[Ffmpeg-devel-irc] ffmpeg-devel.log.20150503

[burek]

[00:59:51 CEST] <cone-666> ffmpeg 03Martin Storsjö 07master:3080b0497ddf: avprobe: Support writing arrays of integers without keys
[00:59:52 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:db4649dd1094: Merge commit '3080b0497ddf8549d86ee99b79ac0c15f44ee382'
[00:59:53 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:2a3270b66c3c: ffprobe: add writer_print_integers() to allow printing formated 1d and 2d arrays of integers
[00:59:54 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:08152f2df2c8: ffprobe: Add Support to print information about AV_FRAME_DATA_DISPLAYMATRIX
[00:59:55 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:a72b61a317b0: ffprobe: support printing AVStream and AVPacket side data
[01:17:30 CEST] <cone-666> ffmpeg 03Martin Storsjö 07master:709e1c91ea7a: avprobe: Print the display matrix from per-stream sidedata
[01:17:31 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:0ba65c9f795a: Merge commit '709e1c91ea7a5bbf2b9a104642572ca7616b224f'
[01:28:50 CEST] <cone-666> ffmpeg 03Martin Storsjö 07master:863ee06809b2: configure: Use the right local variable in the MSVC and ICL probes
[01:28:51 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:02b5af774405: Merge commit '863ee06809b215895ee20cbc557eeceb904cf770'
[01:42:05 CEST] <cone-666> ffmpeg 03Matt Oliver 07master:3edeb7749e42: configure: Allow log2 with MSVC 2013 onwards.
[01:42:06 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:40964e2e894f: Merge commit '3edeb7749e4276c78ad57307b8c3b32dd476d1da'
[01:53:12 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:c8b4ffdf554b: configure: prepend avfilter_deps for *_rect filters
[01:53:13 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:eb1688119975: LICENSE.md: add *_rect filters
[02:06:09 CEST] <cone-666> ffmpeg 03Jérôme Martinez 07master:14605a0b9915: ffv1dec: plane_index is 1 in case of version 4 gray+alpha.
[03:21:54 CEST] <haasn> JEEB: what ever happened to those OpenHEVC intrinsics in ffmpeg?
[03:21:59 CEST] <haasn> Have they ever been merged in?
[03:22:17 CEST] <nevcairiel> we dont do intrinsics in ffmpeg
[03:23:10 CEST] <haasn> Ah, okay
[03:26:45 CEST] <jamrial> most were ported to nasm/yasm syntax and merged
[03:27:13 CEST] <jamrial> only idct and intrapred are missing, afaik
[04:42:46 CEST] <cone-666> ffmpeg 03Michael Niedermayer 07master:29ef54aa9086: avcodec/nvenc: Make pix_fmts_nvenc const
[06:17:56 CEST] <haasn> Can somebody figure out how the const int32_t ff_yuv2rgb_coeffs[8][4] = { in yuv2rgb.c works?
[06:18:12 CEST] <haasn> None of the magic constants are explained
[06:29:31 CEST] <haasn> { 117504, 138453, 13954, 34903 }, /* ITU-R Rec. 709 (1990) */
[06:30:39 CEST] <haasn> The only thing I can decipher is that third number probably has something to do with blue, since it's very small
[06:31:22 CEST] <haasn> But the numbers are also not proportional in the slightest to the luma coefficients BT.709 uses (0.2126, 0.7152, 0.0722)
[06:31:59 CEST] <haasn> (Except that the second is roughly 10 times the third)
[06:33:13 CEST] <haasn> Curiously, there's a second set of parameters I can play with:     { 117579, 136230, 16907, 35559 }  /* SMPTE 240M (1987) */                    
[06:33:32 CEST] <haasn> I know that these correspond to the luma coefficients 0.2122, 0.7013, 0.0865
[06:33:57 CEST] <haasn> And indeed 13954 / 0.0722 * 0.865 H 16718 H 16907
[06:34:35 CEST] <haasn> Although I'm not sure what explains the remaining difference; even if I try using only two digits of preicison it comes out as 16667 or 16861 depending on rounding
[06:35:28 CEST] <haasn> there is this same ~300 difference for the R channel as well (117504 / 0.2126 * 0.2122 H 117282)
[06:36:29 CEST] <haasn> and for the green channel. Maybe it's related to the fact that the fourth parameter is a few hundred higher than for BT.709?
