[Ffmpeg-devel-irc] ffmpeg-devel.log.20191128

[burek]

[00:00:41 CET] <cone-545> ffmpeg 03Paul B Mahol 07master:61dc7add3020: avcodec: add mvdv video decoder
[00:00:41 CET] <cone-545> ffmpeg 03Paul B Mahol 07master:ed58f8475fa8: avcodec: add mvha video decoder
[01:47:38 CET] <montana> i am looking for ffmpeg binary with rav1e support
[11:23:19 CET] <j-b> Hello
[11:24:07 CET] <durandal_1707> bye
[11:25:24 CET] <stevenliu> ouhayo, minasang
[12:15:10 CET] <durandal_1707> https://pastebin.com/aZ8Y2KU9  is this basically doing word swap within dword?
[12:20:09 CET] <jdarnley> I assume you want to avoid pshufb
[12:22:02 CET] <durandal_1707> jdarnley: this is code i'm REing, so i give shit how it is efficient, i just want to know what it does
[12:22:13 CET] <jdarnley> okay then
[12:23:55 CET] <jdarnley> 0x1b is q0123
[12:24:58 CET] <kurosu> in any case, just write some kind of vector containing eg 1 2 - 15, feed it into this code, and watch the result
[12:25:13 CET] <kurosu> probably faster than wrapping your head around it, albeit at the cost of your pride
[12:25:41 CET] <durandal_1707> kurosu: i cant write asm
[12:25:50 CET] <kurosu> you do :p
[12:26:10 CET] <kurosu> but I'm about to leave so can't help
[12:26:40 CET] <jdarnley> So it reverses the words in L and H
[12:26:48 CET] <jdarnley> those words are unpacked bytes
[12:27:01 CET] <jdarnley> does the same operation on all data
[12:27:04 CET] <jdarnley> packs again
[12:27:16 CET] <jdarnley> so yes, a packed bswap
[12:27:42 CET] <jdarnley> not a word swap
[12:28:03 CET] <jdarnley> but just to be sure I will write the test
[12:31:13 CET] <durandal_1707> radare2 is far from complete, dca gives TODO, i cant track functions this way
[13:01:27 CET] <jdarnley> $ ./a.out 
[13:01:27 CET] <jdarnley>  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16
[13:01:27 CET] <jdarnley>  4  3  2  1  8  7  6  5 12 11 10  9 16 15 14 13
[13:01:32 CET] <jdarnley> durandal_1707 ^
[13:21:25 CET] <durandal_1707> switching endianess
[14:24:32 CET] <cone-620> ffmpeg 03Thierry Foucu 07master:eaf566ce0b9f: Add options for spatial layers.
[14:39:21 CET] <durandal_1707> https://0x0.st/zI01.c   do we have this IDCT?
[14:48:33 CET] <durandal_1707> this cyclomatic complexity of ghidra should value SIMD instructions higher if it values it at all, here i get value of 10 for idct8
[14:57:58 CET] <durandal_1707> REing C++ classes is not fun, existing tools cant recover everything
[15:02:18 CET] <Lynne> durandal_1707: is that a float idct?
[15:02:31 CET] <Lynne> oh they clamp to ints
[15:04:30 CET] <Lynne> just looks like a standard 8-point, the mpeg2 one should work I think, or the jpeg one
[15:06:14 CET] <durandal_1707> Lynne: they have few variants to decode to different lowres
[15:06:55 CET] <durandal_1707> Lynne: this is from opencl obfuscated strings, same is used by CPU library
[15:07:23 CET] <durandal_1707> still can not find code that unpacks bitstream to dct coefficients
[15:07:36 CET] <Lynne> still looks like a standard dct which internally uses split-radix (hence the 4,2,2 for loops)
[15:07:51 CET] <Lynne> *split-radix fft
[16:38:25 CET] <cone-620> ffmpeg 03Zhong Li 07master:846e26b8c93a: lavc/rav1e: log and doc updated for const quantizer mode
[16:38:26 CET] <cone-620> ffmpeg 03Zhong Li 07master:1b831bc72975: doc/encoder: add the missing qsv encoders
[16:38:27 CET] <cone-620> ffmpeg 03Linjie Fu 07master:df625057af9b: lavc/qsvenc: Fix some code indentations
[18:50:55 CET] <cone-620> ffmpeg 03Limin Wang 07master:3a6ec10d904f: avfilter/vf_colorconstancy: av_frame_free(&in) in case of error or direct flag is false
[18:50:56 CET] <cone-620> ffmpeg 03Zhao Zhili 07master:ed5cdf3d5a7e: avcodec/vp8: fix multiple ff_thread_finish_setup() calls
[18:50:57 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:df912bbc5f9c: avformat/flacenc: Add const to ff_flac_write_header() parameter
[18:50:58 CET] <cone-620> ffmpeg 03Limin Wang 07master:8aa143eaa89c: avfilter/vf_libvmaf: Check for av_frame_alloc failure
[19:23:27 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:163bb9ac0af4: avformat/aviobuf: Avoid allocating buffer when using dynamic buffer
[19:23:28 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:67ce9e0463e1: avformat/apetag: Avoid allocation for small tags
[19:23:29 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:a31f68fb449e: avformat/av1: Avoid allocation for small headers
[19:23:30 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:c36a3df67692: avformat/avc: Avoid allocation for small SPS/PPS arrays
[19:23:31 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:aad82270b60a: avformat/id3v2enc: Avoid allocation for small tags
[19:23:32 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:3a58ec7c7725: avformat/matroskaenc: Avoid allocation for small dynamic buffers
[19:23:33 CET] <cone-620> ffmpeg 03Andreas Rheinhardt 07master:4d97b2ad2fa6: avformat/movenc: Avoid allocation for small dynamic buffers
[19:23:34 CET] <cone-620> ffmpeg 03James Almer 07master:16fa513392ad: avformat/avc: write the missing bits in the AVC Decoder Configuration Box
[00:00:00 CET] --- Fri Nov 29 2019