[Ffmpeg-devel-irc] ffmpeg-devel.log.20150619

burek burek021 at gmail.com
Sat Jun 20 02:05:04 CEST 2015


[00:12:02 CEST] <philipl> ubitux: thanks
[00:13:41 CEST] <cone-489> ffmpeg 03Andreas Cadhalpun 07master:d216b9debd98: vp9: don't retain NULL as segmentation_map
[01:49:00 CEST] <cone-489> ffmpeg 03Andreas Cadhalpun 07master:590743101dc9: postproc: fix unaligned access
[02:26:50 CEST] <cone-489> ffmpeg 03Michael Niedermayer 07master:76cc8582fd6f: avcodec/fft-test: fix type of cpuflags
[02:26:51 CEST] <cone-489> ffmpeg 03Michael Niedermayer 07master:d8a227f2319a: avutil/adler32: Fix data type in test code
[02:26:52 CEST] <cone-489> ffmpeg 03Michael Niedermayer 07master:fdbad24090e1: avutil/aes: Fix types
[02:26:53 CEST] <cone-489> ffmpeg 03Michael Niedermayer 07master:2c881c7d45bc: avutil/avstring: Fix undefined shift
[02:26:54 CEST] <cone-489> ffmpeg 03Michael Niedermayer 07master:d1dce1c51728: avutil/crc: Fix type of p table so its content fits without overflwoing
[05:33:40 CEST] <cone-661> ffmpeg 03Michael Niedermayer 07master:bb3703a4c306: avcodec/j2kenc: YUV should not use the ICT
[08:31:15 CEST] <rcombs> durandal_1707: so, it turns out BFSTM files always have little-endian ADPC tables, even when the whole rest of the file is big-endian
[08:32:48 CEST] <durandal_1707> good
[09:50:16 CEST] <rcombs> durandal_1707: do you have a sample without an ADPC chunk?
[09:51:06 CEST] <durandal_1707> rcombs: unfortunately no
[09:51:23 CEST] <rcombs> because I've just made a lavc change that would make those really easy to support
[09:51:44 CEST] <durandal_1707> nice
[09:53:10 CEST] <rcombs> https://gist.github.com/2ee9b4bb33f5568031db <-- look sane?
[10:07:01 CEST] <rcombs> now I'm curious who actually cared about x265 on POWER
[10:08:16 CEST] <durandal_1707> rcombs: looks good to me
[10:09:09 CEST] <rcombs> cool, I'll send along with a few revisions to my existing patches on the topic
[10:11:23 CEST] <rcombs> oh, that reminds me, did anyone else have an opinion on wm4's suggestion of using AVStream extradata for loop points, instead of adding members?
[10:11:25 CEST] <rcombs> seemed sane enough to me
[10:12:02 CEST] <wm4> feels better not to litter the context with even more fields
[10:12:12 CEST] <wm4> side data is slightly more organized
[10:12:28 CEST] <rcombs> or, yeah, side data
[10:12:33 CEST] <durandal_1707> well I'm not fan of loop points so I will leave it to others
[10:14:43 CEST] <rcombs> wm4: also, do you think it would make more sense to require the demuxer (or, in the muxing case, user) to specify both points, then?
[10:15:01 CEST] <rcombs> (I'm imagining a struct with 2 int64_t's)
[10:15:04 CEST] <wm4> not sure what you mean
[10:15:16 CEST] <wm4> but yes, a single side data for both values
[10:15:45 CEST] <rcombs> right now, AV_NOPTS_VALUE can be specified for the end loop point to mean "the duration of the file"
[10:16:04 CEST] <cone-563> ffmpeg 03George Boyle 07release/2.7:a9b600cf3995: avcodec/flacenc: Fix Invalid Rice order
[10:16:07 CEST] <wm4> this could remain as-is
[10:16:31 CEST] <rcombs> alright
[10:17:57 CEST] <rcombs> though, waitaminute
[10:18:35 CEST] <rcombs> right now I have it on AVFormatContext because every format that supports this that I'm aware of specifies it at the container level, rather than the stream level
[10:18:51 CEST] <rcombs> and it doesn't seem like it would make sense to have separate loop points for each stream
[10:19:29 CEST] <rcombs> so how would that work at the AVStream level if the file had multiple streams
[10:19:29 CEST] <wm4> these files are single-stream anyway, aren't they?
