1989: Hypertexture
|
Meanwhile, I joined the faculty at NYU and did all sorts of research.
One of the questions I was asking in 1988 and 1989 was
whether you could use procedural textures to unify
rendering and shape modeling.
I started to define volume-filling procedural textures
and render them by marching rays through the volumes,
accumulating density along the way and
using the density gradient to do lighting.
I worked with a student of mine, Eric Hoffert, to produce a SIGGRAPH paper in 1989. The technique is called hypertexture, officially because it is texture in a higher dimension, but actually because the word sounds like "hypertexture" and for some reason I thought this was funny. I offer no redeeming excuse. The image on the left is of a procedurally generated rock archway. Like all hypertextures, it's really a density cloud that's been "sharpened" to look like a solid object. I defined this hypertexture first by defining a space-filling function that has a smooth isosurface contour in the shape of an archway. Then I added to this function a fractal sum of noise functions: at each iteration of the sum I doubled the frequency and halved the amplitude. Finally, I applied a high gain to the density function, so that the transition from zero to one would be rapid (about two ray samples thick). When you march rays through this function, you get the image on the left. |