It seems to me like this allows some incredible augmented reality once a computer vision system can perform do the user-provided analysis of the scene's geometry and light sources.