Difference between revisions of "Talk:Small FFmpeg Tasks"

From MultimediaWiki
Jump to navigation Jump to search
(VC1 timestamps in m2ts)
(move details of mplayer demuxer to small task page)
Line 23: Line 23:
 
: Awesome. See, that's what this collaborative Wiki model is good for-- keeping me honest. :-) I'll get a test in [http://fate.multimedia.cx/ FATE] soon. --[[User:Multimedia Mike|Multimedia Mike]] 02:05, 31 March 2008 (EDT)
 
: Awesome. See, that's what this collaborative Wiki model is good for-- keeping me honest. :-) I'll get a test in [http://fate.multimedia.cx/ FATE] soon. --[[User:Multimedia Mike|Multimedia Mike]] 02:05, 31 March 2008 (EDT)
  
 
== new ffmpeg demuxers ==
 
from mplayer [http://svn.mplayerhq.hu/mplayer/trunk/DOCS/tech/wishlist?view=co wishlist]:
 
Port libmpdemux demuxers to libavformat or write your own from scratch.
 
  libmpdemux is considered deprecated and should eventually be removed.
 
  As of 2008-01-28, the following demuxers are missing from libavformat:
 
 
# TiVo (ty streams, not TiVo To Go)
 
# VIVO
 
# VQF
 
# XMMS
 
# libnemesi
 
# SL support for MPEG-TS
 
 
: Vivo is already on the list. Feel free to add more if you can clearly outline the end goal and the steps toward achieving that goal. --[[User:Multimedia Mike|Multimedia Mike]] 14:50, 4 April 2008 (EDT)
 
  
 
== VC1 timestamps in m2ts ==
 
== VC1 timestamps in m2ts ==

Revision as of 11:45, 24 March 2009

Patch Rules

should we link or quote some ffmpeg patch rules? e.g. tabs are forbidden, use diff -u or svn diff or (insert git command). it seems like this is a common mistake for SoC students. -- Compn 10:53, 29 March 2008 (EDT)

Well the mailing list page has a link to the MPlayer patch guidelines, but it's missing a few things such as:
  • No trailing whitespace
  • Must not break the regression tests
  • If it does change the regression tests, you must update regression.ref
  • Must check bounds on all input and must not allow overflow or other out-of-bounds condition
  • Use meaningful variable names
  • Globals/externs should have ff_ prefix to avoid naming conflicts
  • Doxygen comments
  • No breaking C standards (eg using underscore prefixes)
  • Keep { } tidy
  • No code duplication
  • Workarounds for compiler bugs (eg MSVC) must not break other platforms
Anything else? -- Mkhodor 13:18, 31 March 2008 (EDT)
er, the whole rule list is availible here http://ffmpeg.mplayerhq.hu/general.html#SEC20 , feel free to add it-- Compn 14:09, 31 March 2008 (EDT)

Delphine CIN?

But we already have it in FFmpeg - look at libavcodec/dsicinav.c and libavformat/dsicin.c --Kostya 01:43, 31 March 2008 (EDT)

Awesome. See, that's what this collaborative Wiki model is good for-- keeping me honest. :-) I'll get a test in FATE soon. --Multimedia Mike 02:05, 31 March 2008 (EDT)


VC1 timestamps in m2ts

was: Add support for newer codecs such as Dirac and VC1 to FFmpeg's Matroska muxer.

Dirac should mux perfectly as of r16261. What about muxing VC1 doesn't work? A quick -vcodec copy from wmv didn't throw any errors. -Yuvi 05:31, 3 January 2009 (EST)
Yes, WMV3 works fine. There was a report that muxing VC1 from m2ts didn't work. If this does work then feel free to remove the task. Also the demuxer doesn't recognize V_VC1. -Mkhodor 22:58, 4 January 2009 (EST)
From m2ts, ffmpeg gives "error, pts < dts" which means that the timestamps coming out of the m2ts demuxer are invalid. I dunno if it would be easier to fix the demuxer or just extend the parser to fix/fill them in. For V_VC1, is that actually used by a muxer now? According to http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2007-August/034825.html it was a brief mistake that isn't supposed to be used. -Yuvi 02:50, 5 January 2009 (EST)
I don't know of any other muxers that use it, so I made a note on the Matroska page that it is deprecated. -Mkhodor 10:47, 13 January 2009 (EST)