[Ffmpeg-devel-irc] ffmpeg-devel.log.20160813

burek burek021 at gmail.com
Sun Aug 14 03:05:04 EEST 2016


[00:05:43 CEST] <cone-771> ffmpeg 03Mark Reid 07master:6051bb338c5b: libavcodec/dnxhdenc: add edge emulate for dnxhr
[00:05:43 CEST] <cone-771> ffmpeg 03Mark Reid 07master:44ac2b9b086d: tests/fate/vcodec: add dnxhr edge tests
[00:05:43 CEST] <cone-771> ffmpeg 03Petru Rares Sincraian 07master:777697a62c00: fate: add test for compand filter
[00:23:32 CEST] <Compn> llogan : k
[00:23:47 CEST] <Compn> llogan : i just looked at it today, i couldnt remember if you said something about it earlier :D
[00:25:44 CEST] <llogan> Compn: i can't remember either or confusing it with last trip a few weeks ago
[01:37:00 CEST] <cone-771> ffmpeg 03Carl Eugen Hoyos 07master:bc7031265441: lavc/screenpresso: Fix 32bit pix_fmt, there is no transparency.
[01:46:46 CEST] <cone-771> ffmpeg 03Carl Eugen Hoyos 07master:ceab04fb5db8: lavc/raw: Support QT b64a ARGB64 rawvideo.
[06:54:23 CEST] <fa0> Will we see in any future versions a time progress, showing remaining time left?
[11:05:11 CEST] <cone-832> ffmpeg 03Jan Sebechlebsky 07master:7d5501be1e42: avcodec/bsf: Set EOF flag only in pkt == NULL
[11:05:11 CEST] <cone-832> ffmpeg 03Jan Sebechlebsky 07master:bfdca87ab55c: avcodec/bsf: Forbid packet without payload in av_bsf_send_packet
[11:38:01 CEST] <cone-832> ffmpeg 03Jan Sebechlebsky 07master:2ab52644ee47: doc/bsfs: Fix bsf options divider in documentation
[12:14:27 CEST] <cone-832> ffmpeg 03Carl Eugen Hoyos 07master:79dc94a63b30: configure: Check for arc4random() header definition.
[12:49:45 CEST] <cone-832> ffmpeg 03Jan Sebechlebsky 07master:7ae52f8a6b4a: avcodec/bsf: Add ff_bsf_get_packet_ref() function
[13:01:17 CEST] <cone-832> ffmpeg 03Paul B Mahol 07master:ee56d9bc89c2: avfilter/vf_signalstats: add >8 bit depth support
[13:01:18 CEST] <cone-832> ffmpeg 03Paul B Mahol 07master:3a81775bdee7: avfilter/vf_signalstats: measure video bitdepth
[14:26:30 CEST] <Compn> wasnt there a status bar added to ffmpeg ?
[14:26:34 CEST] <Compn> hmm
[14:27:22 CEST] <Compn> er progress bar
[14:38:18 CEST] <durandal_1707> Compn: no bars please
[15:30:07 CEST] <cone-832> ffmpeg 03Vivekanand 07release/3.0:f3c1a76ffdd8: avformat/allformats: Making av_register_all() thread-safe.
[15:30:08 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:b2a2b1a88b3e: avfilter/af_amix: dont fail if there are no samples in output_frame()
[15:30:09 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:9f1e8b4fb28f: avcodec/bmp_parser: Fix frame_start_found in cross frame cases
[15:30:10 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:fca62cc6a709: avcodec/bmp_parser: Fix remaining size
[15:30:11 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:76a239768f63: avcodec/bmp_parser: reset state
[15:30:12 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:603fd4f77143: avcodec/bmp_parser: Check fsize
[15:30:13 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:b3fe93e73b18: avformat/mpegts: Do not trust BSSD descriptor, it is sometimes not an S302M stream
[15:30:14 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:9cf85419a32d: avcodec/utils: check skip_samples signedness
[15:30:15 CEST] <cone-832> ffmpeg 03Umair Khan 07release/3.0:f50f7adf7a9a: avcodec/alsdec: fix max bits in ltp prefix code
[15:30:16 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:a004e2e4f138: avcodec/alsdec: Check r to prevent out of array read
[15:30:17 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:e7a064efa6ac: avcodec/h264: Fix off by 1 context count
[15:30:18 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:7216068e4255: MAINTAINERs cleanup (remove myself from things i de facto dont maintain)
[15:30:19 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:679e869496ed: avcodec/mpegvideo: Do not clear the parse context during init
[15:30:20 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:5bf11223dd8e: avcodec/cfhd: Set dimensions unconditionally
[15:30:21 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:280587b4ab10: avformat/mp3dec: Increase probe score slightly when the whole data from begin to end is mp3
[15:30:22 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:2577cfd207ee: avcodec/mpc8: Correct end truncation
[15:30:23 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:4e5d1c45f111: avformat/mpegts: adjust probe score for low check_count
[15:30:24 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:694d37eb52ad: avfilter/vf_telecine: Make frame writable before writing into it
[15:30:25 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:19b86db2b32f: avcodec: Add avpriv_codec_get_cap_skip_frame_fill_param()
[15:30:26 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:d13ba3960659: tests/api/api-codec-param-test: Do not directly access caps_internal
[15:30:27 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:13b685222584: avformat/flvdec: Accept last size if its off by 1
[15:30:28 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:96ce6f64cc33: avcodec/flac_parser: Raise threshold for detecting invalid data
[15:30:29 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:c34f0616d9c7: avformat/flacdec: Fix seeking close to EOF
[15:30:30 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:811d560378f3: avformat/format: Fix registering a format more than once and related races
[15:30:31 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:a5680d83af26: avformat/mov: Check sample size
[15:30:32 CEST] <cone-832> ffmpeg 03Sasi Inguva 07release/3.0:6e13acaadd5c: libx264: Increase x264 opts character limit to 4096
[15:30:33 CEST] <cone-832> ffmpeg 03Kacper MichajBow 07release/3.0:6e6609f5dd61: libavutil/opt: Small bugfix in example.
