[mythtv] NuppelVideo format, keyframes, and the cutlist

Geoffrey Hausheer ou401cru02 at sneakemail.com
Wed Jun 18 09:04:00 EDT 2003


On Wed, 18 Jun 2003 02:19:04 -0400, "Isaac Richards ijr-at-po.cwru.edu
|mythtv/1.0-Allow|" <e6kbhbk35r0t at sneakemail.com> said:
> Hmm.. though.. Had you thought of a new frametype that just says 'skip
> the 
> next X frames'?  That way you should be able to just cut on key-frame 
> boundaries, and then just skip up to the real end of the cut
> automatically 
> during playback..  I dunno, though

Actually this is sometghing to think about.  At one time you (or someone)
had said that they didn't want to limit commercial-clipping to keyframe
boundaries.  If we do (at least for 'clipped' files), I do have the
option of either (a) rounding to the nearest keyframe (which means that
you could lose up to 1.2 seconds of your show, or (b)
conservatively-rounding to the next keyframe (which means having to watch
up to 2.4 seconds of commercials).  I had initially just rejected this
outright, but there was no good reason to do so.  If I do this, I don't
think I need a new frametype (I'm going to rewrite the frame-headers
regardless), and all of this code isn't needed.  It also hugely
simplifies the reencoding proces, since I don't need to do any encoding
at all.  Perhaps that is an acceptable tradeoff (trading off user
experience for file/code complexity)?

As for the mplayer patch, I haven't looked at how it does fast-forward,
but fast-froward/rewind are the only things that require the new frame
data, and the worst case is that you'll see ~1 second of garbage before
the next key-frame arives when using these functions (until the patch
gets updated).

The more I think about it, the more I'm inclined to just snap to
key-frames.  I'm thnking I will pull out this code, and do the Reencoder
that way unless there are serious objections.  If the results aren't
appealing after I'm done, perhaps I'll revisit these changes.

Anyone want to weigh in on this?

.Geoff
-- 
  Geoffrey Hausheer
  XXXXXXXXXXXXXXXXXXXXX


More information about the mythtv-dev mailing list