[Ffmpeg-devel-irc] ffmpeg.log.20160416

[burek]

[00:00:50 CEST] <c_14> I think only vp9 has a lossless mode
[00:01:17 CEST] <c_14> You can use -c:v libvpx-vp9 and then add -lossless
[00:01:58 CEST] <iive> h264 for sure have lossless mode
[00:02:17 CEST] <iive> x264 too
[00:03:55 CEST] <snakeryslug> `ffmpeg -i input.gif -c:v libvpx-vp9 -lossless output.webm` says i need an output file, so lossless must take a parameter.
[00:04:05 CEST] <furq> -lossless 1
[00:06:05 CEST] <snakeryslug> thanks. that worked, but the output webm isn't as good as the input gif.
[00:12:42 CEST] <c_14> rgb to yuv conversion maybe?
[01:31:29 CEST] <durango> Is there any option to cut the sound off once the video/images are done cycling?
[01:31:56 CEST] <pzich> -shortest ?
[01:34:49 CEST] <durango> perfecto, ty
[02:17:24 CEST] <Prelude2004c> hey guys.. good day
[02:17:29 CEST] <Prelude2004c> looking for help with " ] Non-monotonous DTS in output stream 0:0; previous: 8585097540, current: 1197762652; changing to 8585097541. This may result in incorrect timestamps in the output file. "
[02:17:35 CEST] <Prelude2004c> not sure what to do about it
[02:18:03 CEST] <TD-Linux> are you making the input file?
[02:18:36 CEST] <Prelude2004c> http://pastebin.com/wCbhxvDc
[02:18:43 CEST] <Prelude2004c> no, its a UDP source from satelite
[02:19:21 CEST] <Prelude2004c> i have tired to put vsync everywhere i could find to try and fix this..
[02:19:24 CEST] <Prelude2004c> even start_zero i tried
[02:33:05 CEST] <TD-Linux> I suspect that's unavoidable with TS, though someone else might know better
[03:41:45 CEST] <FinboySlick> Hello people.  I've got image buffers in raw formats (mosly RGB) that I would like to put together in a compressed video with libav.  I'm trying to figure out two things:  1- Is there a method to describe the data (how many R, G and B bits, the order they're in) and be given a value that libav would understand as a source image format.  2- convert that source image data to a format that the (say x264) encoding functions will be able to deal with.
[03:43:28 CEST] <FinboySlick> I'm mostly looking for a general direction to look into here.  Google isn't being too helpful.
[03:47:26 CEST] <furq> FinboySlick: are you using the api or the command-line
[03:47:47 CEST] <FinboySlick> furq: I'm writing C code.  So API.
[03:48:20 CEST] <furq> well on the plus side you're in the right channel
[03:48:31 CEST] <furq> on the down side i've barely used the api
[03:48:52 CEST] <FinboySlick> furq: Hehe, I peeked in #devel but managed to read the topic ;)
[03:49:05 CEST] <FinboySlick> I mean #ffmpeg-devel
[03:49:25 CEST] <furq> well you said libav so i thought you might be using the false ffmpeg
[03:49:51 CEST] <FinboySlick> furq: The issue I'm trying to solve isn't really ffmpeg-specific, but I figure there has to be code figuring this stuff up in libav already.
[03:50:27 CEST] <furq> afaik if you were using the cli you'd use rawvideo with the appropriate pix_fmt set (ffmpeg -pix_fmts)
[03:50:33 CEST] <furq> so i assume it's the same in the cli, just much more verbose
[03:50:37 CEST] <furq> the same in the api
[03:52:48 CEST] <FinboySlick> furq: My problem is that "pix_fmt" is essentially a number.  I still need some sort of get_pix_fmt(num_red_bits, num_green_bits, num_blue_bits, order) that will spew out a pix_fmt value.
[03:53:23 CEST] <FinboySlick> Because when I want to encode a frame, I need to specify a pix_fmt number.
[03:55:07 CEST] <FinboySlick> Maybe I should look at the x11grab code, it has to figure this out too.
[04:54:39 CEST] <FinboySlick> av_pix_fmt_desc_get_id() looks like it might do what I want.
