Here are a few iterations. Shapely is pretty great for generating Voronoi lines and confining them to an existing polygon—in this case the outline of the remigium.
They look like Voronoi patterns but they don’t look like insect wings right now.
I suspect what I need is Lloyd’s algorithm, which will evenly space the inhibitory centers by iteratively replacing each one with the centroid of its Voronoi polygon.
By the way, I’m smelling some stinky code again. A lot of these calculations happen within the context one of a single remigium—the more I use that word the less I like it. Oof. I’m using it specifically to mean the space between primary veins whereas the remigium is the area of the wing that’s used for propulsion.
No joke that in computer science, naming things is hard.
Well, too bad. I’ve spent enough time figuring out a name for it. I’m sticking with remigium.
Back to the code smell. These calculations happen within a per-remigium context. Inhibitory centers and cross-veins are not affected by other remigia. But right now all inhibitory centers from all remigia are stored in a flat list. Same with the Voronoi polygons. What I should do instead is have a class representing a single remigium and do the calculations for the cross-veins in there.
Right now, the PrimaryVeins class is used like so:
pv = PrimaryVeins(parameters)
if pv.has_collision():
# Regenerate
pass
...
pv.render_to(surf)
Maybe CrossVeins should look like this:
cv = CrossVeins(parameters, pv)
...
cv.render_to(surf)
And within CrossVeins I can maintain a list of Remigium instances.
…the other problem with remigium is it’s a fancy word that nobody knows so everyone has to look it up.