(no subject)
I was trying to figure out how you could generate a three-dimensional model of some object. The best I could come up with is: you put your object in the center of a room with white or blue walls (or anything else that you expect to contrast intensely with the subject.) You have a video camera mounted on a circular track, which rotates around and up while aiming at the subject, tracing a sort of helical corkscrew through the air. You then analyze the footage for the silhouette of the object, and then working backwards it should be relatively straightforward to to combine the silhouettes (perhaps you take a page from sculpting, and start with the whole room being filled by a virtual cube, and then slice away swaths of it in accordance with the outlines) to get a working model of the subject. It should also be pretty easy to map the actual images you get onto the appropriate surfaces of the object, resulting in a highly accurate colored model.
The big downside of this, of course, is that it would only work fully on convex objects; this method won't pick up anything like, say, a nostril. Indentations are tricky. Still, for most things, you could at least get a workable approximation of a model.
The big downside of this, of course, is that it would only work fully on convex objects; this method won't pick up anything like, say, a nostril. Indentations are tricky. Still, for most things, you could at least get a workable approximation of a model.