[07:39:39 CEST] <jml2> hi
[07:39:57 CEST] <jml2> cannot transcode an .ts file to a .wav , anyone want to jab at this one? thanks
[07:42:06 CEST] <furq> that's not an error message
[07:47:55 CEST] <jml2> thing is i have a problematic hardware encoder --- the options I use for 3 gig files work, but not for 9 gig ones
[07:48:15 CEST] <jml2> (options given to ffmpeg)
[07:48:50 CEST] <jml2>  avconv -analyzeduration 2147483647 -probesize 2147483647 -start_at_zero  -ignore_unknown -copy_unknown -copytb 0  -async 1 -avoid_negative_ts 1 -seek_timestamp 1 -hwaccel vdpau -i source.ts -vn -sn -dn -acodec pcm_s24le -b:a 512k -ac 2 out.wav
[07:49:23 CEST] <furq> have you tried with ffmpeg
[07:49:29 CEST] <jml2> same with ffmpeg
[07:50:02 CEST] Action: jml2 (http://pastebin.com/5cyuLKCx)
[07:51:29 CEST] <furq> can you extract the aac track without transcoding
[07:51:47 CEST] <jml2> (ffmeg - http://pastebin.com/HdpmRd64 )
[07:52:54 CEST] <jml2> same
[07:53:04 CEST] <jml2>  ffmpeg -analyzeduration 2147483647 -probesize 2147483647 -hwaccel vdpau   -i source.ts -vn -acodec alac -bsf:a aac_adtstoasc -ab 384k -ar 48000  sound.m4
[07:53:13 CEST] <jml2> well that's alac,
[07:53:14 CEST] <furq> i said without transcoding
[07:53:16 CEST] <furq> -c:a copy
[07:53:31 CEST] <furq> that might get around the need for analyzeduration and probesize
[07:53:57 CEST] <jml2> i can take out video.. but not audio
[07:54:08 CEST] <jml2> i never tried 5+ gig.. I know under 5 i have no issue
[07:54:25 CEST] <jml2> i havent looked at hbrake and vlc for this 9gig yet
[07:55:23 CEST] <furq> also analyzeduration isn't a uint32 in recent versions of ffmpeg
[07:55:29 CEST] <furq> try setting that to 9000M
[07:55:31 CEST] <jml2> so far vlc is extracting it, awaiting its output
[07:55:47 CEST] <furq> looks like probesize isn't either
[07:55:48 CEST] <jml2> furq, how do i use its max?
[07:55:58 CEST] <jml2> furq, there's an exposed range showing e^
[07:56:30 CEST] <jml2> ( Value -1.000000 for parameter 'analyzeduration' out of range [0 - 9.22337e+18] )
[07:56:36 CEST] <jml2> can I use the max?
[07:56:48 CEST] <furq> no idea, i've never needed to set it
[07:56:53 CEST] <furq> if the file is 9GB then 9000M should work fine
[07:57:21 CEST] <jml2> the .wav output is 16 mb..
[07:57:23 CEST] <jml2> (vlc-out.wav)
[07:57:38 CEST] <jml2> it's able to extract audio..
[07:57:44 CEST] <jml2> so i know its do-able
[07:59:16 CEST] <jml2> its 161mb, and so far its audio works --- (vlc-out.wav)
[07:59:29 CEST] <jml2> looks like that 9000M for ffmpeg-out.wav might do the trick..
[14:08:03 CEST] <TheGreatDoc> Hi
[14:09:00 CEST] Last message repeated 1 time(s).
[14:09:37 CEST] <TheGreatDoc> Im having a little issue with video/audio desycn doing a streaming. The source is a DeckLink SDI 4K
[14:09:53 CEST] <TheGreatDoc> Over time, audio start desyncing with video
[14:14:27 CEST] <CruX|_> hi, can I convert flac -> ape with ffmpeg ?
[14:14:36 CEST] <furq> no
[14:14:42 CEST] <furq> ape is decode only
[14:14:53 CEST] <CruX|_> if not which is the best loosles codec that can use ffmpeg ?
