All this was more than a year ago. I have some test code dated January 1998, with more work done six months later. But as usual, my life was a whirling orbit of projects, and this wasn't high on the list. My first efforts didn't work very well, even on simple circular fields, so I let them lie.
Every few months, of course, the idea would pop back up. But I kept getting stuck on how to do it. Mathematically, I mean. I knew how to lay out cubic splines; I've had those principles memorized since college. But how does a spline -- a single curve on a two-dimensional plane -- produce a vector field, which has a value at every point on the plane?
In fact, my 1998 work didn't even get that far. Once you've got your vector field, you have to fill it with pen-strokes. I couldn't even do that respectably. A regular grid of strokes looks dull. A completely random field, white noise, seems full of clumps and gaps. I wanted pink noise. And then, of course, the strokes had to follow the field fluidly, which wasn't easy. You can see the whirlpool sample looks rather spiky around the center.
The random factors move in, the random factors move out. One day, after much work on the Zarcana rules, I said the hell with it; what would happen if I broke down and used calculus?