Decide what you see, and how you see it, with the flick of a switch.

Abstract: We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, ...