Can you go into more detail about the raindrop "simulation" on the window? I love the idea, but it seems like they larger drops are just moving random distances at random times, clearing a path as they go.

That's a nice first approximation, but it would be cool to put some real dynamics in there. For instance, I noticed they are not "absorbing" droplets as they fall; if a smaller droplet rolls over a bigger drop, the bigger one just disappears. Also, the movement for smaller drops should be initiated when another small drop randomly falls into the surface. This should be a poisson process, and I could be wrong but I feel like that's not what's being used here. Lastly, the distance a drop will roll should depend on how "dry" the path is that it travels, although that one seems pretty hard to simulate!