[Ffmpeg-devel-irc] ffmpeg-devel.log.20180903

[burek]

[00:06:15 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:7592e88bfe3d: avcodec/indeo4: Check dimensions in decode_pic_hdr()
[00:06:16 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:93a203662f6f: avcodec/ra144: Fix undefined integer overflow in add_wav()
[00:06:17 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:0c88a5d3eb8c: avcodec/wmv2dec: skip frames that have only skiped MBs
[00:06:18 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:c5e574a0d0f9: avcodec/ffv1dec_template: do not ignore the return code of decode_line()
[00:06:19 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:42f053494ca1: avcodec/zmbv: remove useless zero check on dimensions
[00:06:20 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:a73579c12d66: iavcodec/zmbv: Avoid reallocating cur/prev buffers if they are large enough
[00:06:21 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:b544850a524f: avcodec/h264_refs: Detect more random access points which are not marked
[00:06:22 CEST] <cone-381> ffmpeg 03Michael Niedermayer 07master:697984b9db4d: avcodec/h264_refs: Document last if() in ff_h264_execute_ref_pic_marking()
[00:17:52 CEST] <cone-381> ffmpeg 03Diego Biurrun 07master:e368b0cbfa3f: tls_mbedtls: Use non-deprecated header file
[00:17:53 CEST] <cone-381> ffmpeg 03Zhong Li 07master:35ed7f93dbc7: qsvenc: Add an option to disable MFE mode
[00:17:54 CEST] <cone-381> ffmpeg 03Sven Dueking 07master:ea8ae27a5e11: avformat/libsrt: add payload size option
[00:17:55 CEST] <cone-381> ffmpeg 03Sven Dueking 07master:a507af97eef4: avformat/libsrt: add latency options and deprecate tspbdelay
[00:17:56 CEST] <cone-381> ffmpeg 03Zhong Li 07master:f8060865f3e1: qsvenc: use the compression_level to replace private option
[00:17:57 CEST] <cone-381> ffmpeg 03Zhong Li 07master:4ce701b4e640: qsvenc: remove vcm option on Linux
[00:17:58 CEST] <cone-381> ffmpeg 03James Almer 07master:fb4a12cda403: Merge commit 'e368b0cbfa3fdc627c331e34556f8e39296d04d5'
[00:17:59 CEST] <cone-381> ffmpeg 03James Almer 07master:b4ca32414ea2: Merge commit '35ed7f93dbc72d733e454ae464b1324f38af62a0'
[00:18:00 CEST] <cone-381> ffmpeg 03James Almer 07master:b1b0e532583e: Merge commit 'ea8ae27a5e112d06fd5625f640e40849e6313f0c'
[00:18:01 CEST] <cone-381> ffmpeg 03James Almer 07master:de33b3e457a6: Merge commit 'a507af97eef468238d545ff954a39d7432832e54'
[00:18:02 CEST] <cone-381> ffmpeg 03James Almer 07master:aebc5b2284db: Merge commit 'f8060865f3e1a16c62e0d337ef0979b6ee4ba457'
[00:18:03 CEST] <cone-381> ffmpeg 03James Almer 07master:8e789d244cc9: Merge commit '4ce701b4e640d4723a4005d664f31f8342fac40e'
[00:38:45 CEST] <cone-381> ffmpeg 03John Cox 07master:52fd2afce843: configure: fix inline asm checks
[00:38:46 CEST] <cone-381> ffmpeg 03Martin Storsjö 07master:b93026777aad: libfdk-aac: Use enum names instead of literal numbers for the output format
[00:38:47 CEST] <cone-381> ffmpeg 03Sven Dueking 07master:f25117a42865: libsrt: Pass the correct pointer for the passphrase