[10:20:42 CEST] <rcombs> the formats with native support that I'm aware of are B[RFC]STM and AIFF, which are single-stream, yes
[10:21:23 CEST] <rcombs> so I guess it works for the formats I'm aware of
[11:06:07 CEST] <cone-563> ffmpeg 03George Boyle 07release/2.6:d9dd0cbe632e: avcodec/flacenc: Fix Invalid Rice order
[11:06:08 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:7e5a298bce17: avcodec/hevc_ps: Only discard overread VPS if a previous is available
[11:06:09 CEST] <cone-563> ffmpeg 03banastasov 07release/2.6:df408d213ad1: avcodec/dvbsubdec: Fix buf_size check in dvbsub_parse_display_definition_segment()
[11:06:10 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:a6bdfd9b46bb: avcodec/x86/h264_weight: handle weight1=128
[11:06:11 CEST] <cone-563> ffmpeg 03Paul B Mahol 07release/2.6:81d923f14bed: avcodec/exr: fix crash caused by merge
[11:06:12 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:ecd3cce0ec75: avfilter/x86/vf_hqdn3d: Fix register types
[11:06:13 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:e2054a8b55ee: avcodec/mpegvideo: Factor ff_mpv_reallocate_putbitbuffer() out
[11:06:14 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:678a189e5dd0: avcodec/mjpegenc_common: Use ff_mpv_reallocate_putbitbuffer()
[11:06:15 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.6:8c5d196f5026: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:19:16 CEST] <cone-563> ffmpeg 03Paul B Mahol 07release/2.5:f621ae6a6dd8: avcodec/exr: fix crash caused by merge
[11:19:17 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:e29f68cf9e33: avcodec/x86/h264_weight: handle weight1=128
[11:19:18 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:2d55d3483b8e: avcodec/hevc_ps: Only discard overread VPS if a previous is available
[11:19:19 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:d4c7a1ecb8ce: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:19:20 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:2e210e19b156: avcodec/mpegvideo: Factor ff_mpv_reallocate_putbitbuffer() out
[11:19:21 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:11f597ed8486: avcodec/mjpegenc_common: Use ff_mpv_reallocate_putbitbuffer()
[11:19:22 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.5:e9502ee26ae1: avfilter/x86/vf_hqdn3d: Fix register types
[11:29:40 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.4:65d61a4bbbb9: avfilter/x86/vf_hqdn3d: Fix register types
[11:29:41 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.4:57d094e3e399: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:29:42 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.4:bd28de1b4df5: avcodec/hevc_ps: Only discard overread VPS if a previous is available
[11:29:43 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.4:ff02eeafd8b3: avcodec/x86/h264_weight: handle weight1=128
[11:29:44 CEST] <cone-563> ffmpeg 03Paul B Mahol 07release/2.4:1a4cb3b29cda: avcodec/exr: fix crash caused by merge
[11:35:46 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.3:86c85b6f1a8e: avfilter/x86/vf_hqdn3d: Fix register types
[11:35:47 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.3:7d97587b1ae7: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:35:47 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.3:e1afd6ed6075: avcodec/x86/h264_weight: handle weight1=128
[11:35:48 CEST] <cone-563> ffmpeg 03Paul B Mahol 07release/2.3:17d9157c8229: avcodec/exr: fix crash caused by merge
[11:42:25 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.2:85d0c7355b61: avfilter/x86/vf_hqdn3d: Fix register types
[11:42:26 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.2:59a1e0cba062: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:47:00 CEST] <durandal_1707> michaelni: thx
[11:48:51 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.1:c77d226f7116: avfilter/x86/vf_hqdn3d: Fix register types
[11:48:52 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.1:a444857db318: avformat/mov: Mark avio context of decompressed atoms as seekable
[11:48:53 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.1:e0537e49a45e: avcodec/x86/h264_weight: handle weight1=128
[11:54:14 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.0:4a2be7027f7e: avfilter/x86/vf_hqdn3d: Fix register types
[11:59:01 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/1.2:1218e9d73dc3: avfilter/x86/vf_hqdn3d: Fix register types
[11:59:02 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/1.2:581bfd9c6f3a: avcodec/x86/h264_weight: handle weight1=128
[13:17:48 CEST] <cone-563> ffmpeg 03James Almer 07master:a9af9da6311b: library.mak: Workaround SDL redefining main and breaking fate tests on mingw
[13:30:46 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.7:8749b83e0be1: avcodec/libx264: Avoid reconfig on equivalent aspect ratios
[13:59:44 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07master:d277b05c518a: avcodec/fic: Add skip_cursor AVOption
[14:18:23 CEST] <BBB> so if I want to use special LDFLAGS for one test binary
[14:18:25 CEST] <BBB> how do I do that?
[14:18:37 CEST] <BBB> do we support something like TINY_SSIM_LDFLAGS = .. ?
[14:34:01 CEST] <ubitux> tiny_ssim$(EXESUF): LDFLAGS += ...
[14:34:09 CEST] <ubitux> ah, he left
[14:35:27 CEST] <Daemon404> he waited 7 whole minutes.
[15:13:27 CEST] <cone-563> ffmpeg 03Shivraj Patil 07master:ee3ef5fda2f1: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for hpel functions
[16:10:43 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07master:52e02a9e59f8: avutil/avstring: Do not print NULL
[17:41:27 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07master:8edc17b639c4: avcodec/dpxenc: implement write16/32 as functions
[18:18:54 CEST] <durandal_1707> rcombs: sending patches?
