[Ffmpeg-devel-irc] ffmpeg-devel.log.20151223

[burek]

[00:02:41 CET] <BtbN> I wonder if libyami ever delays frame output. I can't see anything that would indicate that
[00:04:06 CET] <BtbN> seems like the output size is just 0 then
[00:27:24 CET] <BtbN> Well, it does work now.
[00:27:27 CET] <BtbN> But it's horribly slow
[00:27:37 CET] <BtbN> encoding 1080p h264 at 16fps
[00:37:40 CET] <Rokam> rcombs: it did work with ffplay. But not with Kodi
[00:38:06 CET] <rcombs> Rokam: are you setting the "headers" option instead of the "cookies" one
[00:39:44 CET] <Rokam> I did use cookies with ffplay
[00:39:59 CET] <Rokam> and it works
[00:40:38 CET] <Rokam> Now I need to debug kodi with ffmpeg to check why it doesn't update the cookies
[00:41:21 CET] <Rokam> ffplay.exe -cookies "WMPTOKEN_52dfc02cdd23810590000f57=173846766e9c8c8dc4c4121176fb2c6c" "http://live.hls.globo.com/dvr/off24ha/master-playlist.m3u8?h=0411450824875571523425814508283546383372468UipeYDqVOJJ1fPbkdpl1VA&k=html5&a=F&u=52dfc02cdd23810590000f57_605fd6ba-6320-4a5a-b10f-065e5363421d"
[00:43:56 CET] <BtbN> libyami itself copies the frame, again. No wonder it's horribly slow.
[01:04:09 CET] <prelude2004c> hey everyone.. any idea how we can fix the vdpau issue and h264 ?
[03:15:15 CET] <michaelni> ubitux, ive added a aarch64 osx fate client that should force the use of gas-preprocessor, should show up on the next run
[03:23:39 CET] <jamrial> does anyone have an aac sample with ipd/opd data, or know how to create one?
[10:23:32 CET] <cone-393> ffmpeg 03Hendrik Leppkes 07master:b942845beedc: avcodec/libschroedingerenc: add missing AVClass to private context
[12:01:24 CET] <main> hey guys i want to convert a video so i can import it to cinelerra. after converting the video seems to be in the wrong stream so that its not recognised X'(  stream #0.0 is in  #0.1 and stream #0.1 is in #0.2 after conversation how can i set the option whitch allowes me to choose which inputstream goes to whitch outputstream?
[12:02:25 CET] <ubitux> you're looking for the -map option (order in cli defines order in output), but you're on the wrong channel, go to #ffmpeg
[12:02:37 CET] <main> ty
[12:30:06 CET] <kierank> ironically having the same channel map discussion in two places on the mailing list
[12:48:14 CET] <cone-393> ffmpeg 03Paul B Mahol 07master:57787f5ef823: avcodec/s302menc: comment out allowed channel layouts
[13:36:46 CET] <ubitux> alright, rewriting srt demuxer again to handle a maximum of madness
[14:04:16 CET] <Compn> ehe
[14:04:48 CET] <Compn> arent there any srt demuxer libs out there ?
[14:04:51 CET] <Compn> free for the taking ?
[14:10:16 CET] <ubitux> ...
[14:10:31 CET] <ubitux> Compn: demuxing is like 200 lines of code
[14:10:50 CET] <ubitux> any better idea for external libs?
[14:11:19 CET] <ubitux> i'm just changing the logic to be more heuristic based, and cover more cases
[14:26:29 CET] <nevcairiel> I doubt there actually is a set library :P
[14:26:37 CET] <nevcairiel> srt*
[15:30:52 CET] <cone-393> ffmpeg 03wm4 07master:a0ec4aebabe1: avformat/subtitles: treat negative duration like unknown duration
[16:10:50 CET] <Prelude2004c_> hey everyone . still looking for some assistance with vdpau
[16:10:54 CET] <Prelude2004c_> anyone around ?
[16:12:35 CET] <J_Darnley> Yes but I can't help you with that.
[16:13:43 CET] <Prelude2004c_> :) thank you
[16:13:49 CET] <Prelude2004c_> its a tough issue
[16:13:59 CET] <Prelude2004c_> i can't seem to get anyone who knows it well enough to figure out the issue
[16:14:49 CET] <Prelude2004c_> i think ffmpeg has some bug in the code that is causing this .. doesn't make too much sense
[16:15:06 CET] <Prelude2004c_> decoding on h264 sources may not be working
[17:05:16 CET] <ubitux> michaelni: can you upload http://b.pkh.me/ticket5032-rrn.srt and http://b.pkh.me/madness.srt to fate-samples/sub?
