Post
by galinette » Sat Mar 13, 2010 5:33 am
Mmhh, to me, that's more an announcement aimed at raising funds than real breakthrough (and this is very common in innovative technologies!)
Sounds as it's a kind of geometry search in a tree along the camera direction. In that case, it's true that rendering time is not O(number_of_geometry_elements) but rather O(log(number_of_geometry_elements)) and this allows quite a lot more details than standard polygon based scanline renderers. However, I'd be curious of seeing realtime moving geometry instead of having only camera movements. That's often a weakness in tree based rendering.
Off course, I may be wrong, that's only a guess.
Etienne