Re: [Fresco-devel] [ReFresco 06] 3d support in the scene graph

Nathaniel Smith wrote:

> What makes much more sense, and has a clean semantics, is to have both
> 3d subgraphs and 2d subgraphs.  A 3d subgraph can be the child of a
> "pane" 2d graphic, and a 2d subgraph can be the texture of a 3d
> graphic.  In this scheme, you know when you have 2d semantics, you
> know when you have 3d semantics, and you have clear places to describe
> how they interact.

I agree. Now that we don't expose the scenegraph traversal to the client
we can do that more easily, i.e. we can in fact define two entirely
different traversal types (one occlusion / z-buffer based, the other
stack based). This would have been much more difficult with the requirement
to expose all this through IDL.

> This approach also clearly breaks down the problem into pieces; it's
> possible to get 2d working without dealing with 3d at all, for
> instance -- this is useful for getting things working quickly, and
> also in the long run for things like embedded implementations that
> would like to eschew 3d altogether.

youp. There will be 'gateway' nodes that define the embedding of one
type into the other. In fact, that in itself is nothing new, as we
have been discussing such gateways in the past (and there are more
such gateways, for example to map a subgraph that deals with layout
into a node that is rigid, etc., etc.

> (I suppose I should mention somewhere that this is exactly the
> approach taken by Microsoft's Avalon, though the idea came up in our
> discussions years before they ever published anything...)

Then let's find the old discussions so nobody can claim we do something
illegal here. The idea is not that original, but with the patent agony
we are seeing all around us you never know...

Regards,
		Stefan