[Ffmpeg-devel-irc] ffmpeg-devel.log.20150222

burek burek021 at gmail.com
Mon Feb 23 02:05:03 CET 2015

[00:00] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:52a1a55f3014: Add coverity model file
[00:07] <jamrial> ubitux: it's from http://research.microsoft.com/en-us/um/people/kopf/pixelart/supplementary/multi_comparison.html the filter they call "ours"
[00:08] <ubitux> oh i thought it was xbr
[00:08] <ubitux> that's the MS one, ok
[00:11] <ubitux> http://b.pkh.me/yoshi-16br.png this is xbr x16 (2 successive 4br, assuming that was the process they used)
[00:12] <ubitux> and my eyes hurt now
[00:13] <ubitux> http://b.pkh.me/yoshi-hq16.png this is what i get with 2 sucessive hq4x
[00:14] <ubitux> it's somehow different from their results
[00:14] <ubitux> http://research.microsoft.com/en-us/um/people/kopf/pixelart/supplementary/results_hq4x/smw2_yoshi_01_hq4x_16x.png
[00:14] <ubitux> our hqx=4,hqx=4 looks better :p
[00:14] <cone-961> ffmpeg.git 03Peter Meerwald 07master:12655c48049f: libavresample: NEON optimized FIR audio resampling
[00:14] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:db3f6465a631: Merge commit '12655c48049f9a52e5504bde90fe738862b0ff08'
[00:21] <cone-961> ffmpeg.git 03Martin Storsjö 07master:ff394ca087d4: rtpdec_h264: Add a missing closing paren in a log message
[00:21] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:b0b7571a5208: Merge commit 'ff394ca087d41941d2157e7a4e356e3ad312494e'
[00:28] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:1ec28a66f2c8: rtpdec: fix issue with conversion from unsigned to signed
[00:28] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:1fa2d95f6beb: Merge commit '1ec28a66f2c853a7e4aeb21495c0b4d9478f17db'
[00:43] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:4438d1c6ede3: rtsp: parse lang attribute in SDP
[00:43] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:285eea00013f: Merge commit '4438d1c6ede3190d7920513272fb9dc5371648f2'
[00:55] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:5e5583efe416: libavformat: add AC3 RTP depacketization (RFC 4184)
[00:55] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:17384b171980: Merge commit '5e5583efe41662294c5e1b76499ba338557f808d'
[01:03] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:fee2cc39becb: avformat/rtpdec_mpeg4: reassemble fragmented AAC frames
[01:17] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:5cbae5651d7c: mpegaudio_parser: add MP3 ADU headers parser
[01:17] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:dedbefee6043: Merge commit '5cbae5651d7c1ce9b0691dfbf2d474cb2b0ebb9a'
[01:47] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:96084251e57d: libavformat: add robust MPEG audio depacketization (RFC 5219)
[01:48] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:1186e7cf95dd: Merge commit '96084251e57d1738fde02a2b0d37ca609d9efd71'
[01:57] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:dcf969e98d3f: rtpdec_mpeg4: add more verbose error logging
[01:57] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:abf357a9fb2f: Merge commit 'dcf969e98d3f596006a0ddaf077870408b9c09c3'
[02:09] <cone-961> ffmpeg.git 03Gilles Chanteperdrix 07master:44dc138ed29c: rtpdec_mpeg4: reassemble fragmented AAC frames
[02:10] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:20f5a179560c: Merge commit '44dc138ed29c4dfbaf57ebc774da4c75ef113ca4'
[02:37] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:27216bf314c6: avcodec/libtheoraenc: Check for stats allocation failure
[02:37] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:67d4d5f5db06: avcodec/libtheoraenc: Check for av_fast_realloc() failure
[03:04] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:21518f5a0a6b: avcodec/libvpxenc: Check for av_malloc() failure
[03:04] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:066dc0437368: avcodec/libx264: Check for av_malloc() failure
[03:17] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:45e0aa46ca9b: postproc/postprocess: use av_mallocz()
[03:17] <cone-961> ffmpeg.git 03Michael Niedermayer 07master:ac424b23e404: postproc/postprocess: Check for av_mallocz() failure
[03:25] <thardin> äæää
[03:42] <thardin> å
[03:42] <thardin> ä
[03:49] <thardin> oops
[04:02] <thardin> hrj jsg ör ibte tomad
[13:56] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:f9d24ee1dc8e: postproc/postprocess: Check for av_malloc() failure
[13:56] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:48214956b2f1: avcodec/jpeglsenc: Check for memory allocation failures
[14:08] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:642c49c983df: avcodec/libaacplus: Check for av_malloc() failure
