Infinigen is a cutting-edge procedural generator of 3D scenes created by Princeton Vision & Learning Lab, optimized for computer vision research. It generates high-quality training data based on Blender and is free and open-source. What sets Infinigen apart is its ability to create diverse objects and scenes in the natural world with real geometric detail, avoiding the use of bump/normal-maps and full-transparency to ensure accurate 3D ground truth. Infinigen also offers automatic high-quality annotations for various computer vision tasks. Their research papers have been presented at prestigious conferences, with plans to expand into generating indoor scenes in the future.
https://infinigen.org/