3D / Point Cloud
Objaverse
800K+ annotated 3D models for training 3D and multimodal models.
100K–1M·CC-BY·Point Cloud (LAS/PLY)
WorldRenderer Dataset Test Dataset Summary WorldRenderer Dataset Test is a synthetic multi-scene 3D rendering dataset designed for research in: Novel View Synthesis (NVS) Neural Rendering Geometry-aware Generation Multi-view Representation Learning World Models 3D-conditioned Generative Modeling The dataset contains 27 textured 3D scenes.Each scene is rendered using a predefined monocular camera trajectory consisting of 401 frames. For every frame, aligned multi-modal… See the full description on the dataset page:
Source: Hugging Face Hub (Tengpaz/worldrenderer-dataset-test). Metadata imported from the dataset’s Hub tags.