[Ffmpeg-devel-irc] ffmpeg.log.20150521

[burek]

[02:31:09 CEST] <bornpilot> if I had to iFrame only video h.264 and I wanted to concat them would I need to covert them to raw intermediate format?
[02:31:48 CEST] <bornpilot> ^that is I have two of them :)
[03:27:51 CEST] <ac_slater> hey all. Anyone any simple decoding code that decodes at the source framerate?
[03:29:51 CEST] <ac_slater> sure, I can probably say 30fps == some ms delay. But is there a better way?
[03:30:03 CEST] <klaxa> use the -re flag
[03:30:19 CEST] <ac_slater> I mean using libavformat/codec
[03:30:27 CEST] <ac_slater> klaxa: but yes, -re in ffmpeg.c does it
[03:30:47 CEST] <ac_slater> they have a mess of structures/abstractions that do PTS/DTS calculations
[03:30:58 CEST] <ac_slater> (and mess refers to size, not quality)
[03:35:54 CEST] <klaxa> ac_slater: have a look at: https://git.videolan.org/?p=ffmpeg.git;a=blob;f=ffmpeg.c;h=a89ae399bfa1d04dc0b414f67713ef48933d7d63;hb=HEAD#l3498
[03:36:07 CEST] <klaxa> f->rate_emu is true for an input file if -re was used
[03:37:21 CEST] <ac_slater> klaxa: right, searching for rate_emu you see some functions that calculate some PTS vs now then return EAGAIN. Makes sense except figuring out how st->start_time is calculated is pretty difficult :(
[03:37:54 CEST] <ac_slater> hmm  actually, it's not that bad
[03:38:10 CEST] <klaxa> see: https://git.videolan.org/?p=ffmpeg.git;a=blob;f=ffmpeg.c;h=a89ae399bfa1d04dc0b414f67713ef48933d7d63;hb=HEAD#l2612
[03:38:36 CEST] <ac_slater> interesting
[03:38:48 CEST] <ac_slater> I was looking at 3631
[03:42:05 CEST] <ac_slater> klaxa: so maybe you know. I have a non-b frame source, so I really only need PTS. The lines you references only use a wall clock to determine if a delay should happen. The `av_recale` at like 3502 is confusing as I have no DTS
[03:42:13 CEST] <ac_slater> or, rather, DTS == PTS. Should it still work?
[03:43:28 CEST] <klaxa> hmm... apparently so in that case? i don't know to be honest
[03:43:43 CEST] <ac_slater> you were my only hope! ;)
[03:43:54 CEST] <ac_slater> I've been struggling with PTS/DTS for a few months now
[04:16:57 CEST] <ac_slater> I guess I wanted to use a decoded AVPacket
[04:17:13 CEST] <ac_slater> I guess I wanted to use a decoded AVPacket's PTS and DTS to calculate delay... but seems like ffmpeg.c just uses wall clock
[04:17:32 CEST] <ac_slater> I dont know how to translate PTS ticks to wall clock ticks :(
[04:28:03 CEST] <Nemo377307289> Could ffmpeg add fade effect for watermark picture?
[04:28:04 CEST] <Nemo377307289> I means, increase or reduce the watermark alpha dgree.
[04:28:04 CEST] <Nemo377307289> Thanks a lot.
[05:26:18 CEST] <Score_Under> When recording the desktop and stereo mix, how can I avoid visual tearing and sound crackling? I've recorded this to demonstrate some game crashing ( https://youtu.be/qAtzdSsfRGc ) but it is very poor quality both visually and audibly. The crackling artifacts in the sound don't appear in arecord. I used "ffmpeg -video_size 1920x1080 -f x11grab -thread_queue_size 1024 -i :0+1024+0 -f alsa -thread_queue_
[05:26:24 CEST] <Score_Under> size 8192 -ac 2 -i front -c:a copy -c:v huffyuv test.mkv" to grab this. The huge thread queue size was me getting fed up with thread queue size warnings and just turning it way up. It printed a lot of lines like "Past duration 0.999046 too large" too.
