Because they are computed, generative art systems necessarily explore a state space, but many of their state spaces exhibit wild variations in visual interestingness across their domain. This makes them unsatisfying to use in live performance, since they'll lapse into the boring parts and resetting the state to restore it is visually unpleasant.
In this talk I'll give an introduction to generative art with a focus on iterated fractal systems in particular, which have some absolutely fascinating portions of state space alongside very boring ones. I'll present Many Worlds, a system that visualizes state space traversals ahead of time remotely to allow the user to find a compelling state path in an interactive, iterative way