[Ffmpeg-devel-irc] ffmpeg-devel.log.20180224

[burek]

[00:20:28 CET] <cone-904> ffmpeg 03Rodger Combs 07master:0419623cdca9: lavfi/vf_transpose: fix regression with semiplanar formats
[00:23:43 CET] <gagandeep_> atomnuker: hello Rostislav, you are also mentoring cineform, right?
[00:24:22 CET] <atomnuker> I'm just backup mentor, for when kierank isn't there and you have questions
[00:25:00 CET] <gagandeep_> okay
[01:48:58 CET] <atomnuker> jkqxz: I think I've completed the hwcontext if you want to review or test - https://github.com/atomnuker/FFmpeg/tree/exp_vulkan
[01:49:32 CET] <atomnuker> takes advantage of pretty much everything available to upload/download as fast as possible and can kinda import drm surfaces
[01:50:35 CET] <atomnuker> I was thinking of writing a chromatic aberration filter first but I think I'll go with a good old unsharpen
[01:50:55 CET] <atomnuker> as a compute shader rather than a vertex/fragment pair
[01:55:25 CET] <cone-904> ffmpeg 03James Almer 07master:4f273b97b046: avcodec/Makefile: add missing dependencies for some cuvid decoders
[03:09:58 CET] <jamrial> ubitux: https://github.com/nagadomi/waifu2x
[04:32:20 CET] <cone-904> ffmpeg 03James Almer 07master:9482ec1b203e: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:44:02 CET] <cone-904> ffmpeg 03James Almer 07release/2.8:26755a8c6f6d: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:44:04 CET] <cone-904> ffmpeg 03James Almer 07release/3.0:e0750d2067dc: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:44:06 CET] <cone-904> ffmpeg 03James Almer 07release/3.2:0a340924ef17: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:44:09 CET] <cone-904> ffmpeg 03James Almer 07release/3.3:e2edfff9a8f1: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:44:11 CET] <cone-904> ffmpeg 03James Almer 07release/3.4:e4b95f710ba4: avformat/hvcc: zero initialize the nal buffers past the last written byte
[04:51:24 CET] <cone-904> ffmpeg 03Jun Zhao 07master:b141902872bb: doc/bitstream_filters: documents delete_filler option.
[04:51:25 CET] <cone-904> ffmpeg 03Michael Niedermayer 07master:0293663483ab: avcodec/smc: Check input packet size
[04:51:26 CET] <cone-904> ffmpeg 03Jun Zhao 07master:74afa545286b: ffmpeg_opt: fix max_error_rate help info display issue.
[05:00:57 CET] <philipl> BtbN: why does nvenc say it accepts 0rgb rather than argb as documented?
[07:38:06 CET] <cone-904> ffmpeg 03Vishwanath Dixit 07master:ffe7cc89d0ce: avformat/dashenc: opening a segment file when its first frame is ready
[07:38:07 CET] <cone-904> ffmpeg 03Vishwanath Dixit 07master:28924f4b488a: avformat/dashenc: chunk streaming support for low latency use cases
[10:47:44 CET] <durandal_1707> i contacted stefano long ago, still expecting reply
[16:43:31 CET] <jamrial> why the fuck are there avpriv_ functions in avio.h?
[16:46:03 CET] <jamrial> da8b228977 the commit message makes no sense. they are still in an installed header
[17:11:15 CET] <philipl> BtbN: And on the subject of pix fmts, I still have my fix for the yuv444p10 problem sitting on a branch. I went back through the old review comments and no one objected, but you didn't say explicitly that you were happy with it.
