[Ffmpeg-devel-irc] ffmpeg-devel.log.20131011
burek
burek021 at gmail.com
Sat Oct 12 02:05:02 CEST 2013
[01:13] <cone-389> ffmpeg.git 03Michael Niedermayer 07master:a72bf5fd1188: ffmpeg: set the source_index for trivial filter graphs
[03:29] <BBB> llogan: yes, need exactly sse or sse2 only, not pre-sse or post-sse2
[08:28] <cone-717> ffmpeg.git 03Maxim Poliakovski 07master:be7641504737: atrac3: Remove unused gain compensation tables
[08:28] <cone-717> ffmpeg.git 03Maxim Poliakovski 07master:ed796fba761f: atrac3: Better name for IMDCT window initialization
[08:28] <cone-717> ffmpeg.git 03Michael Niedermayer 07master:964f9ca22994: Merge commit 'ed796fba761f4794bec7735d467c1b2c8e1858fe'
[09:20] <cone-717> ffmpeg.git 03Tim Walker 07master:5f5ada3dbf97: shorten: Fix out-of-array read
[09:21] <cone-717> ffmpeg.git 03Michael Niedermayer 07master:bb8ce36dc285: Merge commit '5f5ada3dbf97e306a74250ba8dcf8619ad59b020'
[09:28] <cone-717> ffmpeg.git 03Matthieu Bouron 07master:054454c63a2e: mxf: Add jpeg2000 codec to intra only codecs
[09:28] <cone-717> ffmpeg.git 03Michael Niedermayer 07master:08b22c0ea536: Merge remote-tracking branch 'qatar/master'
[09:29] <michaelni> mateo`, do you want to review "1007 18:22 To FFmpeg devel (1.2K)
[09:30] <michaelni> or should i just apply them ?
[09:30] <michaelni> they are of course tested ...
[10:24] <cone-717> ffmpeg.git 03Carl Eugen Hoyos 07master:d24da748c308: Add H.264 fourcc GAVC for GeoVision cameras.
[11:38] <durandal_1707> why frame_count starts from 0?
[11:38] <durandal_1707> bah
[12:04] <TheFluff_> what are you, a lua programmer?
[12:09] <durandal_1707> lua programmers counts from 1?
[12:09] <ubitux> yes
[12:10] <ubitux> array indexing starts at 1 iirc
[12:11] <Daemon404> yep.
[12:12] <durandal_1707> i am ex-lua programmer so i think lavfi should start counting from 1
[12:15] <Daemon404> durandal_1707, vf_lua.c time?\
[12:15] <durandal_1707> nope i'm using AVExpr and exposing link->frame_count via it
[12:16] <Daemon404> because everybody loves typing several kb of shit on the cmd line
[12:18] <durandal_1707> nobody comes with patch that gives alternative
[12:21] <durandal_1707> and what would vf_lua.c do if you can't call filter within filter
[12:21] <durandal_1707> of if filter can't control filtergraph
[12:22] <durandal_1707> and i guess recreating filtergraph resets all frame counts thus you are damned
[13:07] <ubitux> Daemon404: you can write the filtergraph in a file and you know it
[13:07] <ubitux> it's just more convenient on the cli 95% of the time
[13:08] <durandal_1707> what filtergraph in while?
[13:08] <ubitux> parse error
[13:09] <durandal_1707> can you give example of such filtergraph in file?
[13:10] <ubitux> i was refering to -filter_script and -filter_complex_script durandal_1707
[13:11] <durandal_1707> i use that, and that is just command line
[13:11] <ubitux> yeah but i was talking to Daemon404
[13:12] <durandal_1707> ok let me paraphrase what he said: because everybody loves typing several kb of shit in file
[13:13] <ubitux> maybe i misunderstood everything :)
[13:22] <Daemon404> [12:07] <@ubitux> Daemon404: you can write the filtergraph in a file and you know it
[13:22] <Daemon404> i actuall did not.
[15:26] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:2a7bdbf67efb: ffserver: strip odd chars from html error messages before sending them back
[15:26] <cone-717> ffmpeg.git 03Reinhard Tartler 07release/1.1:58287d3b10a2: update Changelog
[15:26] <cone-717> ffmpeg.git 03Reinhard Tartler 07release/1.1:bb81b2b2e06a: Fix top-level description
[15:27] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:835bc39b2608: Merge remote-tracking branch 'qatar/release/9' into release/1.1
[15:27] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:e31e66948d0a: Delete changelog
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:4ad0330b3dd1: doc/encoders: reformat libmp3lame doc
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:ed2c15eadc3b: doc/encoders: reformat and add some clarification in libtwolame doc
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:852ee0e0ad26: doc/encoders: Remove options that were not there when branch was cut from master
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:3eee21406a9f: doc/encoders: improve libvo-aacenc doc
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:c42fd4c6eee4: doc/ffmpeg-formats: Add documentation for 2 parameters that have been missing
[15:35] <cone-717> ffmpeg.git 03Timothy Gu 07release/1.1:a4acb5b90038: doc/encoders: add doc for AAC encoder
[15:35] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:eb3330b050ba: Merge remote-tracking branch 'TimothyGu/release/1.1' into release/1.1
[15:44] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:f0bb0aaaa7a8: avcodec/ffv1enc: update buffer check for 16bps
[15:44] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:0efb4ff86c20: avcodec/parser: reset indexes on realloc failure
[15:44] <cone-717> ffmpeg.git 03Michael Niedermayer 07release/1.1:4bc7c1ba8e9a: update for 1.1.7
[16:24] <cone-717> ffmpeg.git 03Carl Eugen Hoyos 07master:a12ab37d1f7a: lavf/riff.c: Fix GeoVision H.264 fourcc.