[18:49:32 CEST] <cone-563> ffmpeg 03wm4 07release/2.7:93d076b4fde1: avformat: clarify what package needs to be compiled with SSL support
[18:49:33 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:250cf2d4dae0: vp9: don't retain NULL as segmentation_map
[18:49:34 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:bea4894d0c42: postproc: fix unaligned access
[18:49:35 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.7:23614d09d511: avutil/avstring: Do not print NULL
[18:49:36 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.7:157dd527008c: avcodec/dpxenc: implement write16/32 as functions
[18:51:49 CEST] <BBB> no comments on my tiny_ssim_avf? :(
[18:51:58 CEST] <BBB> I was hoping somebody would pick it up and actually include it
[19:03:04 CEST] <durandal_1707> does it crash?
[19:10:14 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07release/2.7:7cbceb16ade6: Update for 2.7.1
[19:16:39 CEST] <BBB> durandal_1707: no, it just reads two files and outputs a ssim on stdout, pretty boring stuff
[19:16:52 CEST] <BBB> fairly similar to tiny_ssim I s'pose
[19:18:01 CEST] <durandal_1707> what if files have different dimensions?
[19:22:46 CEST] <durandal_1707> michaelni: how that code in dpxenc could crash?
[19:25:06 CEST] <BBB> I suppose it should error (same for colorspaces), but I didnt bother to test/implement
[19:25:18 CEST] <BBB> it could also support raw yuv with private options but I didnt do that either
[19:25:24 CEST] <BBB> long todo as you can see
[19:26:52 CEST] <durandal_1707> my plan is to add lavfi one with slice threading
[20:07:35 CEST] <michaelni> durandal_1707, one of the arguments was doing ++
[20:08:59 CEST] <lglinskih> kierank: now my test calculates crc of a frames, but two last frames are always lost. av_read_frame() just doesn't give me two last packets. I don't know why this is happening.
[20:20:55 CEST] <BBB> lglinskih: do you flush?
[20:21:14 CEST] <BBB> lglinskih: as in, do you call av_decode_videoN() with NULL?
[20:24:45 CEST] <lglinskih> BBB: no, no, explain it to me please )
[20:25:28 CEST] <BBB> https://ffmpeg.org/pipermail/ffmpeg-devel/2015-June/174340.html
[20:25:34 CEST] <BBB> look for the function read_frame()
[20:25:50 CEST] <BBB> it calls av_read_frame and then calls avcodec_decode_video2
[20:26:04 CEST] <BBB> then if av_read_frame fails (because of end of file), it calls avcodec_decode_video2 with NULL
[20:26:11 CEST] <BBB> this flushes the internal queue of packets
[21:06:11 CEST] <thrwaway> anybody feel like helping a newbie like myself to hunt down an issue very similar to https://trac.ffmpeg.org/ticket/2882 (ffmpeg can't save h264 stream from v4l2 webcam into file without recoding)? see the problem on a rapsberry pi with pi camera module. same problem on ffmpeg 1.2.12, ffmpeg 2.6.1 and git master. currently experimenting with st->need_parsing = AVSTREAM_PARSE_FULL vs AVSTREAM_PARSE_H
[21:06:17 CEST] <thrwaway> EADERS.
[21:24:34 CEST] <lglinskih> BBB: thank you!! it works ^_^
[21:24:59 CEST] <BBB> lglinskih: good!
[21:25:27 CEST] <BBB> thrwaway: sounds like you need parsing yes
[21:30:35 CEST] <anoop_r> how to crop 10 pixels from the left, 30 pixels from the right 20 pixels from the top and 5 pixels from the bottom borders of a video using ffmpeg
[21:31:26 CEST] <Compn> anoop_r : ask in #ffmpeg
[21:34:32 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:d89cd16afa97: matroskadec: check audio sample rate
[21:34:33 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:973e67e5d2d4: matroskadec: validate audio channels and bitdepth
[21:34:33 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:73e7fe8e648a: configure: make makeinfo_html check more robust
[21:34:35 CEST] <cone-563> ffmpeg 03Andreas Cadhalpun 07release/2.7:620197d1ffea: doc: avoid incorrect phrase 'allows to'
[22:05:44 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07master:9a2fc170b10e: avcodec/j2kenc: Support writing JP2 encapsulation
[22:05:45 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07master:7e2bdea764a5: avcodec/j2kenc: Use JP2 format by default
[22:11:59 CEST] <thrwaway> BBB: Chaning AVSTREAM_PARSE_HEADERS to AVSTREAM_PARSE_FULL in libavdevice/v4l2.c:v4l2_read_header() indeed seem to fix issue for me.  Here's a writeup with logs -> https://pastebin.mozilla.org/8837436
[22:17:21 CEST] <wm4> michaelni: can we have the 2.7.1 release already?
[22:17:30 CEST] <wm4> the complaints are starting to pile up
[22:36:21 CEST] <ubitux> BBB: 14:34 <ubitux> tiny_ssim$(EXESUF): LDFLAGS += ...
[22:36:57 CEST] <BBB> ooh nice, ty
[22:58:55 CEST] <cone-563> ffmpeg 03Michael Niedermayer 07n2.7.1:HEAD: avcodec/j2kenc: Use JP2 format by default
[00:00:00 CEST] --- Sat Jun 20 2015


More information about the Ffmpeg-devel-irc mailing list