You can see on the cube: The above images were independently scaled to power-of-2 dimensions (e.g., 512x1024) in order to make OpenGL (graphics/Mesa) happy. Then a program (see the editors/emacs window with source code, upper right?) read the files and had graphics/Mesa render the images as textures on the facets. Program controls let me put the eye & lights where I thought they had the best effect. (The program was one that I wrote this fall, and readily adapted to the purpose at hand. )

Image contributed by Richard Rauch <rkr@rkr.kcnet.com>.