I suppose some very very very efficient opengl could do a few of those pieces in realtime, but unfortunately i'm not very proficient at opengl. most of the pieces, due to complexity (sheer numbers of things moving while looking at what all the other objects are doing at the same time), took anywhere from 18hrs to several weeks to simulate.

Still, having each iteration be similar but variable would be the ultimate. Even if i could run it on macminis, the cost difference between 14 macminis and 14 dvd players is considerable...

In an ideal world, Brandon shouldn't have to make the choice between DVD loops and a stack of CPUs - the local art scene *should be* mature enough for him to get real support from grant-givers AND an engaged art-purchasing community. Alas...