[Ffmpeg-devel-irc] ffmpeg-devel.log.20160810

[burek]

[02:55:45 CEST] <jamrial_> is trac down for anyone else?
[03:00:25 CEST] <michaelni> jamrial_, seems working again, dont ask me what the issue was, i dont know
[03:07:12 CEST] <jamrial_> michaelni: yeah, works now, thanks
[07:58:20 CEST] <kd_dev> Hello!, I had 2 questions:
[07:58:39 CEST] <kd_dev> 1. Is some kind og demosaicing algorithm implemented as filters in ffmpeg?
[07:58:54 CEST] <kd_dev> 2. If not, any ideas on how to proceed about doing so?
[07:59:07 CEST] <kd_dev> found some discussion here: https://ffmpeg.org/pipermail/ffmpeg-devel/2010-July/090493.html
[09:49:12 CEST] <jsebechlebsky> Hi, I'm trying to make a mingw build of FFmpeg to test failing fifo muxer test, but I can't make configure detect pthreads... Does someone have experiece with mingw+pthreads build?
[10:03:49 CEST] <cone-311> ffmpeg 03Paul B Mahol 07master:4506f911858b: avfilter/vf_remap: >8 bit support
[10:19:59 CEST] <Timothy_Gu> jsebechlebsky: mingw or mingw-w64?
[10:20:32 CEST] <Timothy_Gu> the former is very old (i686-pc-mingw32). don't use it
[10:20:52 CEST] <Timothy_Gu> mingw-w64 bundles pthreads methinks
[10:20:58 CEST] <jsebechlebsky> it's mingw-w64, but I've just figured it out...
[10:21:17 CEST] <jsebechlebsky> --disable-w32threads 
[10:22:22 CEST] <jsebechlebsky> enabled pthreads... I've originally supposed pthreads are preferred
[11:38:15 CEST] <cone-311> ffmpeg 03Paul B Mahol 07master:80262d8c86e9: avfilter/avf_showwaves: add sqrt and cbrt amplitude scalers
[12:58:05 CEST] <robUx4> nevcairiel: DXVA2 doesn't support FF_PROFILE_H264_HIGH_10 ?
[12:58:45 CEST] <robUx4> H.264 variable-length decoder, film grain technology has P010 output so I suppose it should
[12:59:02 CEST] <nevcairiel> there is no h264 decoder with 10-bit support
[13:00:05 CEST] <robUx4> you mean hardware decoder 
[13:00:10 CEST] <robUx4> +?
[13:00:17 CEST] <nevcairiel> of course, thats what we're talking about isnt it? :)
[13:00:24 CEST] <robUx4> I have some hi10 samples that decode
[13:00:25 CEST] <robUx4> yes
[13:00:33 CEST] <robUx4> why the P010 output then ?
[13:00:54 CEST] <nevcairiel> never seen a device that offers that
[13:01:54 CEST] <robUx4> the Xbox One does
[13:01:58 CEST] <nevcairiel> which devices even offer FGT? that feature never really took off
[13:02:12 CEST] <nevcairiel> maybe the 10-bit output is related to FGT somehow
[13:02:38 CEST] <robUx4> possibly
[13:04:49 CEST] <robUx4> here's a log from the Xbox (not the S) http://pastebin.com/aMQ1GTCX
[13:06:13 CEST] <nevcairiel> or AMD gave every device P010 output for some reason, who knows
[13:07:44 CEST] <nevcairiel> in any case the specification for those devices clearly states the profiles which are supported, and main10 is not among them
[13:08:01 CEST] <robUx4> is there an easy to force the HW decoding with hi10 sources ?
[13:08:08 CEST] <robUx4> or it's just too different ?
[13:10:07 CEST] <nevcairiel> might be able to trick avcodec into offering that
[13:43:46 CEST] <Compn> nevcairiel : i thought there was some nvidia cards that did 10bit h264 
[13:43:50 CEST] <Compn> but probably i am wrong
[13:46:01 CEST] <Compn> oh no, guess i'm wrong
[13:46:07 CEST] <Compn> bunch of software decoding going on haha
[13:46:59 CEST] <durandal_1707> you always wrong
[13:52:04 CEST] <BtbN> Pascal should support that though?
