It appears that the placement of the perspective points must be customized to align with the deeper elements (I.E. the alleyway) to create multiple vanishing points. This image cleverly reveals only one such VP. Would this system work as well in a consistently deep space like a bowling alley? What happens if the facades are not parallel, as in the acropolis?

There is a program called “stitcher” that uses a similar interpolative system for single panoramic captures. I would think you could combine the two systems for some interesting results. Hmmm.