This is cool and all, but are there any uses for it?
Representations like this are also sometimes used for automatic shot boundary detection (vertical edges in this type of image = an edit to a different camera shot) in video content retrieval systems.