I wanted to make an ellipsoid, following the same premises as for the spheres: the Voronoi cells and Delaunay triangulation of points on the surface.

After a little thought, I realised this was a lot harder than it seemed. For starters, all along, I've been playing fast and loose with whether I was considering the sphere a two dimensional object, or a three dimensional one.

In particular, my code assumes that the angle (from the center) between two points is a useful metric for how far apart they are, which is no longer the case, if we stretch the sphere into an ellipsoid.

Since the concept of making an ellipsoidal model had been kicking
around in my head for a long time, and I wanted to just get on with
making one I decided to go with simply stretching the spherical model.
What's constructed isn't truely the Delaunay triangulation, and its dual.
The display code ended up being quite a headache, since the obvious
way to get an ellipse in postscript is to stretch the coordinate space,
then draw a circle. I had to keep track of what angle I wanted in both
coordinate spaces, and wrap everything gsave/grestore pairs to keep the
different transformations from interfering with each other.

The construction of the first ellipsoidal segment brought up an unanticipated problem; where before it had been simple to eyeball where the tabs should go, and to score the folds, the uneven looking pieces suddenly made that a lot harder. I went back and added an extra dotted segment to give the direction of the tab (I still don't use more than half an inch of it), and an x that's where the center would be, relative to that piece. The x is just to line up the ruler against; if it lands on top of another strip, it doesn't matter. Here's the postscript files: Delaunay, Voronoi.

There was a bug in my script that made one of the pieces of the Voronoi cell model broken, having two of its edges swapped and mangled. I didn't notice it until after I'd made the rest of the model. Since I was generally disappointed with them, the pair sat around incomplete and half painted for several months, until while I was working on the documentation, I made a replacement strip and out the gold red and blue color scheme (based on the number of edges each polygon has) for the Voronoi model.

I'm still unhappy with the green and gold color scheme, and dark colored
models seem to photograph poorly, so I'm not bothering with any close
up pictures of it.