I'm working in this field, and I would love to hear how he's generating video from these networks that don't understand what they're making. It can barely draw the same person twice, much less have the camera move through a 3d space.