[00:38:48 CEST] <cone-381> ffmpeg 03Dmitry Rogozhkin 07master:c1bcd321ea2c: avcodec/qsv: fix async support
[00:38:49 CEST] <cone-381> ffmpeg 03wm4 07master:c011beda2611: avconv: make sure packets put into the muxing FIFO are refcounted
[00:38:50 CEST] <cone-381> ffmpeg 03Martin Storsjö 07master:e1e3a1224234: libopenh264: Add support for decoding of b-frames
[00:38:51 CEST] <cone-381> ffmpeg 03James Almer 07master:3eea8edf6180: Merge commit '52fd2afce8436c59c05765f3a6e95f9adb6f9f2f'
[00:38:52 CEST] <cone-381> ffmpeg 03James Almer 07master:2d62e06ff6a9: Merge commit 'b93026777aada7742583d8c5ab079e9f4dfe9a5d'
[00:38:53 CEST] <cone-381> ffmpeg 03James Almer 07master:768b07e3bca7: Merge commit 'f25117a4286505b38c12466ef04459471de3c1b0'
[00:38:54 CEST] <cone-381> ffmpeg 03James Almer 07master:7167ac33a8f2: Merge commit 'c1bcd321ea2c2ae1765a1e64f03278712221d726'
[00:38:55 CEST] <cone-381> ffmpeg 03James Almer 07master:c0a647644f27: Merge commit 'c011beda2611acfeb6f67d4fdf30d1eceed9e62f'
[00:38:56 CEST] <cone-381> ffmpeg 03James Almer 07master:022fa7a24ea8: Merge commit 'e1e3a12242347dd11174b2fb9ddac8dc8df16224'
[04:34:26 CEST] <cone-155> ffmpeg 03James Almer 07master:11cec348291d: avformat/av1: zero initialize the seq_params struct
[06:05:58 CEST] <atomnuker> haasn: wow, intel's vulkan drivers are shit, and they used to be the best
[06:06:23 CEST] <atomnuker> I'm signalling and waiting on semaphores in the right locations and they freeze and crash the gpu
[06:07:51 CEST] <atomnuker> and I'm still waiting on each queue to finish just so I don't make mistakes, so there are no races here, just needless semaphore signalling
[09:48:18 CEST] <cone-035> ffmpeg 03Marton Balint 07master:23fe072e43c2: avformat/libsrt: fix merge of payload_size option
[09:55:51 CEST] <atomnuker> durandal_1707 / jkqxz:updated vulkan branch, removed shaderc dep (at the cost of a c++ glslang wrapper)
[09:56:37 CEST] <atomnuker> should be a lot easier to compile now, since glslang and spriv-tools are actually packaged by pretty much everyone
[10:21:02 CEST] <j-b> atomnuker: that is interesting.
[10:33:10 CEST] <durandal_1707> what? ffv2? or from numerous other from never ending list of unfinished projects?
[11:48:15 CEST] <BtbN> Is there a config file for clang-format or other tools to make code into the ffmpeg style?
[12:06:57 CEST] <wbs> BtbN: http://martin.st/temp/uncrustify-libav.cfg - this is diego's config for uncrustify, iirc. should mostly match what's used, although I think this config is a bit excessive with the amount of vertical alignment it tries to do
[12:10:13 CEST] <BtbN> looks good, thanks
[12:25:19 CEST] <haasn> atomnuker: nice
[12:25:23 CEST] <haasn> seems like all drivers are the same now
[12:25:47 CEST] <haasn> but show the call sequence I guess?
[12:27:57 CEST] <kierank> michaelni: does chunks and slice threading work together?
[12:42:53 CEST] <BtbN> Unless I'm missing something, the transpose filter nvidia submitted doesn't transpose, but rotate.