[17:05:27 CET] <ubitux> 36a2ea0edb6eddd21425c6f4e412fccc  ticket5032-rrn.srt
[17:05:29 CET] <ubitux> 713094eecb94d277666edaaf09931086  madness.srt
[17:06:01 CET] <ubitux> patch fixing those is done, i'll submit later with a fate test
[17:16:23 CET] <ubitux> actually,
[17:16:25 CET] <ubitux> 713094eecb94d277666edaaf09931086  madness.srt
[17:16:54 CET] <ubitux> i mean b1420e94e5bfad62cfd03a33af060074
[18:07:52 CET] <michaelni> ubitux, uploaded
[18:24:34 CET] <cone-393> ffmpeg 03Ganesh Ajjanagadde 07master:520a5d33f0ea: lavfi/af_aemphasis: remove unnecessary complex number usage
[18:24:35 CET] <cone-393> ffmpeg 03Ganesh Ajjanagadde 07master:e29db08cf7f6: lavu/libm: add exp10 support
[19:18:18 CET] <ubitux> michaelni: thanks
[19:40:08 CET] <cone-393> ffmpeg 03Paul B Mahol 07master:3689b58a5a1c: avfilter/af_biquads: display clipping warnings once per filtered frame
[19:59:52 CET] <Properrr_> Hello to all. May somebody help me? I want get PSI tables from demuxer of ffmpeg
[20:00:05 CET] <Properrr_> What should i do?! ffmpeg 2.8
[20:01:03 CET] <Properrr_> PAT, CAT, PMT, EIT tables...
[21:23:40 CET] <JEEB> what kind of time line does FFmpeg have regarding the recent changes by elenril into the internals? if I have patches that would have to be partially redone with those changes, should I start poking at it now, or rather post them as-is?
[21:32:52 CET] <durandal_1707> ubitux: do you know how to use asendcmd filter?
[21:37:45 CET] <Compn> JEEB : only nevcairiel is doing merges iirc, so ask him
[22:39:19 CET] <cone-393> ffmpeg 03Michael Niedermayer 07master:e42e0b11f1df: avfilter/x86/vf_maskedmerge: Clear upper part of width
[22:46:33 CET] <durandal_1707> michaelni: thanks!
[23:15:12 CET] <tmm1_> is anyone aware of work to optimize the yadif implementation for arm64?
[23:24:42 CET] <J_Darnley> I haven't seen any chatter about it
[23:24:55 CET] <J_Darnley> but then I tend to ignore arm discussions
[23:27:35 CET] <fritsch> tmm1_: weren't there recently some improvement for w3fdif (also a very good deinterlacer)
[23:29:57 CET] <jamrial> yes, but for x86 afaik
[23:30:11 CET] <fritsch> i need to look into that again
[23:30:18 CET] <fritsch> in kodi we have yadif sw deinterlacing
[23:30:32 CET] <fritsch> last time I added w3fdif even a core i5 3 ghz was too slow
[23:30:35 CET] <fritsch> for 1080i50 live tv
[23:36:54 CET] <tmm1_> yea they both perform well on x86 due to SIMD optimizations, but not so much on arm64
[23:37:38 CET] <tmm1_> i started on a patch but i don't have much experience with NEON, so i'm looking for someone to collaborate with, or even sponsor the development of it
[23:39:10 CET] <fritsch> arm chips like the PI have dedicated firmware methods for deinterlacing
[23:39:19 CET] <fritsch> amlogic do it themselves
[23:39:26 CET] <fritsch> you don't really have an influence
[23:39:43 CET] <fritsch> depends on the use case
[23:41:22 CET] <tmm1_> ah i didn't realize some chips had dedicated deinterlacing built in
[23:41:58 CET] <tmm1_> i'm just interested in making the yadif s/w deinterlacer faster w/ SIMD, just like it is on x86
[23:42:04 CET] <fritsch> oki
[23:42:10 CET] <fritsch> yeah, that's cool
[23:42:51 CET] <fritsch> perhaps you can "copy" the relevant ideas used for the x86 code
[23:43:02 CET] <fritsch> to use the neon instructions similarly
[23:47:42 CET] <tmm1_> yep that's what i'm trying to do, write a neon asm version based on the x86 and C versions
[23:48:40 CET] <tmm1_> are there any docs or papers etc on the yadif algorithm itself? would help to understand what its doing before i start reimplementing it
[23:48:58 CET] <fritsch> i think the source code is the best doku
[23:49:58 CET] <fritsch> but the author is still here :-)
[23:50:14 CET] <fritsch> so I am sure you get answers when you get detailled questions
[23:53:59 CET] <J_Darnley> tmm1_: I once drew an ascii diagram of what yadif did
[23:54:08 CET] <J_Darnley> Do you want me to see if I can find it?
[23:55:29 CET] <tmm1_> J_Darnley: yea that'd be great if you have it handy
[23:55:39 CET] <J_Darnley> Not handy but I'll search for it
[00:00:00 CET] --- Thu Dec 24 2015