[15:30:34 CEST] <cone-832> ffmpeg 03Kacper MichajBow 07release/3.0:c5757266219c: libavformat/rtpdec_asf: zero initialize the AVIOContext struct
[15:30:35 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:8b3d9c33d617: avcodec/vp9_parser: Check the input frame sizes for being consistent
[15:30:36 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:7876fd679d98: ffplay: Fix invalid array index
[15:30:37 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:2a8441e949ee: avformat/oggdec: Fix integer overflow with invalid pts
[15:30:38 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:e7c00a281423: avcodec/ffv1enc: Fix assertion failure with non zero bits per sample
[15:30:39 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:ab9616428ac0: avcodec/raw: Fix decoding of ilacetest.mov
[15:30:40 CEST] <cone-832> ffmpeg 03Hendrik Leppkes 07release/3.0:09d185406387: cmdutils: remove the current working directory from the DLL search path on win32
[15:30:41 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:534dc63089d9: Update for 3.0.3
[15:30:42 CEST] <cone-832> ffmpeg 03Michael Niedermayer 07release/3.0:66bd740e0ba5: avcodec/h264: Put context_count check back
[17:46:47 CEST] <ubitux> durandal_1707: opinion on having some lavfi filter-test progs for finer grain tests?
[17:47:20 CEST] <ubitux> i'm back on nlmeans and some code i'm writing is relatively sensitive; i'm currently using an external prog to test very specific part
[17:47:31 CEST] <durandal_1707> ubitux: hmm
[17:47:37 CEST] <ubitux> i'm wondering about integrating that test code inside the filter scoped into a #if TEST ...
[17:48:03 CEST] <durandal_1707> what it tests?
[17:48:33 CEST] <ubitux> it's basically computing a bunch of integrals with all kind of parameters
[17:49:18 CEST] <ubitux> https://github.com/ubitux/FFmpeg/blob/3731bc3872bbf8a870d67196b440cf48aeaaba94/ii.c
[17:49:42 CEST] <ubitux> (not actually doing any test currently, but computing and printing; i'm going to add some checks now)
[17:52:01 CEST] <durandal_1707> integrals, uh oh
[17:52:28 CEST] <ubitux> yeah i was trying to make it possible to make it much faster later on
[17:52:42 CEST] <ubitux> so i had to split boundary area computation from the rest
[17:53:30 CEST] <ubitux> the unsafe version covering most of the area is thus very simple now and relatively easy to simdify
[17:54:01 CEST] <ubitux> i'll need to push it further to have it modN width though, but that will come later 
[17:54:15 CEST] <ubitux> even in C it should be much better
[17:54:37 CEST] <ubitux> anyway, back to the topic, i was planning to have that main() in between TEST ifdefery within the filter
[17:54:46 CEST] <ubitux> so it could be tested specifically
[17:56:57 CEST] <durandal_1707> but?
[17:57:40 CEST] <ubitux> but that will be the first filter to do so
[17:57:48 CEST] <ubitux> so i was just poking
[18:00:27 CEST] <ubitux> i hope to be done with the filter relatively soon btw
[18:00:39 CEST] <ubitux> i just didn't have time to focus on recently
[18:02:28 CEST] <durandal_1707> it could become useful in future
[18:03:11 CEST] <cone-832> ffmpeg 03Rostislav Pehlivanov 07master:c92cc2d559de: aacenc: add a faster version of twoloop as the "fast" coder
[18:03:13 CEST] <cone-832> ffmpeg 03Rostislav Pehlivanov 07master:fb0abb34cb5a: aacenc: unmark the fast coder as experimental
[18:07:08 CEST] <durandal_1707> is vlc ts demuxer better than ours? Sorry if asked already.
[18:13:15 CEST] <durandal_1707> Anybody tried surroundscope? With 5.1 audio.
[18:18:30 CEST] <jamrial> durandal_1707: doesn't vlc use lavf?