[12:43:20 CEST] <BtbN> Not at home right now, so no nvidia machine near to test though
[12:53:24 CEST] <durandal11707> got rasc decoder almost working
[13:00:15 CEST] <michaelni> kierank, frame and slice threading in mpeg4 might, iam not sure but pthread.c enables only one i think
[13:00:26 CEST] <kierank> h264
[13:02:15 CEST] <durandal11707> only one is enabled at time, and you only need one type anyway
[13:02:44 CEST] <nevcairiel> in chunk mode you do send slices as chunks right, not any other arbitrarily cut parts
[13:02:50 CEST] <nevcairiel> i think it should work
[13:03:02 CEST] <michaelni> kierank, h264 should be the same
[13:03:22 CEST] <kierank> I will ask January to add a fate test then
[13:03:31 CEST] <kierank> nevcairiel: send NALs yes
[15:15:47 CEST] <durandal11707> Binary |.Net DLL and/or Com DLL -->>> I'm not interested in Windows
[15:32:13 CEST] <durandal11707> anybody have comments for SCPR v3 support patch? It is very big one.
[15:39:11 CEST] <durandal11707> funny thing i discovered, you can disable auto images downloading for twitter: and then for some cases, their blur is adding green artifacts at bottom, i wonder who coded that...
[16:29:08 CEST] <January> 'invalid mixed IDR / non IDR frames cannot be decoded in slice multithreading mode', what am I supposed to be encoding for slice threaded decoding to work
[16:30:10 CEST] <kierank> January: they aren't mixed idr non idr
[16:30:19 CEST] <kierank> your input is wrong
[17:29:48 CEST] <kierank> January: ask michaelni questions about h264.c
[19:32:57 CEST] <atomnuker> NVIDIA Performance Primitives video <- wat?
[19:33:29 CEST] <atomnuker> as if cuda isn't already heavy enough, they now have a library for this
[19:33:37 CEST] <BtbN> They always did
[19:33:48 CEST] <BtbN> It's actually quite handy for that stuff
[19:33:54 CEST] <durandal11707> ban hardware accelerators
[19:34:24 CEST] <BtbN> But their transpose filter just rotates lol
[19:38:56 CEST] <jkqxz> The format conversion stuff feels like it shouldn't be in that filter.
[19:39:32 CEST] <BtbN> it's a bit odd, yeah. They just copied it from the scale filter I guess
[19:39:33 CEST] <jkqxz> And given that it only rotates by right angles, what is that interpolation algorithm argument actually doing?
[19:40:05 CEST] <BtbN> The rotate function they use supports it, as it can do arbitrary angles. And as this was copied from the scale filter, it was kept.
[19:40:15 CEST] <jkqxz> Rotate should just move pixels around for those angles.
[19:41:03 CEST] <jkqxz> atomnuker:  Was the avgblur failure an Intel issue somehow, then?
[19:41:18 CEST] <BtbN> But there really should be a format conversion filter, and then a scale and transpose filter. Cause right now, if you scale _and_ transpose NV12, you will end up with 4 entirely pointless conversions back and fourth
[19:44:05 CEST] <durandal11707> Compn: where are the samples?
[19:44:13 CEST] <atomnuker> jkqxz: yeah, its an intel issue, I opened a bug way back when I discovered it
[19:44:45 CEST] <durandal11707> huh? how averaging pixels can be so problematic?
[19:48:54 CEST] <jkqxz> So the nondeterminism on RADV was probably unrelated.
[19:51:50 CEST] <Compn> durandal11707 : how is your rasc code ?
[19:51:52 CEST] <Compn> codec
[19:51:54 CEST] <Compn> er sample
[19:57:29 CEST] <atomnuker> durandal11707: vulkan multiplanar formats tend to do that, as I believe I have the only written and working code that uses them
[20:41:51 CEST] <durandal11707> Compn: i have sample. why you ask?
[21:05:46 CEST] <Compn> durandal11707 :Did you put it on samples.ffmpeg.org ?
[21:08:47 CEST] <durandal11707> Compn: cant, dunno info to login
[21:10:06 CEST] <Compn> put file online , i can copy it over
[00:00:00 CEST] --- Tue Sep  4 2018