[Ffmpeg-devel-irc] ffmpeg-devel.log.20180914

burek burek021 at gmail.com
Sat Sep 15 03:05:03 EEST 2018


[00:00:00 CEST] --- Fri Sep 14 2018
[00:08:54 CEST] <kierank> kurosu: the use-case is for smart TVs and apps where LGPL for reasons I may not agree with isn't doable
[00:09:10 CEST] <kierank> I'd rather see a BSD decoder that makes people use av1
[00:09:18 CEST] <kierank> Than an LGPL decoder in FFmpeg they can't use
[00:09:55 CEST] <JEEB> and the hwaccel stuff shouldn't be insurmountable
[00:28:09 CEST] <cone-506> ffmpeg 03Rick Kern 07master:a259501e5a6e: lavd/avfoundation: Fix skewed video output
[03:06:32 CEST] <cone-506> ffmpeg 03Shiyou Yin 07master:a55adf24b91a: avcodec/mips: [loongson] fix bug of svq3-watermark failed in fate test.
[03:06:33 CEST] <cone-506> ffmpeg 03Tristan Matthews 07master:f3b430e0e496: lavfi/silencedetect: fix spelling
[04:18:06 CEST] <cone-506> ffmpeg 03Martin Storsjö 07master:2a9e1c122eed: libfdk-aac: Don't use defined() in a #define
[04:18:07 CEST] <cone-506> ffmpeg 03James Almer 07master:1b98bfb932ad: Merge commit '2a9e1c122eed66be1b26b747342b848300b226c7'
[13:37:43 CEST] <durandal_1707> what idiot came to idea to remove some nal units from h264 stream, hardcode them and name new codec IMM5 which then instert them back
[13:38:33 CEST] <J_Darnley> The Chinese?  They ripped off h264 once before
[13:38:56 CEST] <J_Darnley> And Did I read the history right? Arch Linux has libaom without assembly?
[13:39:22 CEST] <nevcairiel> so someone said
[14:19:01 CEST] <January> where can I output a fate test's data? or am I expected to just hash the data to stdout 
[14:19:51 CEST] <thardin> I think some tests write temporary files just where the scripts are
[14:20:09 CEST] <thardin> if they really have to
[15:11:48 CEST] <funman> av500: you come to VDD next week?
[15:12:08 CEST] <funman> January: you missed wbs 
[15:12:45 CEST] <funman> he's an IBC noob too :)
[17:52:02 CEST] <kurosu> kierank, reminds me of Google licensing Ittiam's decoder for android
[17:52:39 CEST] <kurosu> though I don't think that's this side of the licensing that would condition manufacturer's usage
[17:53:21 CEST] <kurosu> I mean, only as a second step
[18:26:39 CEST] <jamrial> huh, cone joined but failed to post the commit notifications
[18:39:23 CEST] <BradleyS> cone-736, you're drunk
[18:53:26 CEST] <BtbN> It's not registered, and lost its hostname, so it's muted, due to the spam wave.
[18:54:33 CEST] <BtbN> albiero.videolan.org got an excemtpion from the ban, but cone just has its IP
[19:20:35 CEST] <atomnuker> nevcairiel: yep, that's right
[20:31:26 CEST] <durandal_1707> michaelni: does h264 have mode where inter frame stuff is always made on keyframe and not on prev frame?
[20:34:51 CEST] <michaelni> you can use a long term ref that points to the keyframe. Its not a seperate "inter mode" though
[20:41:01 CEST] <durandal_1707> michaelni: i ask, because parsing of imm5 files seems fully working, yet I see artifacts in decoded output for non-keyframes. One of their dlls uses internal ffmpeg and another different h264 decoder.
[20:42:56 CEST] <michaelni> unmodified ffmpeg ? or they modified the h264 decoder ?
[20:45:09 CEST] <durandal_1707> michaelni: stripped down ffmpeg with h264, i know that becuse in old days H264 codec value was 28, and it uses avcodec_open(28), not with that exact name but you get it
[20:48:48 CEST] <durandal_1707> and there are even strings with "ffmpeg" in it
[20:56:22 CEST] <michaelni> if its ffmpeg, do we have the source code for that ? if not did we ask for the source code ?
[20:58:20 CEST] <atomnuker> yep, if lavc is not in a separate dll they have to put the entire dll source code up somewhere
[20:58:46 CEST] <durandal_1707> michaelni: no, good luck finding corp that have source
[20:59:03 CEST] <JEEB> we can ask for it since it's either LGPL or GPL
[20:59:10 CEST] <JEEB> if they will reply is a separate problem
[21:08:33 CEST] <durandal_1707> [idvr] [icodec_decode_ffmpeg] undefined size!!
[21:12:02 CEST] <durandal_1707> i cant find any info about those who created vcmimm5.dll in 2012
[21:12:18 CEST] <JEEB> what's the format called?
[21:12:36 CEST] <JEEB> and the DLL probably has the usual DLL metadata fields?
[21:19:20 CEST] <durandal_1707> JEEB: just korean resource
[21:20:43 CEST] <michaelni> where do you have the binary from (i assume you have a binary), that person who distributes a binary must give you the source too if you ask
[21:20:50 CEST] <JEEB> seems like it's infinity cctv :P
[21:32:20 CEST] <durandal_1707> Your location (HR) has been blacklisted.
[21:38:29 CEST] <durandal_1707> michaelni: preson who distributes vcmimm5.dll is samples.ffmpeg.org
[21:57:48 CEST] <michaelni> durandal_1707, file has been removed from samples.ffmpeg.org
[21:58:03 CEST] <durandal_1707> michaelni: which file?
[21:58:32 CEST] <michaelni> vcmimm5.dll and vcmimm6.dll
[21:58:41 CEST] <durandal_1707> why?
[21:59:37 CEST] <michaelni> because you reported a copyright violation
[21:59:42 CEST] <durandal_1707> lol
[22:00:37 CEST] <durandal_1707> so you removed file because it links to some very old internal ffmpeg?
[22:02:06 CEST] <durandal_1707> michaelni: i nowhere mentioned vcmimm6.dll
[22:02:18 CEST] <durandal_1707> i even never looked at it
[22:04:26 CEST] <durandal_1707> thank God i have backup
[23:11:47 CEST] <durandal_1707> libavformat/sdp.c:394:40: warning: implicit conversion from 'int' to 'char' changes value from 254 to -2 [-Wconstant-conversion]
[23:13:44 CEST] <atomnuker> char == signed char?
[23:13:52 CEST] <atomnuker> in that case 254 - 128 - 128 = -2
[23:14:44 CEST] <wbs> no reason for the config pointer to be plain char there, since av_base64_encode takes a const uint8_t*
[00:00:00 CEST] --- Sat Sep 15 2018


More information about the Ffmpeg-devel-irc mailing list