[09:45:47 CEST] <pagios> hello i am trying to capture video from a usb capture card and nothing happens: http://pastie.org/10199919
[09:45:50 CEST] <pagios> it just hangs
[09:45:51 CEST] <pagios> any idea?
[11:14:58 CEST] <Jonas__> I'm looking for details on what general variables there are and what functions I can use to transform them, ie like in rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
[11:15:13 CEST] <Jonas__> is there a documentation page for it somewhere or a couple of search terms I can use to find it?
[11:16:35 CEST] <Jonas__> (I want to rotate a video 90 degrees for the first 40-or-so seconds, then interpolate that from 90 to 0 and keep 0 for the rest of the video)
[11:20:29 CEST] <Jonas__> found it, https://ffmpeg.org/ffmpeg-utils.html
[14:44:14 CEST] <halogene> Is there by any chance an estimate about when 2.6.4 will be released? I just wondered if I might be able to make use of a bugfix in the near future without having to compile from git
[14:45:19 CEST] <halogene> I know that the answer is usually "it will be released when it's ready"... but maybe it's like just about to be released or smth
[14:45:38 CEST] <halogene> in which case I would wait for it :o)
[14:45:51 CEST] <BtbN> 2.6 is branched and maintained seperately. bugfixes will be most likely backported though
[14:46:19 CEST] <BtbN> In fact, only bugfixed are backported, no new features will be added anymore.
[14:46:38 CEST] <iive> the /topic doesn't list 2.6.3 as released
[14:46:48 CEST] <halogene> yes, I noticed and wondered
[14:46:59 CEST] <BtbN> The topic is wrong
[14:47:03 CEST] <BtbN> or outdated rather
[14:47:06 CEST] <halogene> https://ffmpeg.org/download.html shows 2.6.3
[14:47:11 CEST] <BtbN> 2.6.3 was tagges 4 days ago
[14:47:14 CEST] <BtbN> *d
[14:47:47 CEST] <halogene> BtbN: I don't really understand your answer about 2.6 being branched and separately maintained
[14:49:17 CEST] <halogene> I have 2.6.3 on my system (arch linux), I would expect 2.6.4 to be available upon its release?
[14:50:18 CEST] <iive> if 2.6.3 have been released just few days ago... it is probably going to take a while for next release
[14:50:30 CEST] <halogene> In the bug report it reads along the lines of "fixed, will probably be in 2.6.4"
[14:50:31 CEST] <halogene> oh ok
[14:50:33 CEST] <halogene> thank you
[14:50:36 CEST] <iive> unless there is a major bug, security etc...
[14:50:50 CEST] <halogene> that's all information I needed. Thanks!
[14:50:51 CEST] <iive> you could ask michaelni , if you want to be sure
[14:51:36 CEST] <halogene> nah, I just wanted to know if it was common knowledge around here that 2.6.4 is about to be released in the next few days
[14:52:02 CEST] <iive> are you sure the bug is not in .3 ?
[14:52:15 CEST] <halogene> how you mean, "not in .3"?
[14:52:19 CEST] <iive> 2.6.3
[14:52:22 CEST] <halogene> I encountered the bug in 2.6.3
[14:52:38 CEST] <halogene> researched, found a fix from 2 days ago with comment "will be in 2.6.4"
[14:52:41 CEST] <iive> i see
[14:52:55 CEST] <halogene> so I wondered if that's something to be available very soon :o)
[14:53:53 CEST] <halogene> (this one: http://trac.ffmpeg.org/ticket/4568)
[14:54:05 CEST] <halogene> anyway, thanks for taking your time to elaborate.
[14:54:33 CEST] <iive> well, if i am to estimate from the current releases and that major releases are 3 months apart, i'd say that there is minor release every month
[14:56:03 CEST] <halogene> :oD
[14:58:31 CEST] <spaam> good guy iive fixing stuff
[15:44:13 CEST] <spectromas> what is the difference between the sample_fmts with 'p' and those without 'p', like s16 and s16p ?
