Nonlinear depth scaling for immersive video applications

In this paper we discuss the minimum required accuracy of depth or disparity information for a typical immersive video application. Our approach benefits from the fact that in many cases the navigation range of the virtual camera is limited. We develop a simple geometric problem formulation and compare it with the more general plenoptic sampling approach. In this way we bridge the gap between classical 3D warp applications and the more general plenoptic sampling theorem. From this we derive general rules for the design of a typical immersive video application scenario. We provide a solution for the scalability of depth or disparity map designs in dependency of application scenarios.