[17:35:55 CET] <thardin> time to push the codec2 patchset I think
[17:40:31 CET] <wm4> jamrial: haha
[17:40:44 CET] <wm4> jamrial: they should probably be moved to a private header
[17:41:03 CET] <wm4> maybe url.h
[17:41:08 CET] <jamrial> yeah, but they still require a deprecation period since they have been effectively public for three years now
[17:41:15 CET] <jamrial> avio_internal.h IMO
[17:41:17 CET] <wm4> I wouldn't let that count
[17:41:37 CET] <nevcairiel> not really, its symbols marked as private
[17:41:40 CET] <wm4> they're on the same level as fields marked internal in public structs
[17:42:40 CET] <jamrial> alright
[17:48:05 CET] <jamrial> lovely, one of the avio examples in the documentiation uses avpriv_io_delete()
[17:48:22 CET] <jamrial> using only public api otherwise
[17:48:55 CET] <jamrial> actually, it uses both
[17:48:58 CET] <jamrial> so what now?
[17:50:15 CET] <wm4> ...
[17:58:50 CET] <atomnuker> they're still priv
[18:07:17 CET] <nevcairiel> fix the example and call it a day
[18:29:32 CET] <cone-056> ffmpeg 03Tomas Härdin 07master:f958f431eced: Add libcodec2 en/decoder
[18:29:32 CET] <cone-056> ffmpeg 03Tomas Härdin 07master:5655048c73f7: Add muxer/demuxer for raw codec2 and .c2 files
[18:29:32 CET] <cone-056> ffmpeg 03Tomas Härdin 07master:5caae2797998: Don't complain about codec2's 700 bit/s modes in ffmpeg.c
[18:29:32 CET] <cone-056> ffmpeg 03Tomas Härdin 07master:5f36c546ec4e: Add myself as codec2 maintainer
[18:34:20 CET] <jamrial> nevcairiel, michaelni: so what do i do with that av_parser_change() call? remove it altogether, leave it there as a noop, enabling it just for mpeg4?
[18:35:52 CET] <jamrial> to remove inband headers during codec copy you can use the revome_extradata bsf or even extract_extradata with the remove avoption enabled, which may be a better idea than forcing it based on output format
[18:36:55 CET] <jamrial> although gpac for example seems to remove them if you do a simple mp4box remux
[20:11:43 CET] <atomnuker> jkqxz: denoise_vaapi seems broken
[20:11:54 CET] <atomnuker> all settings result in a black screen
[20:30:55 CET] <cone-056> ffmpeg 03Marton Balint 07master:dcb2ef2211fd: avformat/aviobuf: add ff_read_line_to_bprint and ff_read_line_to_bprint_overwrite functions
[20:30:56 CET] <cone-056> ffmpeg 03Marton Balint 07master:ffabff1baa08: avformat/concatdec: add support for very long line sizes
[21:26:08 CET] <hathinns> 4:54 PM <hathinns> 4:42 PM <hathinns> Hi, having problem while trying to read video frames from ffmpeg and piping it to python for further processing in opencv. Processing is going fine but when I pipe the output to an output using another ffmpeg subprocess, the video frames are scrolling vertically from bottom to top.
[21:26:59 CET] <hathinns> the python script https://usercontent.irccloud-cdn.com/file/kOYUeNTS/processing.py
[21:27:31 CET] <RiCON> hathinns: wrong channel -> #ffmpeg
[21:28:37 CET] <hathinns> And the log
[21:28:51 CET] <hathinns> https://usercontent.irccloud-cdn.com/file/qJajDvaN/processing_log.txt
[21:29:39 CET] <hathinns> Any suggestions on how to solve this vertical scrolling issue
[21:30:42 CET] <wm4> wrong channel
[21:43:39 CET] <cone-056> ffmpeg 03Martin Vignali 07master:ea03f295ec79: avcodec/exr : add support for long name flag and be more explicit about unsupported flag
[21:43:40 CET] <cone-056> ffmpeg 03Martin Vignali 07master:c8552b78584a: fate/exr : add test for long name flag
[21:46:16 CET] <cone-056> ffmpeg 03Martin Vignali 07master:e3fc36a84c43: checkasm/vf_blend : add depth param in order to add test for 16 bit version
[21:46:17 CET] <cone-056> ffmpeg 03Martin Vignali 07master:7590d58b6146: avfilter/x86/vf_blend : reorganize init in order to add 16 bit version
[21:46:18 CET] <cone-056> ffmpeg 03Martin Vignali 07master:6c6c9d14a841: avfilter/x86/vf_blend : indent
[21:46:19 CET] <cone-056> ffmpeg 03Martin Vignali 07master:53a03b5c8c7d: avfilter/x86/vf_blend : add 16 bit version for BLEND_SIMPLE, phoenix, difference for SSE and AVX2 (x86_64)
[21:46:20 CET] <cone-056> ffmpeg 03Martin Vignali 07master:c0919c49855b: checkasm/vf_blend : add test for blend_simple_16, phoenix_16 and difference_16
[21:49:11 CET] <beastd> hi
[21:53:36 CET] <beastd> I am preparing a simple diagram listing functionality of FFmpeg by listing the main features / components per lib.