[15:44:51 CEST] <iive> my guess is that p stands for planar. meaning samples from same channel are grouped together. aka
[15:45:21 CEST] <iive> instead of having "lr, lr, lr ..." you have "lll...., rrr..."
[15:45:50 CEST] <spectromas> thanks iive, I'm not sure what that means in relation to this so I guess I don't need it
[15:45:51 CEST] <iive> where l and r are 2 byte signed words
[15:46:19 CEST] <iive> it's just internal way to represent audio data.
[15:47:07 CEST] <spectromas> thank you :)
[15:49:55 CEST] <nyuszika7h> is it possible to remove silence from the end of m4a files (preferably without reencoding)?
[17:31:01 CEST] <Fyr> how do I convert MID into m4a?
[17:51:34 CEST] <RobotsOnDrugs> Fyr: you'll have to feed the midi into an instrument first if you want sound
[17:52:08 CEST] <Fyr> ok, is there a workaround in ffmpeg?
[17:52:27 CEST] <RobotsOnDrugs> midi is just notes and some parameter information
[17:52:27 CEST] <Fyr> I've managed to have the file opened and played with a player.
[17:53:15 CEST] <RobotsOnDrugs> the player uses something to play the notes, but i couldn't tell you what
[17:53:30 CEST] <Fyr> but I have no idea how to open it with ffmpeg.
[17:53:57 CEST] <RobotsOnDrugs> well, it's not audio
[18:13:09 CEST] <Anoia> Fyr: youcan interactively convert it using audacity
[18:13:48 CEST] <Fyr> ok
[19:07:30 CEST] <DeadSix27> can someone properly explain what "Past duration 0.999992 too large" means, and why I get that error? (full output and command line -> http://pastebin.com/raw.php?i=fPKZ43A3 )
[20:35:01 CEST] <toomuchtvrotsurb> hi
[21:21:03 CEST] <rjp421> using the latest git static build on centos6, how can i optimize this cmd http://pastebin.com/vN7PP0DM so it still creates livestream1 2 and 3 in the same order, but duplicates the source stream without wasting resources out to both ustream and livestream3?
[21:22:16 CEST] <rjp421> i may need to eventually specify a bitrate for livestream3, and leave live/livestream the same as the source
[22:22:48 CEST] <Emn1ty> Curious if anyone knows of a way to do video processing on a gpu?
[22:23:11 CEST] <Emn1ty> using ffmpeg / avconv
[22:23:32 CEST] <Emn1ty> may be a dumb question, just wondering if it's possible
[22:25:55 CEST] <toomuchtvrotsurb> what kind of video processing?
[22:26:01 CEST] <toomuchtvrotsurb> certain types for sure
[22:29:41 CEST] <Emn1ty> processing different formats into mp4
[22:29:47 CEST] <Emn1ty> as well as applying watermarks
[22:29:57 CEST] <Emn1ty> toomuchtvrotsurb ^
[22:37:22 CEST] <toomuchtvrotsurb> i m wrong :( - ffmpeg faq "Regardless, FFmpeg today doesn't support any means of GPU encoding, outside of libx264."
[22:39:32 CEST] <Anoia> does anyone here use the zeranoe win32 builds in visual studio?
[22:39:53 CEST] <Emn1ty> hm...
[22:39:55 CEST] <Emn1ty> thanks toomuchtvrotsurb
[22:40:10 CEST] <Emn1ty> that's what I found, too but wasn't sure if someone here might know better
[23:27:14 CEST] <rjp421> using the latest git static build on centos6, how can i optimize this cmd http://pastebin.com/vN7PP0DM so it still creates livestream1 2 and 3 in the same order, but duplicates the source stream without wasting resources out to both ustream and live/livestream?
[00:00:00 CEST] --- Fri May 22 2015