Mike Raiford wrote:
> I have a neat idea ..
> 
> Given some sort of structure formula, simulate the growth of a crystal 
> from one or several seed points.
> 

This is the closest I could find: http://gamma.cs.unc.edu/HYB_ICE/, the 
same group at UNC has a few other related papers as well.  For your 
purposes you probably could do without the fluid dynamics solver (or 
just use a simple heuristic) which would simplify the implementation.

I think a more interesting question along the lines that you suggest 
would be to provide simple high-level user controls of the shape of the 
crystal an then grow a crystal to match those requirements (instead of 
just being tailored for ice like this one).

I've also seen some things on building crystals with cellular automata 
of Voronoi diagrams or such, but I think they're equally difficult to 
control and don't give an nice results, but maybe they could be improved 
in this respect.

Post a reply to this message