[Daniel Geng] and others have an interesting system of generating multi-view optical illusions, or visual anagrams ... diffusion models driven by modern machine learning techniques.