But you guys keep missing the point: computer graphics is _not_ 3d.
It's all 2d _projections_ of stuff. Just because you can't build a 3d
model of many Escher works, does _not_ mean you can't create
continuously moving 2d perspective illusions which resemble his works.
There is probably a mathematically correct way to do this. Even if
there isn't, it can probably be simulated through extra processing of
the viewer's perspective as he/she moves. So you run out of illusion
after traversing 180 degrees or whatever. Big deal, you can always
put up a wall to block the user from moving any farther than that.
Well, actually that's getting into implementation specifics, which is
somewhat outside of the charter for this list, I believe. Sterographic
viewing is NOT strictly 3d transformed to 2d, and may be what many
people invision when they think of VR. Also, the problem with doing
the Escher transform, is that the actual physical representation of
the space would change based on the 2d results as the viewer moved,
basically a reverse of how the transform normally works. The point is
developing a 3d model to represent Esherian space and have a standard
viewing transform for it, which is most likely not possible. What your
suggesting would involve more than one viewing engine, basically one
for every view produced, as there is no standard algorithm for
producing Escherian effects.
What would be more interesting, I think, is to take the 3d space and
play games with representing 4d space... i.e. extrapolate what Escher
might do with the same tools.
Ray