Virtual reality, extended Depth-of-Field

We propose an end-to-end joint optimization method to learn a DOE placed in front of a projector lens and a compensation network for deblurring.