Extracting 3-dimensional structure from real-world imagery and rendering it from unrestricted viewpoints is an important problem in computer vision, and increasingly, computer graphics. Despite many years of research, a system that automatically recovers
