As far as moving away from the monolithic image goes, I'm curious how 
objects that *have* to be shared will be referenced, like Symbols.
Will objects like these be specially know to the image segmentation 
primitives?

-- Mike

mklein at alumni.caltech.edu