[18:19:53 CEST] <durandal_1707> jamrial: not always
[18:20:26 CEST] <ubitux> http://git.videolan.org/?p=vlc.git;a=tree;f=modules/demux/mpeg;hb=HEAD
[18:20:34 CEST] <jamrial> yeah, just looked at the git repo. lavf wrapper but also a bunch of other demuxer
[18:23:14 CEST] <jamrial> heh, custom matroska demuxer. nobody likes lavf's :p
[18:23:27 CEST] <ubitux> ordered chaptering maybe?
[18:23:32 CEST] <jamrial> yeah, most likely
[18:23:55 CEST] <ubitux> i wonder if mpv ended up switching or not
[18:24:00 CEST] <RiCON> atomnuker: "and some may be removed in the future"  should've lost the "some"?
[18:24:43 CEST] <jamrial> i question the need for a custom tta demuxer in vlc, though. it's a very simple format and lavf supports it perfectly
[18:25:07 CEST] <jamrial> guess they wrote it before lavf's was commited and they never bothered to replace it
[18:25:30 CEST] <durandal_1707> ubitux: mpv have own version, lavfilters own, and so on
[18:25:43 CEST] <RiCON> ubitux: https://github.com/mpv-player/mpv/wiki/libavformat-mkv-checklist
[18:25:52 CEST] <ubitux> durandal_1707: yeah i know mpv has its own one, but i remember mpv wondering about switching to lavf one
[18:26:07 CEST] <ubitux> RiCON: ah, there you go; thanks
[18:26:32 CEST] <Compn> we should steal all other demuxers ...
[18:26:34 CEST] <Compn> make wrapper
[18:26:36 CEST] Action: Compn runs
[18:28:03 CEST] <ubitux> Compn: you need to understand that fundamental issues involve api thinking
[18:28:21 CEST] <jamrial> ubitux: doubt it since wm4 was interested in ordered chapters functionality
[18:28:23 CEST] <ubitux> it's not just a bucket in which you throw building blocks code
[18:29:00 CEST] <ubitux> jamrial: according to the wiki it's handled at a higher level; so maybe exporting as side data or whatever may do
[18:29:05 CEST] <Compn> adding more demuxers requires api changes ?
[18:29:06 CEST] <Compn> heh
[18:29:25 CEST] <ubitux> but then again, you probably want to design a side data that can handle more than just matroska
[18:29:49 CEST] <Compn> most players out there (mpv, mplayer, vlc) can use different demuxer libs (lavf, libmatroska, libogg etc)
[18:29:49 CEST] <ubitux> Compn: the reason most players use their own mkv demuxer is because of api limitation
[18:30:19 CEST] <Compn> not just talking about mkv
[18:30:22 CEST] <Compn> but all demuxers
[18:30:38 CEST] <Compn> nevermind
[18:35:46 CEST] <cone-832> ffmpeg 03Rostislav Pehlivanov 07master:6612d0493356: aacenc: fix various typos and an error message
[18:35:50 CEST] <kierank> vlc uses its own demuxes for everything apart from flv afaik
[19:55:11 CEST] <kierank> atomnuker: any idea where we did uyvy -> planar asm?
[19:57:14 CEST] <kierank> ah void ff_uyvy_to_planar_10_c(uint16_t *y, uint16_t *u, uint16_t *v, const uint16_t *l, const int64_t width)
[20:36:54 CEST] <kierank> aballier: https://github.com/kierank/ffmpeg-sdi/blob/master/libavcodec/x86/sdienc.asm#L256
[20:36:58 CEST] <kierank> atomnuker: ^
[20:37:00 CEST] <kierank> what does that SWAP do?
[21:05:01 CEST] <jamrial> kierank: pre processor macro. swaps m1 with m0 for every following instruction
[21:05:13 CEST] <kierank> yes, but why is the question
[21:05:16 CEST] <kierank> as well
[21:05:21 CEST] <jamrial> kinda pointless in this case. the mova bellow could just use m0
[21:05:37 CEST] <kierank> yeah
[21:05:40 CEST] <kierank> might be some cruft
[21:06:16 CEST] <jamrial> guess it was copied from the SBUTTERLFY macro
[21:07:04 CEST] <kierank> maybe
[21:26:18 CEST] <durandal_1707> michaelni: what about adding int64 sample format?
[21:27:41 CEST] <kierank> for audio?
[21:27:43 CEST] <kierank> seriously
[21:28:04 CEST] <durandal_1707> sample was uploaded
[21:28:29 CEST] <durandal_1707> for audiophile killer feature
[21:28:53 CEST] <kierank> link?
[21:41:05 CEST] <durandal_1707> http://trac.ffmpeg.org/ticket/5649
[21:41:43 CEST] <durandal_1707> kierank:^
[21:49:52 CEST] <kierank> omg what a joke
[21:51:57 CEST] <nevcairiel> even as a processing format int64 makes no sense, you would use double then
[22:04:10 CEST] <durandal_1707> its export format like u8
[22:05:51 CEST] <durandal_1707> and its bitexact unlike dbl across platforms
[00:00:00 CEST] --- Sun Aug 14 2016


More information about the Ffmpeg-devel-irc mailing list