Using random video input for music generation
I was recently reading about the special effects for the upcoming movie "The Fountain", which were done using microphotography of particles in fluid.
Apparently the result was incredibly dense, rich visuals of starfields & such, with a texture difficult to match by CG.
I was thinking of how to process such imagery, and use it to generate a matching soundscape. Perhaps by taking live video of the tank, and running it through a computer vision application that would generate sound?