Ah yes I understand what you mean although I'm not sure it would bend time (4d) well apart from the long render time required.
I already tried making a highly subdivided Spline mesh, with alternating faces removed, reduced in size, then cloned and joined several times. it creates quite a large and slow to render model because so many calculations are required.
A bit like the holes in my lamps, where many extra light path calculations are required. So I would imagine that is why they use bitmaps to help simplify and reduce the number crunching process.
With bitmaps only using one poly or two triangles for a flat surface it keeps things very simple.
Ideally vectors would be better so they were resolution independent, plus the model file sizes would be smaller, but I guess at the moment it would add to much extra processing?
db4tech
