We present LookingGlass, an interactive experience that gives a generative twist to classical anamorphoses—illusions that reveal their true form only from a specific viewpoint or through a mirror or lens. While these distortions date back to the 17th century, traditional anamorphic images lose meaning when viewed directly. Our method uses latent diffusion models to create ambiguous anamorphoses—images that remain interpretable in both distorted and undistorted forms. Central to this is Laplacian Pyramid Warping, a frequency-aware image warping technique for high-quality synthesis. Differing from previous works, we extend generative perceptual illusions to latent space models with broader spatial transformations while maintaining image quality. LookingGlass invites Laval Virtual attendees to explore these illusions firsthand, blending geometry, optics, and AI creativity.