Apple’s new AI model recreates 3D objects with realistic lighting effects from a single image

Apple researchers developed LiTo, an AI model that reconstructs detailed 3D objects from a single photograph while preserving realistic lighting effects like reflections and highlights from multiple viewing angles. The system compresses both geometry and light interaction data into a mathematical "latent space," then uses a decoder to recreate the full 3D object with accurate visual effects. Unlike previous methods requiring multiple images from different angles, LiTo achieves reconstruction from one image by learning from thousands of objects rendered under various lighting conditions and viewing angles.