RE: How about concretization?

Andy Norris (andyn@texas.net)
Mon, 14 Jan 1980 18:48:18 -0500

At 04:46 PM 4/25/95 EDT, Brandon Van every wrote:
>
> I don't mean this in a discouraging way, but it seems like it ought to be
> inherently impossible to represent Escher in 3D. His work plays off the
> unique aspects of representing perspective objects in 2D artwork, and as
> soon as you put it in 3D, you lose the ambiguity between different ways of
> seeing depth into perspective artwork.
>
>Your point then, is that you can't build a 3D model and expect it to
>look like 2D Escher. Fortunately, computer graphics isn't really
>about 3D models, but 2D projections.

[and in another post]

>Then we have to modify the scene as the viewer moves, so as to retain
>the illusion. Maintaining the illusion for multiple viewers would be
>a real challenge. Essentially, you'd have to have everyone view a
>"4D" world: that's 3D + extra processing.

I do like your idea of a 4D world--where you process to get "impossible"
perspective shifts as you move through the world. I think that could probably
get you the famous Escher work with the different-perspective staircases
(please excuse my ignorance of the title).

OTOH, there are some 2D perspective issues that inherently make no sense
from other perspectives: take constellations, for example. A collection
of stars may look like the Big Dipper from Earth, but if you move very far
around the galaxy, it looks nothing like that, because the stars are
different distances from you.

Many Escher works rely on analogous effects, like the picture of the hands
drawing the hands. If you had a true 3D, with a flat plane drawing visibly
shifting into a 3D rendering, it would be a very different effect, because
the flat drawing would only look like 3D from certain angles. You could
design it so that the plane of the drawing was always perpendicular to the
viewing angle, but that too, would yield a very different effect.

One last thought: would any of it look right through a stereoscopic display?
I would think that would eliminate any of the necessary ambiguity.

--Andy
andyn@texas.net