[14:14:55 CEST] <CruX|_> flac ?
[14:15:11 CEST] <furq> flac is what i use
[14:15:30 CEST] <CruX|_> ok thx
[14:16:59 CEST] <furq> CruX|_: ffmpeg -codecs | grep \ .EA..S
[14:17:05 CEST] <furq> will list all lossless audio encoders
[14:21:00 CEST] <CruX|_> great
[14:21:05 CEST] <CruX|_> for i in *; do ffmpeg -y -i "$i" -vn -acodec wavpack "$i.wavpack"; done
[14:21:16 CEST] <CruX|_> btw what's wrong in this command ?
[14:22:37 CEST] <furq> looks fine to me other than the fact that wavpack is normally .wv
[14:22:51 CEST] <furq> and also you're using wavpack in 2016
[14:23:34 CEST] <CruX|_> I want to test all loosless formats
[14:23:48 CEST] <furq> https://xiph.org/flac/comparison.html
[14:23:52 CEST] <furq> there's a nice comparison there
[14:31:17 CEST] <CruX|_> furq: nice thanks
[15:56:18 CEST] <dongs> is there some ffmpeg default filter that can do 3d transform to video
[15:56:29 CEST] <dongs> like, tilt/rotate it slightly
[15:57:04 CEST] <dongs> http://i.ytimg.com/vi/uVPNjEFsuQs/maxresdefault.jpg something like this etc
[16:13:19 CEST] <durandal_170> perspective
[16:13:55 CEST] <dongs> 3d warp / perspective yeah.
[16:14:00 CEST] <durandal_170> and remap if you have maps...
[16:14:13 CEST] <dongs> perspective is a filter?
[16:14:29 CEST] <dongs> nice
[18:08:18 CEST] <evan_> Hi all, is it possible to use ffmpeg to re-encode and stream a video file on the fly from a VPS server to a pc that is behind nat (adsl) if I'm not able to open the ports on my adsl router .. ?
[23:35:06 CEST] <__raven> hi
[23:35:52 CEST] <__raven> how to concatenate several .mov chunks (same type) into one video file ("the right way")? ^^
[23:36:04 CEST] <__raven> i just get "invalid data found when processing input" at the moment
[23:36:33 CEST] <__raven> could that be caused since concat just works for containerless files?
[23:37:30 CEST] <BtbN> remux them all to ts or mkv, and concat those.
[23:38:07 CEST] <__raven> BtbN: pipe possible this way?
[23:38:15 CEST] <BtbN> what?
[23:38:56 CEST] <__raven> doing that in one step: ffmpeg remux | ffmpeg concat.mkv?
[23:39:36 CEST] <BtbN> how would you pipe more then one file? oO
[23:39:52 CEST] <BtbN> and if you only have one file, what exactly do you want to concat?
[23:49:14 CEST] <kepstin> for several 'mov' files which all contain the same audio/video formats, you'd probably want to use the "concat demuxer" https://www.ffmpeg.org/ffmpeg-formats.html#concat-1
[23:49:56 CEST] <BtbN> it doesn't work exactly well for mp4
[23:50:09 CEST] <kepstin> concat demuxer should work perfectly fine in mp4
[23:50:20 CEST] <JEEB> really?
[23:50:25 CEST] <kepstin> it reads each file individually with a proper format-aware demuxer
[23:50:31 CEST] <kepstin> don't get it confused with the concat protocol
[23:50:35 CEST] <JEEB> right
[23:50:45 CEST] <JEEB> protocol, demuxer and filter :P
[23:51:02 CEST] <JEEB> I mostly used the filter because it seemed like the thing that might fuck up least
[23:51:06 CEST] <BtbN> I had even the demuxer screw up on mp4 files. remuxing to ts fixed it
[23:51:31 CEST] <kepstin> well, filter will always work, but requires re-encoding.
[23:52:03 CEST] <BtbN> filter also isn't exactly nice when you want to concat several 100 files
[23:52:18 CEST] <JEEB> yeah, thankfully for my use case that was a requirement and I think I supported up to 5 or so files
[00:00:00 CEST] --- Sun Apr 17 2016