[21:55:45 CET] <beastd> I want to use it on a poster at chemnitz linux days. I would be grateful if some other developers could have a look and tell me about typos, errors or more things I could mention.
[21:57:32 CET] <beastd> Here is a sketch with what I have so far: https://www.dropbox.com/s/i98ptge5xmobiu8/sketch-ffmpeg-libs.png?dl=0   (I will pretty it up later, just need to finalize what's in it before to avoid doing the layouting again and again.)
[22:11:30 CET] <Compn> ooo
[22:11:35 CET] <Compn> sounds fun
[22:12:39 CET] <Compn> beastd : for libavdevice, i think you might want to change the "input devices" and "output devices" to something more descriptive
[22:14:39 CET] <Compn> beastd : you might also  be able to arrange them with an example video so that the video gets processed as it moves through the diagram
[22:14:52 CET] <atomnuker> lavu does a _lot_ of things
[22:15:29 CET] <atomnuker> hardware context creation, uploading and downloading, vector operations using SIMD on ints and floats
[22:15:39 CET] <beastd> Compn: What do you have in mind in/out devices? In general I want to go more for the abstract. If people ask, I can always tell them a bunch of examples.
[22:16:36 CET] <beastd> atomnuker: true. i'm not sure if it is a good idea to list all those on a poster.
[22:18:22 CET] <Compn> beastd : i'm not sure, i am thinking...
[22:18:42 CET] <Compn> they are definitely audio/video devices
[22:18:47 CET] <Compn> instead of input/output devices :P
[22:21:12 CET] <beastd> atomnuker: i mainly want to achieve 2 things: 1) FFmpeg has code for lots of stuff and more important (2) if a developer looks at it he might find out the right lib to start looking at for whatever he wants to find out/wants to implement. So I think the interested person will start looking at e.g. libavcodec and then realize the stuff you mentioned is part of lavu
[22:21:43 CET] <Compn> i thought mike melanson made some kind of graph like this before :D
[22:21:48 CET] <Compn> or some other powerpoint presentation :D
[22:21:52 CET] <Compn> someone else made somewhere
[22:21:53 CET] <Compn> hmmm
[22:23:36 CET] <beastd> Compn: good point. i never mentioned audio / video in there explicitly. maybe i will add it all over or in the text part that will be on the right
[22:29:57 CET] <Compn> yeah some quick google searches i cant find the diagram i remember, so ignore that
[22:32:29 CET] <durandal_1707> beastd: libresample is not ffmpeg lib
[22:32:43 CET] <durandal_1707> you want libswresample
[22:33:18 CET] <atomnuker> fucking hell does aurel live in his little own world where everyone uses the crappiest codecs and has devices which don't support them and there are no specifications
[22:33:35 CET] <atomnuker> so many fucking exceptions for those piece of shit special snowflake codecs
[22:33:59 CET] <durandal_1707> atomnuker: whhats issue now?
[22:34:05 CET] <beastd> durandal_1707: right. it should read libswresample. i'll fix it. thank you!
[22:34:49 CET] <atomnuker> durandal_1707: he thinks _all_ coding tools are optional and not required to be implemented by decoders, so of course it should be up to users to set them for idividual devices
[23:53:01 CET] <philipl> atomnuker: bluetooth audio is its own special pile of snowflakes.
[00:00:00 CET] --- Sun Feb 25 2018