Hey guys,

I've created a frame accurate seeking library based on the standard  
ffmpeg libraries that I've convinced my company to LGPL and release.  
It's meant to support video processing applications that frequently  
require random access in a codec/format independent way. I've tried to  
preserve the full breadth of the formats and codecs that  
av_seek_frame() works with. I've tested it pretty extensively with our  
internal video set and with http://samples.mplayerhq.hu. I've uploaded  
a *very* basic version to sourceforge. You can see the SVN here:

http://ffmpeg-fas.svn.sourceforge.net/viewvc/ffmpeg-fas/

NOTE: This version still contains quite a bit of code that was meant  
for our internal use and so some of the design choices may seem  
awkward or totally unnecessary and will require a good deal of  
cleaning up. I'm releasing it in this state because I want to gauge if  
there is any interest in a library like this before refining it into a  
more.. shall we say.. usable state.

It provides a very simple interface for video processing applications  
to decode video, including seeking. It works by building a seek table  
through normal decoding of a video that can be saved and loaded later.  
This table provides the necessary information for frame-accurate  
seeking.

Some useful files:
dump_frames.c - shows how the simplified decoding interface works  
(without seeking)
seek_test.c - shows the test I've been using to test frame-accurate  
seeking and serves as a seek example
generate_seek_table.c - shows the current way I create seek tables  
using libavformat and libavcodec

I'd love to hear any questions/comments/concerns on this set of  
functionality and its usefulness. Also any improvements to the  
methodology would be greatly appreciated.

Regards,
Louis Brandy
_______________________________________________
libav-user mailing list
[email protected]
https://lists.mplayerhq.hu/mailman/listinfo/libav-user

Reply via email to