Virtual Computational Camera

In computational camera research, people design unconventional optics to sample the light field in a way that is radically different from the traditional camera and preserve new forms of visual information for various applications. When one has an idea of optics design, it is very often that an evaluation based on simulation has to be conducted first before building a real system. However, how to quickly simulate an unconventional camera is often a challenging problem even with professional optical design software (e.g. Zemax). First, these software is designed for optics researches and usually do not support light field, which is a central concept in computational camera research. Secondly, researchers have to spend plenty of time in designing detailed physical profiles of optical elements. In this project, we develop a framework of virtual computational camera for light field processing. With this framework, one can easily simulate the light field processing of a virtual computational camera. In the proposed framework, we formulate three categories of optics, including linear transform optics (lenses and prisms), dotproduction optics (photomasks), and convolution optics (diffusers and some phaseplates). We propose an efficient computation strategy to render light fields at a decent resolution.