DaVinciCodice

A piece of History

We use an image-to-image translation network to encode motion priors of natural scenes collected from online videos so that for a new photo, we can synthesize...