Math, not artificial intelligence, powers new infinite world generator

Princeton researchers have unveiled Infinigen, a open-source tool that generates limitless photorealistic 3D scenes using pure mathematical rules – no AI or reliance on models trained on existing videos or images.

Unlike traditional 3D generators that rely on existing art assets, Infinigen builds everything procedurally from scratch. "Every shape, texture, and material is created through randomized mathematical formulas," reports the Princeton Vision & Learning Lab team in their CVPR 2023 paper. The system can generate diverse natural elements like plants, animals, terrain, and even complex phenomena such as fire, clouds, rain, and snow – all with true geometric detail rather than visual tricks.

From the paper, "Infinite Photorealistic Worlds using Procedural Generation":

We introduce Infinigen, a procedural generator of photorealistic 3D scenes of the natural world. Infinigen is entirely procedural: every asset, from shape to texture, is generated from scratch via randomized mathematical rules, using no external source and allowing infinite variation and composition. Infinigen offers broad coverage of objects and scenes in the natural world including plants, animals, terrains, and natural phenomena such as fire, cloud, rain, and snow. Infinigen can be used to generate unlimited, diverse training data for a wide range of computer vision tasks including object detection, semantic segmentation, optical flow, and 3D reconstruction. We expect Infinigen to be a useful resource for computer vision research and beyond.

The platform has already expanded beyond its nature-focused roots. A new CVPR 2024 paper introduces "Infinigen Indoors," bringing the same procedural generation approach to interior environments. As project lead Jia Deng and team note, while the initial focus was on natural scenes "because mammalian vision evolved in the natural world," Infinigen is actively growing to cover built environments and artificial objects.