use -mbd 2
[01:36:37 CEST] <aram535> Can anyone helping me with a concat with an odd audio map please. I would like to have 1 mkv (h264 is possible) with just two English audios (one of them will be 5 minutes short). What I have are 2 files: part 1 is AVC + Audio (0 - French, 1 - English, 2 - English) + Subs ... then part 2 is MPEG + Audio ( 0 - French, 1 - English .... no 2nd English track).  Part 2 is only ~5m long ... no idea why they did this.  Here is the ffmpeg -i out
[01:36:37 CEST] <aram535> puts: https://pastebin.com/GBHr1C9a
[01:38:54 CEST] <JustLandedOnMars> so I tryed presets "fast" it took 8.2 seconds and "ultrafast" it took 1.2 seconds, why such drastic difference, what that "ultrafast" preset doing such that its so fast ?
[01:41:51 CEST] <michaelni> saurabh702, it should be logged, but it seems something went wrong, ill mail mladen again
[01:42:41 CEST] <giaco> which linux video player client in your opinion can give me the shortest latency over rtsp?
[01:45:45 CEST] <H-M-H> giaco: You could try `mpv --profile=low-latency --no-cache --untimed`.
[01:48:11 CEST] <giaco> H-M-H: great!
[01:48:17 CEST] <giaco> thanks a lot, it's perfect
[01:48:30 CEST] <aram535> Nevermind on my question ... I did it one part at a time and removed what I didn't want, converted the 2nd part to avc and then concat works fine
[01:57:42 CEST] <c_14> JustLandedOnMars: look at the filesize difference between the two
[01:57:45 CEST] <c_14> that's what it's doing
[01:58:02 CEST] <c_14> (assuming you're using crf)
[05:08:57 CEST] <m00n> so im trying to convert some .wavs to flac and add music metadata from another file (either read it from text or from another some), i can get them to convert but im not sure how to add the metadata
[05:09:05 CEST] <m00n> any recommendations?
[05:14:07 CEST] <furq> m00n: -i foo.wav -i bar.mp3 -map 0 -map_metadata 1 baz.flac
[05:14:54 CEST] <furq> https://www.ffmpeg.org/ffmpeg-formats.html#Metadata-1
[05:14:59 CEST] <furq> also that if you want to use plaintext
[05:26:37 CEST] <m00n> thats awesome
[05:26:50 CEST] <m00n> thanks furq
[05:49:51 CEST] <m00n> is there a better way to filter which tags i copy other than the bash piping im doing now?
[05:52:36 CEST] <m00n> like, im trying to get it formatted to what ffmpeg will understand per that link, because i dont want all the tags from the source, just album/artist/disc
[07:04:22 CEST] <MatCat> I have ffmpeg on a pi UDP streaming to my home server, on it I am trying to run: ffmpeg -f mpegts -i udp://lovecat.m42.pw:9969 -vcodec copy -acodec copy -f flv rtmp://a.rtmp.youtube.com/live2/ BUT I am getting:  Cannot open connection tcp://a.rtmp.youtube.com:1935
[07:04:24 CEST] <MatCat> Any ideas?
[07:05:01 CEST] <MatCat> also non-existing PPS 0 referenced
[07:05:05 CEST] <MatCat> decode_slice_header error
[07:05:08 CEST] <MatCat> no frame!
[07:28:11 CEST] <MatCat> HOLY CRAP I manage to do it, I manage to get an image overlay working! https://www.youtube.com/watch?v=gSo2UaM3Ckk
[07:28:31 CEST] <MatCat> But its encoding faster then it should
[07:28:52 CEST] <MatCat> dunno how a live UDP stream can encode at faster then 1x
[09:29:35 CEST] <pagios> hi