Single channel video 2022


VQGAN hallucinating a conversation between circles.

VQGAN is a neural network architecture used to generate new images from a set of existing images. It can then sequence these new images to produce a video from the original images. I forced the neural network to give me a video from exactly one image.