[Ffmpeg-devel-irc] ffmpeg-devel.log.20180807

[burek]

[05:29:58 CEST] <cone-970> ffmpeg 03Zhong Li 07master:6088b7b0373f: lavc/qsvenc: expose qp of encoded frames
[05:29:58 CEST] <cone-970> ffmpeg 03Zhong Li 07master:6434b84460bb: fftools/qsv: add extra_hw_frames support
[06:32:21 CEST] <cone-970> ffmpeg 03Jacek Jendrzej 07master:2f45378ba144: avformat/dashdec: fix for ticket #7336
[06:32:22 CEST] <cone-970> ffmpeg 03Jacek Jendrzej 07master:0cf5e6b5b4e1: avformat/dashdec: Fix memleak when resolve_content_path
[08:42:20 CEST] <cone-970> ffmpeg 03Steven Liu 07master:d54ae9b782c8: avformat/dashdec: remove unused functions and process code
[10:05:20 CEST] <cone-970> ffmpeg 03Steven Liu 07master:d37125fb910e: avformat/flvdec: add flv_ignore_prevtag option into flvdec
[10:05:21 CEST] <cone-970> ffmpeg 03Steven Liu 07master:3b99bb388903: avformat/flvdec: reindent code for previous commit
[10:05:22 CEST] <cone-970> ffmpeg 03Steven Liu 07master:243ecadad5e5: avformat/flvdec: add flv_full_metadata option into flvdec
[18:30:19 CEST] <cone-091> ffmpeg 03Sergey Lavrushkin 07master:4eb63efbdaea: libavfilter: Adds on the fly generation of default DNN models for tensorflow backend instead of storing binary model.
[18:30:20 CEST] <cone-091> ffmpeg 03Sergey Lavrushkin 07master:9d87897ba84a: libavfilter: Code style fixes for pointers in DNN module and sr filter.
[18:30:21 CEST] <cone-091> ffmpeg 03Sergey Lavrushkin 07master:a66e74306a36: libavfilter: Fixes warnings for unused variables in dnn_srcnn.h, dnn_espcn.h, dnn_backend_tf.c.
[18:34:57 CEST] <atomnuker> ...less of evil is still evil, there's still binary weights
[18:43:25 CEST] <BtbN> Isn't that stuff also built by default? And it's insanely large?
[19:00:44 CEST] <durandal_1707> anybody interested in reversing SIMD code to C for me?
[19:02:12 CEST] <atomnuker> paste it here, as long as it isn't arm simd code someone will figure it out
[19:03:11 CEST] <iive> pastebin might be better idea ;)
[19:06:52 CEST] <durandal_1707> https://pastebin.com/YeUqpBZn
[19:07:53 CEST] <durandal_1707> it is idct very xvid alike, but looks like uses different constants, and causes wrong interframe decoding
[19:07:54 CEST] <kierank> probably copy pasted skal idct
[19:08:16 CEST] <durandal_1707> kierank: skal?
[19:08:34 CEST] <atomnuker> main xvid maintainer
[19:10:49 CEST] <durandal_1707> i nowhere see that xvid idct uses -2.031250476844434 constant
[19:39:55 CEST] <iive> it seems to be integer code, not float, and the pastebin doesn't show the constants value.
[19:43:30 CEST] <iive> durandal_1707, have you tried all idct variants that exist in ffmpeg?
[19:45:22 CEST] <durandal_1707> iive: xvid, simple, faan, int
[19:47:54 CEST] <iive> there are more
[19:48:25 CEST] <iive> some are codec specific
[19:48:49 CEST] <durandal_1707> yea, they are much harder to reach
[19:53:27 CEST] <durandal_1707> right, code it is using integers, and 49152 constant among bunch of others
[19:54:11 CEST] <durandal_1707> it is probably slightly modified xvid idct
[19:54:53 CEST] <Yagiza> durandal_1707, maybe I have to ask my question at this channel?
[19:56:18 CEST] <durandal_1707> Yagiza: nope
[19:56:51 CEST] <Yagiza> durandal_1707, are you sure?
[19:58:33 CEST] <durandal_1707> Yagiza: you can open bug report, this is channel for developement
[19:59:29 CEST] <Yagiza> durandal_1707, before opening a bug report I want to make sure that I'm doing everything correct.
[20:00:19 CEST] <Yagiza> durandal_1707, so I need to ask those, who developed the feature.
[20:00:27 CEST] <durandal_1707> Yagiza: i'm not rtp expert
[20:01:15 CEST] <Yagiza> durandal_1707, are there some of them around?
[20:01:37 CEST] <durandal_1707> dunno
[20:02:43 CEST] <Yagiza> durandal_1707, ok. At which time of day most of FFMpeg developers available on the channels?
[20:03:01 CEST] <durandal_1707> Yagiza: some are not using irc at all
[20:03:23 CEST] <Yagiza> durandal_1707, do you know a way to contact them?
[20:04:28 CEST] <durandal_1707> Yagiza: seek with git over source code who is responsible for feature you want to use
[20:05:47 CEST] <Yagiza> durandal_1707, it seems it's Martin Storsjö
[20:06:20 CEST] <Yagiza> durandal_1707, https://patchwork.libav.org/patch/32593/
[20:08:09 CEST] <durandal_1707> Yagiza: he is not on this channel
[20:09:13 CEST] <atomnuker> he is I think, wbs 
[20:09:27 CEST] <Yagiza> atomnuker, wbs?
[20:42:09 CEST] <cone-091> ffmpeg 03Carl Eugen Hoyos 07master:81f25a105136: lavc/libx264: Support monochrome encoding.
[20:43:58 CEST] <cone-091> ffmpeg 03hwren 07master:f563e7868fb9: avs2: correct avs2 description
[21:34:56 CEST] <jamrial> atomnuker: if you didn't think that set solved the main issue, why didn't you say as much in the relevant thread?
[21:42:39 CEST] <atomnuker> lately I've had very little time to spend on reading the ML
[23:45:13 CEST] <cone-580> ffmpeg 03James Almer 07master:b955a33314d4: ffmpeg: simplify refcounting packets for the muxing queue
[00:00:00 CEST] --- Wed Aug  8 2018