[Ffmpeg-devel-irc] ffmpeg-devel.log.20191207

burek burek at teamnet.rs
Sun Dec 8 03:05:03 EET 2019


[01:36:09 CET] <cone-512> ffmpeg 03Andreas Rheinhardt 07release/4.2:40123639fe7e: avformat/av1: Fix leak of dynamic buffer in case of parsing failure
[01:36:09 CET] <cone-512> ffmpeg 03Andreas Rheinhardt 07release/4.2:2722fc2bcfae: avcodec/extract_extradata_bsf: Don't unref uninitialized buffers
[08:50:35 CET] <cone-515> ffmpeg 03Jun Zhao 07master:46d2a67f80e6: lavfi/avf_showspectrum: Fix the memory leak in error handle path
[08:50:35 CET] <cone-515> ffmpeg 03Jun Zhao 07master:70e292becf3f: lavf/rtmpproto: Don't unref uninitialized buffers
[14:11:41 CET] <thardin> jpeg has the ability to say "all remaining coefficients are zero", right?
[14:15:40 CET] <thardin> EOB.. but is that only per block? or also for progressive JPEGs?
[14:37:06 CET] <thardin> "One can combine these two features to generate an image with a very large number ofprogressive scans (say, 80,000), each of which uses the EOB run feature to represent millionsof zero­value pixels using less than 100 bytes"  seems I'm not the first one to note this
[16:43:09 CET] <cone-156> ffmpeg 03Andreas Rheinhardt 07master:af50f0a515d8: avformat/matroskadec: Fix use-after-free when demuxing ProRes
[16:43:09 CET] <cone-156> ffmpeg 03James Almer 07master:70efa31ba1d1: fate/matroska: add a demux test for ProRes using zlib compression
[17:39:00 CET] <cone-156> ffmpeg 03James Almer 07master:ade7f4c60c00: fate/matroska: fix dependencies for fate-matroska-prores-zlib test
[18:31:08 CET] <Mysoft> ok kinda advanced question here, i'm using ffmpeg to convert to webm (libvpx), but i send to ffmpeg raw images
[18:31:34 CET] <Mysoft> which input format would be optimal for ffmpeg to not need to convert, (as well reducing bandwidth when sending to ffmpeg)
[18:32:30 CET] <Mysoft> i assume it uses some form of yuv but i dont know which one, and i assume i may want to -vf "scale=in_range=pc:out_range=pc"
[18:32:53 CET] <Mysoft> to prevent it from re-ranging as well
[18:33:32 CET] <Mysoft> if all my assumptions are true, which format for yuv i should use for input for ffmpeg to not need to convert?
[18:35:36 CET] <BtbN> Whatever format(s) libvpx accepts.
[18:38:18 CET] <Mysoft> ok yeah found this
[18:38:18 CET] <Mysoft> https://trac.ffmpeg.org/ticket/3271
[18:38:25 CET] <Mysoft> so yuv420p good :)
[19:37:20 CET] <cone-156> ffmpeg 03hwrenx 07master:b7583230ae68: MAINTAINERS: add myself as libxavs2 maintainer
[19:37:21 CET] <cone-156> ffmpeg 03Michael Niedermayer 07master:e1d836d2375c: avcodec/atrac9dec: Check q_unit_cnt more completely before using it to access at9_tab_band_ext_group
[20:06:33 CET] <durandal_1707> how much is simple_idct different from standard one?
[20:19:49 CET] <phh> x
[21:11:58 CET] <durandal_1707> how does idct works with 32 size of block?
[21:20:56 CET] <Lynne> pretty sure we only have 8x8 dcts
[21:22:15 CET] <durandal_1707> but is there 4x8 idct in wild?
[21:29:18 CET] <Lynne> yeah, av1
[21:31:45 CET] <Lynne> a dct is spatially separable though so its just a 4-point dct followed by an 8-point dct, or the opposite
[00:00:00 CET] --- Sun Dec  8 2019


More information about the Ffmpeg-devel-irc mailing list