2 million executable 3D programs with renders and descriptions
Huge corpus of executable 3D code (Blender, Three.js, CAD) with renders and instructions for 3D generation.
- Text → 3d
- Image → 3d
- Text generation
- English
Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.
Answered by Gemini 4 · the chat is kept only in this tab
Huge corpus of executable 3D code (Blender, Three.js, CAD) with renders and instructions for 3D generation.
Unified 3D dataset and benchmark for predicting spatial occupancy in autonomous driving tasks.
10,000 to 100,000 examples featuring text, images, and 3D code for evaluating and training 3D generation models.
Multimodal corpus of parametric 3D models from Objaverse featuring sand3d code, OpenUSD scenes, and images.
A set of 48 views for each of 83,296 3D objects for training 3D generation models.
10,000 realistic indoor 3D scenes with 565K objects for training and evaluating embodied AI.
24,495 3D objects and mechanisms with sand3d code descriptions, renders, and OpenUSD.
Dataset for image-to-3D or text-to-3D generation tasks from the UnCommon Objects in 3D research.
Millions of English-language examples for generating and reconstructing 3D objects from images.
Synthetic images and texts for point tracking, 3D reconstruction, and robotics tasks.
106 hours of free-camera flights across 17 games for 3D reconstruction and video generation.
10,000 hours of real human work video with head and hand pose annotations in video format for robotics.
9,209 everyday items, mechanisms, and CAD parts provided as sand3d code, OpenUSD (.usda) files, and renders.
Images and 3D data of animals with shape, texture, and fur maps for image-to-3D tasks.
Three 3D object collections with renders and points for text-to-3D and image-to-3D.
Large multi-domain image-text corpus for 4D world modeling and robotics.
Large dataset of synchronized multi-view egocentric videos for 3D scene synthesis and depth estimation.
A large dataset of aerial urban panoramic images and matching depth maps for 3D scene understanding.
Monocular videos of static and dynamic scenes with 4D annotations for 3D and 4D generation.
Preprocessed RE10K data for image-to-3D scene reconstruction models.
1M executable CAD programs synthesized by an agent without real data, for text-to-3D and image-to-3D.
A complete snapshot of the Mixamo library: 2,317 animations and 114 rigged 3D characters with a unified skeleton for motion generation.
3D meshes with physical properties, grasp points, and descriptions for robot training.
Open-source portion of the CADGenBench benchmark featuring English input data for evaluating STEP format CAD model generation.