[Ffmpeg-devel-irc] ffmpeg-devel.log.20170205

[burek]

[00:00:52 CET] <cone-182> ffmpeg 03Marton Balint 07master:a91cedf79a92: avfilter/ebur128: do not allow null ebur128 context in ff_ebur128_relative_threshold
[00:00:54 CET] <cone-182> ffmpeg 03Marton Balint 07master:fd57f70c9788: avfilter/ebur128: fix relative threshold calculation for multiple contexts
[01:27:41 CET] <cone-182> ffmpeg 03Michael Niedermayer 07release/3.0:9797929749d0: avcodec/interplayvideo: Move parameter change check up
[01:27:42 CET] <cone-182> ffmpeg 03Michael Niedermayer 07release/3.0:f9a96bac2804: avcodec/pngdec: Check trns more completely
[01:27:43 CET] <cone-182> ffmpeg 03Michael Niedermayer 07release/3.0:c595b1da4f93: Changelog: Update
[02:08:54 CET] <cone-182> ffmpeg 03James Almer 07release/3.0:e8b94e5ce48f: configure: bump year
[02:17:14 CET] <cone-182> ffmpeg 03Marton Balint 07n3.0.6:HEAD: avfilter/ebur128: fix relative threshold calculation for multiple contexts
[04:02:44 CET] <Compn> durandal_1707 : i was up til 3 am looking for the clearvideo encoder
[04:02:47 CET] <Compn> i give up i think
[04:03:33 CET] <Compn> no one has a mirror of ftp.iterated.com. i cant even find the filename of the encoder evaluation
[04:04:09 CET] <Compn> did that file we have that says its the compressor work in win95 vfw 1.1 ?
[04:08:23 CET] <Compn> some of these codecs only work on the old os...
[04:08:54 CET] <Compn> i found a new file though
[04:09:07 CET] <Compn> at least i think its new
[04:10:05 CET] <Compn> clramd.ax ClearVideo ® Decompression Filter 
[04:10:53 CET] <Compn> cfusion.ocx , the clear fusion netscape plugin
[07:58:30 CET] <Compn> theres an interesting mail
[07:58:43 CET] <Compn> Dear the WebMaster,
[07:59:24 CET] <Compn> then a long request for a website counter to track commits
[11:43:53 CET] <durandal_1707> trac database is locked
[12:17:06 CET] <michaelni> durandal_1707, seems fine here
[14:51:32 CET] <Compn> wow ami found a vdo sample! awesome
[14:51:54 CET] <JEEB> wtf is that?
[14:52:04 CET] <JEEB> something from the early 2000s?
[14:52:13 CET] <JEEB> seemed like ASF
[14:52:46 CET] <Compn> VDO is super old, one of the first. like vivo , rm10 , clearvideo
[14:52:59 CET] <JEEB> ah
[15:09:11 CET] <Compn> oh , we had vdo samples
[15:09:18 CET] <Compn> http://samples.ffmpeg.org/V-codecs/VDOM/
[15:25:04 CET] <Compn> https://trac.ffmpeg.org/ticket/2222 seems to work now? carl can you confirm ?
[15:36:24 CET] <Compn> carl : the patches here were sent to ffmpeg-devel ? https://trac.ffmpeg.org/ticket/3233
[15:41:47 CET] <kierank> michaelni: I'm going to add myself and others who wish to the Google oss-fuzz list
[15:42:10 CET] <kierank> Because it's crazy that you are sole contact
[15:47:24 CET] <michaelni> ffmpeg-security at ffmpeg.org is the contact
[15:47:35 CET] <michaelni> or rather were things are sent to
[16:02:53 CET] <kierank> michaelni: github says otherwise
[16:04:01 CET] <kierank> Ah right, can you add me to ffmpeg-security
[16:04:27 CET] <BtbN> That cinepak dude is weird.
[16:04:42 CET] <BtbN> He has been told multiple times that his approach is not wanted, and still goes on about it.
[16:05:49 CET] <kierank> Sounds familiar
[16:06:39 CET] <BtbN> He basically wants to fix a broken proprietary ffmpeg using application by patching ffmpeg to make it read environment variables for him to supply settings that application doesn't support.
[16:06:59 CET] <BtbN> And also wants to put pixel format conversion into a decoder instead of swscale
[16:11:26 CET] <jkqxz> I don't think the different RGB pixel format (RGB32, RGB24, RGB565, PAL8) output in the decoder are insane, but quadruplicating large amounts of the code to support them isn't helpful.
[16:11:44 CET] <jkqxz> Embedding a fixed RGB-to-YUV conversion is insane, though.
