[theora-dev] Theora streaming

[Dan Miller]

pages are limited to ~2^16 bytes, which is (for usual resolutions)
enough for a keyframe but not for a series of frames.  Right now, one
page = one frame, period.

Therefore, there will need to be a mechanism to see if a frame in a page
is a keyframe or not.  I think Monty has thought about this issue but
I'm not sure there's any way right now other than knowing what the
codec's private header data looks like.

> -----Original Message-----
> From: Arc [mailto:arc at indymedia.org]
> Sent: Thursday, February 27, 2003 3:14 PM
> To: theora-dev at xiph.org
> Subject: Re: [theora-dev] Theora streaming
> 
> 
> On Thu, Feb 27, 2003 at 02:58:33PM -0500, Dan Miller wrote:
> > 
> > > I thought
> > > you had to back up to the previous complete frame,
> > > then apply the differences in the following frames
> > > to generate a new complete frame where the splice
> > > was to happen. Are you going to handle editing to
> > > arbitrary frames? I believe you have to do something
> > > analogous in theora/ogg.
> > 
> > yup.
> 
> It's not really an issue of dropping "frames" off, but rather,
> Ogg packets/pages.  Trim a few off the beginning and suddenly you're a
> few fractions of a second into the clip.
> 
> A slightly different question would be how well VP3's keyframes are
> being matched to the beginning of Ogg packets/pages.  How would an
> editor know which packet contained the next keyframe, so it 
> knew how far
> it would have to trim to?
> 
> Of course, if the editor was really intelligent, it could potentially
> render a new keyframe at any hypothetical location, but this would
> require codec-specific coding.  It would be much better if every
> page started with a keyframe, thus cutting at any page border would be
> legal.
> 
> 
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'theora-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.