princeton-vl/infinigen

Health v2 · maintenance only

Adoption not part of the score

7222 stars · 607 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Infinigen is a procedural generator of infinite photorealistic 3D worlds and scenes, built on Blender by the Princeton Vision & Learning Lab. It produces diverse high-quality 3D training data (RGB, depth, segmentation, surface normals) for computer vision research, including indoor scenes, nature scenes, and articulated simulation assets.

Use cases

• generate synthetic training data for computer vision models
• create photorealistic 3D indoor scenes procedurally
• generate infinite nature landscapes with ground truth annotations
• export articulated assets to simulators
• render depth maps and instance segmentation from generated scenes
• create diverse 3D datasets for machine learning research

When to choose

• you need large-scale synthetic 3D training data with perfect ground truth labels
• you want procedurally generated photorealistic indoor or outdoor scenes
• you need simulation-ready assets for robotics or RL environments
• you are doing computer vision research requiring diverse rendered scenes

When to avoid

• you need a lightweight real-time 3D engine for games
• you want hand-crafted art assets rather than procedural generation
• you lack GPU/compute resources for Blender-based rendering
• you need non-photorealistic stylized content

Facets

library · maturity active

simulation graphics image-processing machine-learning data-generation computer-vision graphics simulation machine-learning artificial-intelligence python cross-platform procedural-generation blender synthetic-data 3d-scenes photorealistic training-data cvpr linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="princeton-vl/infinigen")