[13:52:22 CEST] <nevcairiel> what makes you thin that, h264 support hasnt changed for years
[13:53:37 CEST] <BtbN> So it supports 10bit HEVC decoding, but not h264? oO
[13:54:06 CEST] <nevcairiel> indeed
[13:54:27 CEST] <nevcairiel> codecs are quite different, assuming one means the other is not how it works .D
[13:54:56 CEST] <BtbN> would have expected that they share a lot of the decoding hardware, so if they add 10bit, it would work for both.
[14:03:13 CEST] <Compn> probably the 8bit h264 chips are cheap
[14:03:20 CEST] <Compn> while the 10bit h264 chips cost another nickel
[14:04:18 CEST] <nevcairiel> its part of the main GPU chip so its integrated into there somewhere, i dont know if they source those logic circuits from somewhere else or develop them tho
[14:05:10 CEST] <nevcairiel> but there is really no point for them in264 10-bit support, there is literally zero commercial content
[14:09:00 CEST] <robUx4> AVCHD is not 10 bits ?
[14:09:30 CEST] <nevcairiel> isnt that blu-ray compatible video
[14:09:45 CEST] <robUx4> ah no AVCHD is 8 bits
[14:23:25 CEST] <durandal_1707> lot of love in ffmpeg-user lately
[15:35:31 CEST] <robUx4> nevcairiel: for the record hi10p decoding doesn't work, neither in P010 nor NV12
[18:32:15 CEST] <omerjerk> Hi
[18:32:25 CEST] <omerjerk> I've added alsenc here - https://github.com/omerjerk/FFmpeg/blob/alsenc_first_patch/libavformat/alsenc.c
[18:32:36 CEST] <omerjerk> Still I'm not able to encode to out.als
[18:32:39 CEST] <omerjerk> for e.g. 
[18:32:57 CEST] <omerjerk> ffmpeg -i in.wav -c:a als out.als gives error
[18:33:13 CEST] <omerjerk> Error = Could not write header for output file #0 (incorrect codec parameters ?): Operation not permitted
[18:33:38 CEST] <omerjerk> Is it even possible to output into .als file or not?
[18:39:10 CEST] <durandal_1707> omerjerk: only if you wrote als muxer 
[18:40:38 CEST] <durandal_1707> write into mp4
[18:41:48 CEST] <cone-931> ffmpeg 03Burt P 07master:16ea0bca1473: avfiltergraph.c: restore disabling of auto conversions
[18:43:40 CEST] <durandal_1707> so we gonna remove hacks which blocks codecpar transformation?
[18:45:30 CEST] <jamrial_> can that be done without negatively affecting output files?
[18:46:00 CEST] <durandal_1707> hacks are for invalid files anyway
[18:47:30 CEST] <jamrial_> poke michaelni, afaik he added most of those hacks and he probably has example use cases or files
[18:51:00 CEST] <durandal_1707> michaelni: see above?
[18:57:30 CEST] <michaelni> anything i should test ? but i dont have a too good feeling with ffmpeg->codecpar to be "just working"
[19:16:54 CEST] <jamrial_> michaelni: in ffmpeg.c transcode_init(), the stream copy section does some crazy calculations only for avi/mp4 output using avctx->ticks_per_frame and avctx->time_base that with codecpar will not be possible
[19:17:41 CEST] <jamrial_> not sure how you could test aside from dropping that chunk when converting ffmpeg to codecpar and seeing how it ends up muxing that kind of file
[19:28:15 CEST] <michaelni> speaking about ticks_per_frame, theres also [FFmpeg-devel] [PATCH 2/2] doc/examples/muxing: Fix packet duration
[21:02:48 CEST] <durandal_170> omerjerk: still around?
[21:02:58 CEST] <omerjerk> yes.
[21:03:07 CEST] <omerjerk> There's some bug in the ALS muxer code.
[21:03:13 CEST] <omerjerk> I'm trying to fix that.
[21:03:35 CEST] <durandal_170> what about into mp4 muxer?
[21:03:44 CEST] <omerjerk> that works perfect.
[21:03:56 CEST] <durandal_170> nice
[21:13:59 CEST] <BtbN> uhm, that sounds kinda expected.
[21:14:44 CEST] <durandal_170> but it works with mencoder I think
[21:18:43 CEST] <iive> i hope these are for different avi files
[21:19:06 CEST] <durandal_170> :) job for Carl
[00:00:00 CEST] --- Thu Aug 11 2016