For modelling you could look into the catmull clarke formula for subdivisions (basicly smooth polys) just that having knowlage of algorithms some times makes a big difference, for instance I saw a presentation about the neuclus solver by Jos Stam, I now understand how it works and it makes a big difference in getting the best out of it, with minimal effort on the computer and/or the user as well as getting more accurate results!
"No pressure, no diamonds" Thomas Carlyle