[14:35] <cone-325> ffmpeg.git 03Gilles Chanteperdrix 07master:22bb5bd7a3b5: avformat/rtsp: punch holes again after pause
[16:42] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:c3b6454d8016: avutil/fixed_dsp: Check for av_malloc() failure
[17:50] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:ca86e91dd660: tools/trasher: Check fseek returns
[17:50] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:568381030203: avformat/rmenc: Check for av_malloc() failure
[18:09] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:9b8152bf047b: avformat/oggdec: Check for av_malloc() failure and forward the error code
[18:19] <kurosu> jamrial: http://ffmpeg.org/pipermail/ffmpeg-devel/2013-May/143520.html
[18:19] <kurosu> when I did that, the benchmark were not particularly enticing
[18:22] <jamrial> save for order 4 it was faster when i benched. i didn't consider the memset overhead but using AVZERO should deal with that
[18:23] <kurosu> yeah, I was surprised when I saw the results, but didn't want to bother
[18:24] <kurosu> *saw my results (at the time)
[18:25] <wm4> nevermind that you can't seek correctly in tak (AFAIK), which completely ruins user experience
[18:25] <wm4> or maybe it was some other format
[18:31] <cone-325> ffmpeg.git 03Paul B Mahol 07master:586ba24ff294: avcodec/exr: fix memset first arg in reverse_lut()
[19:43] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:40adcf576f7d: avformat/oggdec: Check for ost allocation failure
[19:47] <jamrial> kurosu: that patch of yours could be applied. no need to bother adding unaligned AV_ZERO macros for this
[19:48] <jamrial> what did you use to bench back then? the fate sample?
[19:48] <jamrial> because that's probably too short to get proper numbers
[19:49] <jamrial> i encoded a 22 minutes long song using the encoder the author distributes
[19:56] <kurosu> jamrial, I don't quite remember
[19:57] <kurosu> up to you which one should be applied (eg not sure mine still applies)
[19:58] <kurosu> hum, actually I'm sure it wasn't the fate sample, but not sure it was the ones on samples.mphq
[19:59] <kurosu> on a related subject, do you have any idea how feasible is porting to yasm ?
[20:00] <jamrial> porting what?
[20:00] <kurosu> the remaining mlp dsp function
[20:01] <kurosu> you know, that duff device, where jump labels are stored into regs
[20:01] <jamrial> ah, filter_channel
[20:01] <kurosu> yeah
[20:02] <kurosu> I had looked at it like 3 years ago wondering if sometimes shortcut could be taken (like int32_t*int32_t->int32_t)
[20:02] <kurosu> also not sure if doing 2 samples (2 channels) per xmm would be possible
[20:03] <jamrial> something like we're doing with flac?
[20:03] <kurosu> haven't looked at it
[20:03] <kurosu> it probably requires changing a lot of buffers anyway
[20:04] <kurosu> but well, that looks so useless, seeing how mlp is not very used afaik
[20:04] <kurosu> wasn't it hddvd only for instance?
[20:04] <jamrial> no, it's used on blu-ray as well
[20:04] <kurosu> ok
[20:05] <jamrial> it's used on dvd-audio, probably also blu-ray audio, and for lossless tracks on blu ray movies (with ac3 as fallback)
[20:10] <kurosu> I thought dtshd-ma was more frequent
[20:10] <nevcairiel> it is, its still present however
[20:15] <kurosu> https://github.com/kurosu/ffmpeg/commits/mlpdsp <- useless stuff I had worked on
[20:15] <kurosu> refreshed that today, I think now it slows things down...
[20:16] <kurosu> the intent was to cleanup a bit buffers as a start point
[20:33] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:3518925a9127: avformat/movenc: Check for memory allocation failures
[20:33] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:b7e7ee6231bc: avfilter/buffer: Check for qp_table allocation failure
[20:49] <cone-325> ffmpeg.git 03Christophe Gisquet 07master:e0c7ba40cf0b: takdec: pad filter coeff buffer for DSP functions
[22:13] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:8559a714155a: avcodec/wmaenc: Check extradata allocations
[22:13] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:2234a1cd1668: avcodec/vorbisdec: Fix memleak, call cleanup on memory allocation failure
[22:38] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:a6baef6517ec: avcodec/mpegvideo: Fix memleak of mvs
[22:38] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:704c980294aa: avcodec/aacenc: Fix sample rate check
[23:33] <cone-325> ffmpeg.git 03Michael Niedermayer 07master:0a32a1b4bbcd: tools/graph2dot: Check for av_malloc() failure
[00:00] --- Mon Feb 23 2015

More information about the Ffmpeg-devel-irc mailing list