[vorbis-dev] video compression using textured polygons?

[Lourens Veen]

"Timothy J. Wood" wrote:
> 
[snip]
>    If you have previous frame, however, you could look at how the pixel
> values change as the edge moves over them.  This could lead you to
> determine that the edge is solid and as the edge moves over background
> pixels, they quickly take on the color of the surface attached to that
> edge.

It would seem that having the triangles to be non-overlapping is
actually rather restrictive. Allowing overlapping triangles as well
would improve compression I think, but it would also make the encoder a
lot more complicated. Then again, if we follow the standard way of
defining a bitstream/decoder format only then encoders can improve with
time.
 
>    Or, they might take on a function of the color of the surface attached
> to that edge (if you have a translucent surface), for example.  In
> either case, by exploiting frame coherence, you might be able to
> eliminate noise on the edges (or having to transmit extra cleanup data)
> by figuring out whether you can compute the data using alpha blending.

Ouch, I hadn't thought about translucent objects yet. Actually that will
probably severely hamper compression, since the encoder will see it as
an ever changing texture, unable to get any meaningful edge information
from it.

This is going to be very tricky methinks. Imagine 3 translucent layers
on top of each other. How is the encoder ever going to figure out what
is what? Needs heavy thought...

Lourens

--- >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 'vorbis-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.