[16:17:05 CET] <JEEB> of course we already do such conversion in some subtitle decoders, but it doesn't make it any better
[16:17:32 CET] <nevcairiel> subtitles are a bit special since we only support one specific output format
[16:17:37 CET] <nevcairiel> its not like we can output anything else
[16:17:43 CET] <nevcairiel> but its legacy and hopefully fixed some day
[16:17:59 CET] <wm4> sigh is that still going on
[16:18:20 CET] <JEEB> oh, so you can't just output 4:4:4 YCbCr from a subtitle decoder? interesting
[16:18:40 CET] <nevcairiel> i dont think so
[16:18:45 CET] <nevcairiel> has to be pal8 iirc
[16:22:27 CET] <wm4> well it's not really defined
[16:23:25 CET] <wm4> yeah, API users have to guess
[16:25:18 CET] <nevcairiel> which is why everything outputs the s ame format, i guess
[16:25:33 CET] <nevcairiel> of course it helps that all bitmap subtitle formats are indeed paletted
[16:25:40 CET] <nevcairiel> even if some use a yuv palette
[16:41:50 CET] <cone-712> ffmpeg 03Mark Thompson 07master:df3b17eba47e: ffmpeg_vaapi: Unreference global device before setting it
[16:41:51 CET] <cone-712> ffmpeg 03Mark Thompson 07master:3420b34a8a63: Revert "avutil/hwcontext_vaapi: fix SEGV in vaTerminate when vaInitialize fails"
[16:41:52 CET] <cone-712> ffmpeg 03Mark Thompson 07master:55a7e7e3986e: ffmpeg_vaapi: Always set hwaccel_ctx, so it gets properly cleaned up on error
[16:41:53 CET] <cone-712> ffmpeg 03Mark Thompson 07master:326b1ed93e25: vf_scale_vaapi: Add missing return value checks
[17:12:14 CET] <Compn> dang, how many more codecs are hiding in ffmpeg trac? :\
[17:50:38 CET] <Compn> https://trac.ffmpeg.org/ticket/5285 might be a good qualification task
[17:50:43 CET] <Compn> or starting point for new developer
[17:51:31 CET] <BtbN> Or for anyone wanting 250$ I guess
[17:51:34 CET] <JEEB> I think MS has a sample collection of its own of that
[17:51:45 CET] <JEEB> + the smooth streaming spec
[17:52:12 CET] <JEEB> ismt is pretty much yet another XML in fragmented ISOBMFF with MS custom boxes thing
[17:52:25 CET] <JEEB> so writing is simpler than reading :D
[17:52:50 CET] <Compn> dont we just parse the xml as little as possible anyway
[17:53:00 CET] <JEEB> we don't, at all
[17:53:16 CET] <Compn> i mean, convert xml tags to srt tags internally... :P
[17:53:19 CET] <JEEB> I don't think we have a single XML-based format demuxer/decoder
[17:53:31 CET] Action: Compn looks at mp4 nonsense
[17:53:51 CET] <JEEB> that one adobe thing got cuckblocked with bikeshedding about which XML library to use
[17:54:02 CET] <Compn> oh thats what happened
[17:54:11 CET] <Compn> also cuckblocked, nice
[17:54:52 CET] <JEEB> basically to write semi-sane limited amount of XML you can just sprintf. but to read...
[17:54:56 CET] <JEEB> not so simple
[17:55:27 CET] <Compn> ah
[17:55:39 CET] <Compn> just one big ass lookup table? 
[17:55:43 CET] <Compn> we add as we go? no ?
[17:55:56 CET] <Compn> its probably all nested :\
[17:56:18 CET] <JEEB> let's not go there, if you want XML you just use one of the maintained libraries and be content with that
[17:56:20 CET] <Compn> <xml version "1.0 wecanthavenicethings=1>
[18:05:51 CET] <Compn> this would also be a good project for new developer, seeing as how it is based on open source format, https://trac.ffmpeg.org/ticket/1959
[18:06:05 CET] <Compn> someone remind me to create new developer page or email
[18:06:14 CET] Action: Compn afk
[20:09:15 CET] <ashk43712> Hi, I am new to FFmpeg. I was working on integrating Netflix's "VMAF - Video Multi-Method Assessment Fusion" to the libavfilter so that by using the command 'ffmpeg -i encoded_file -i source_file -lavfi vmaf -f null -' one can get the results similar to Netflix's vmaf's output. But I am not sure about where (which files) to make changes so that I can call vmaf from the command line.
[20:18:54 CET] <durandal_1707> ashk43712: is this for gsoc?
[20:19:25 CET] <ashk43712> yeah
[20:20:45 CET] <durandal_1707> ashk43712: see related commits in libavfilter directory which adds new filter
[20:21:04 CET] <durandal_1707> there is bunch of examples
[20:21:33 CET] <durandal_1707> you need to edit libavfilter/Makefile
[20:21:56 CET] <durandal_1707> and libavfilter/allfilters.c
[20:22:48 CET] <durandal_1707> and also have vf_vmaf.c
[20:23:21 CET] <durandal_1707> in libavfilter directory
[20:23:58 CET] <ashk43712> okay, thanks.
[20:37:47 CET] <ubitux> durandal_1707: doc/writing_filter.txt is less redirection effort :p
[20:39:08 CET] <durandal_1707> is it updated to not call old api no longer available?
[20:46:14 CET] <durandal_1707> ashk43712: see vf_psnr and vf_ssim.c in libavfilter directory
[20:46:41 CET] <durandal_1707> you will need to use fframesync internal api
[20:49:22 CET] <ashk43712> yeah, looking at it
[20:53:25 CET] <ubitux> durandal_1707: it's not much mentioning the old api
[20:53:36 CET] <ubitux> s/old//
[20:53:46 CET] <ubitux> what "old" api are you refering to?
[20:54:15 CET] <durandal_1707> ubitux: pre avframe era
[20:54:24 CET] <ubitux> it never mentioned that
[20:54:28 CET] <ubitux> it's a doc i wrote "recently"
[20:54:39 CET] <ubitux> (2014)
[20:54:51 CET] <durandal_1707> ahh there is other file than?
[20:56:48 CET] <durandal_1707> ashk43712: see also doc/writing_filters.txt
[22:59:12 CET] <cone-363> ffmpeg 03Marton Balint 07master:e57fd926b091: avfilter/af_pan: fix null pointer dereference on empty token
[00:00:00 CET] --- Mon Feb  6 2017