[16:41] <cone-717> ffmpeg.git 03Stefano Sabatini 07master:f02bab6d0978: doc/filters/scale: do not explicitly state the default swscale flags value
[16:54] <durandal_1707> if nobody wants to review telecine patches its time now (an no reviews like 'ok if tested' are not reviews)
[16:55] <durandal_1707> i will also make big changes to MA* file
[16:55] <cone-717> ffmpeg.git 03Luca Barbato 07master:c0de9a23c708: prores: Reject negative run and level values
[16:55] <cone-717> ffmpeg.git 03Michael Niedermayer 07master:e7fa0417b31a: Merge remote-tracking branch 'qatar/master'
[16:56] <durandal_1707> michaelni: are you gonna push qp?
[17:11] <cone-717> ffmpeg.git 03Michael Niedermayer 07master:dfb89fee2d2b: doc/codecs: document skip_alpha
[17:48] <durandal_1707> ubitux: why is decimate changing pts when it does not change time_base ?
[17:48] <ubitux> maybe it's broken
[17:49] <durandal_1707> it should change time_base?
[17:49] <durandal_1707> well it could... but there is no need ....
[17:50] <durandal_1707> also the fps thing seems fishy, filter just eats random frames...., why care about fps
[20:06] <cone-717> ffmpeg.git 03Timothy Gu 07master:a55e813e5394: doc/codecs: Cosmetics in the flags2 description
[20:06] <cone-717> ffmpeg.git 03Timothy Gu 07master:db4a677433a6: doc/codecs: Remove no longer existing options
[20:06] <cone-717> ffmpeg.git 03Timothy Gu 07master:1fdae1017c5b: doc/codecs: Add ignorecrop
[20:06] <cone-717> ffmpeg.git 03Timothy Gu 07master:024bf3a1c2d3: doc/codecs: Add missing mpeg2 aac profiles
[20:10] <ubitux> durandal_1707: maybe we should just divide the pts by the same frame count divider..
[20:10] <ubitux> in decimate
[20:11] <durandal_1707> why?
[20:11] <durandal_1707> test with audio where first pts != 0
[20:11] <durandal_1707> and where pts are not increasing one by one
[20:12] <durandal_1707> it probably should be rescaled once time_base is changed too
[20:12] <durandal_1707> but i do thing its not required
[20:12] <durandal_1707> *think
[20:12] <ubitux> mmh
[20:13] <durandal_1707> well if you want each frame have same duration than yes, you need to change time_base too
[20:13] <durandal_1707> and rescaled every pts
[20:15] <ubitux> but where is the pts adjustement supposed to happen?
[20:18] <durandal_1707> pts are a, b, c, d, e; when you drop for example c, you rescale all other frames to take same timeframe, which current code does, but it resets pts to 0 thus causes async
[20:22] <ubitux> yes i understand the problem with the current solution
[20:22] <ubitux> but i don't understand how you address the issue by simply dropping the pts negociation
[20:26] <durandal_1707> i don't :)
[20:27] <durandal_1707> i could pick min and max pts between frames and interpolate using them...
[20:27] <ubitux> durandal_1707: then maybe try frame->pts *= outlink->frame_rate
[20:28] <ubitux> ah er
[20:28] <ubitux> well, pick the correct multiplie
[20:28] <ubitux> r
[20:40] <ubitux> frame->pts = frame->pts * (dm->cycle - 1) / dm->cycle; something like this probably
[20:53] <durandal_1707> that cant work
[20:55] <cone-717> ffmpeg.git 03Timothy Gu 07fatal: ambiguous argument 'refs/tags/n1.1.7': unknown revision or path not in the working tree.
[20:55] <cone-717> Use '--' to separate paths from revisions
[20:55] <cone-717> refs/tags/n1.1.7:HEAD: doc/codecs: Add missing mpeg2 aac profiles
[22:30] <durandal11707> michaelni: can this be solved at all?
[22:30] <durandal11707> i see vf_fps code doesn't do anything similar
[00:00] --- Sat Oct 12 2013
More information about the Ffmpeg-devel-irc
mailing list