[theora-dev] Analogue artifact estimation

[jbradford at dial.pipex.com]

> > What I'm thinking of is a true psycho-visual model that tracks the human
> > eye/brain's ability to perceive things in various frequency/spatial
> > domains, ie least noticeable differences and such.  I have yet to see any
> > compelling research in this area that has been effectively reduced to
> > practice in designing compression algorithms.
> 
> This would certainly be extremely cool, and likely have huge benefits,
> especially if a free software project were to get to this before anyone else.

That's half the reason I posted my ideas to the list, because I cannot afford to do patent searches, etc, so I thought at least making them public stops other people from patenting them, (or does it?).

> Of course, it would likely also involve a lot of work, but it sounds like
> we've got time and at least a few people who know what they're doing. 

I think that some of these posts should be being crossposted between Theora-dev and Tarkin-dev.

> This might be a good time time for me to go off-topic a little bit and ask
> about how this project is supposed to work. As I'm understanding it, and I
> may very well be totally wrong, the purpose of Theora is to create a way to
> put the VP3.2 video codec and the Vorbis audio codec into the Ogg container,
> thus creating an essentially patent-free audio and video format to compete
> with MPEG, Windows Media, RealMedia and the like.

Essentially, yes, that's right.

> Meanwhile, there is another project, Tarkin (aka W3D) which is trying to
> create a totally new, multi-dimensional video codec, with a release date
> sometime in the far-off future. Tarkin, as it is, seems a bit dead in the water.

I thought it was dead too, but I think there just isn't much to say - all of the developers are just sitting around thinking about new ideas at the moment, and the list is just used to bat them off each other.

> What I guess I'm getting around to is this: Is this kind of work
> (psycho-visual models, filtering, etc.) on the video codec itself outside
> the scope of Theora?

No.

> Is Theora really 2 projects, one for the container, one for the video codec?

Not really.

> Do we want to make drastic changes to the codec, or should we hold off until
> Theora 1.0 is out the door? According to the FAQs:
> 
> "Theora will be almost entirely based upon the VP3 codec designed by On2." 

That doesn't stop us from doing anything we like in the pre-processing stage.

The way I see it, some of the ideas that are destined for Tarkin, could be tried out on Theora in the form of a pre-processor.  Certainly my reverse 3-2 pulldown, and dropout compensation ideas could be implemented easily in that way.

> Do we really think that it will take a year to figure out how to put VP3
> into the Ogg container?

Well, there are issues with streaming, like making the stream gracefully degrade over lower and lower bandwidth, so it's not a five minute job, but a year seems like a long time to me...

> Isn't a lot of this work already done with the "OGM" format that is being
> used in the "DVD Back-Up" scene? (http://www.bunkus.org/videotools/ogmtools/)

No idea about that, sorry.

> Am I totally missing something here?

The VP3 codec can be implemented in different ways, some better than others, just as there are good and bad MPEG encoders, there could be good and bad VP3 encoders.  If somebody has an idea for improving the encoder, that's very relevant.

John
--- >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.