[06:36:45 CEST] <haasn> Since they're both 3D spaces, presumably the fourth is some sort of constant offset?
[12:15:17 CEST] <cone-138> ffmpeg 03Muhammad Faiz 07master:a141948a8be7: avfilter/avf_showcqt: optimize gamma calculation
[14:22:40 CEST] <nevcairiel> haasn: if you know the calculations for yuv->rgb, the table contains crv, cbu, cgu, cgv, in that order, ie. the first step of the coeff calculation has already been hardcoded
[16:28:11 CEST] <BtbN> haasn, those are the fixed point constants for calculating the RGB components from yuv, for diffrent yuv color standards.
[16:28:43 CEST] <BtbN> They are just defined to be that way
[17:18:45 CEST] <cone-425> ffmpeg 03Michael Niedermayer 07master:52835cb8e16c: avutil/opt: Avoid division by 0
[17:18:45 CEST] <cone-425> ffmpeg 03Michael Niedermayer 07master:042260cde4ec: avcodec/wavpack: Check L/R values before use to avoid harmless integer overflow and undefined behavior in fate
[17:37:11 CEST] <cone-425> ffmpeg 03Andreas Cadhalpun 07master:287dbb0771d5: pngdec: check s->last_picture.f->data[0] before using it
[20:36:49 CEST] <cone-425> ffmpeg 03Andreas Cadhalpun 07master:8f760be4d312: pngdec: return correct error code from decode_frame_common
[21:02:20 CEST] <anshul_mahe> ubitux: I saw ur comment regarding style in mail,can u throw more light
[21:02:43 CEST] <anshul_mahe> which style, you did not like
[21:02:59 CEST] <ubitux> http://ffmpeg.org/developer.html#Coding-Rules-1
[21:03:30 CEST] <ubitux> if(ret < 0) {
[21:03:32 CEST] <ubitux> this is wrong
[21:03:37 CEST] <ubitux> if( !ctx->pktbuf) {
[21:03:40 CEST] <ubitux> this as well, differently
[21:03:51 CEST] <ubitux> if ( ret == 0 )
[21:03:55 CEST] <ubitux> this as well, differently again
[21:04:44 CEST] <ubitux> it's not important for that patch, but it would be nice to have it fixed at some point
[21:10:33 CEST] <anshul_mahe> ubitux: correct way is "if ( ret == 0 )" or if(ret == 0)
[21:10:54 CEST] <ubitux> look at the link i pasted
[21:10:57 CEST] <ubitux> look at other files too
[21:12:53 CEST] <anshul_mahe> ok from other file, I should use if (ret == 0). but nothing related to it in doc
[21:16:49 CEST] <ubitux> good point
[21:16:56 CEST] <ubitux> well, just keep it consistent with the other file
[21:17:04 CEST] <ubitux> or at the very least within the same file
[21:24:03 CEST] <anshul_mahe> ok, I will atleast get this done in webvtt patch
[21:24:21 CEST] <anshul_mahe> there are also similar style problem
[21:25:58 CEST] <anshul_mahe> but in condition like this if ((ret=0)),  its more readable when space is left(my opinion) like if ( (ret = 0) )
[21:34:22 CEST] <anshul_mahe> bye
[22:04:34 CEST] <cone-425> ffmpeg 03Andreas Cadhalpun 07master:372aa0777aaa: pngdec: don't use AV_PIX_FMT_MONOBLACK for apng
[22:59:42 CEST] <cone-425> ffmpeg 03Michael Niedermayer 07master:f9e802018e54: ffplay: unify displaymatrix based rotation code
[23:39:16 CEST] <cone-425> ffmpeg 03Anshul Maheshwari 07master:a6b630e9b6d8: avcodec/ccaption_dec: correcting line breaks in cc
[00:00:00 CEST] --- Mon May  4 2015