The project was created from AI-generated Depth Maps and animated in AE.
Here's one more animated environment created by Nude Robot using the same techniques:
And if you are interested in Nude Robot's method of turning generated 2D images into Depth Maps, we also recommend checking out this presentation on the Consistent Video Depth Estimation algorithm, shared by a team of researchers from Facebook, the University of Washington, and Virginia Tech during the SIGGRAPH 2020 conference.
The presentation describes a method that can reconstruct dense, geometrically consistent depth for all pixels in a monocular video, leveraging a conventional structure-from-motion reconstruction to establish geometric constraints on pixels in the video. This neural network was trained for single-image depth and is able to handle hand-held captured input videos with dynamic motion.
Click here to check out more of the artist's projects.