how do we determine the threshold between "low", "medium", and "high" frequencies? is there a standard number somewhere?
This looks really good in comparison to the other stuff...what are the advantages/disadvantages of using component breakdown methods versus sampling?
I'm assuming we're using Fourier transforms of a kind here to split the image into its components?
Is this only useful for storing the image (but not drawing the image)? If the frequency components are calculated from the samples, it doesn't really seem to help with the sampling problem.
I'm not entirely sure how this would be extended to 3d, since meshes tend to be sharp edges in 3d space - would you use polar coordinates instead?
When the signals are mixed, does Fourier analysis comes into play?