Point-e

Point-E is the official code and model release for the paper 'Point-E: A System for Generating 3D Point Clouds from Complex Prompts'. Main features and usage: 1. Image to point cloud (image2pointcloud): sample a point cloud conditioned on some example synthetic view images. 2. Text to point cloud (text2pointcloud): use a small, worse quality pure text-to-3D model to produce 3D point clouds directly from text descriptions. This model's capabilities are limited, but it does understand some simple categories and colors. 3. Point cloud to mesh (pointcloud2mesh): use an SDF regression model for producing meshes from point clouds. Evaluation and rendering: provides P-FID and P-IS evaluation scripts, as well as Blender rendering code. Resources: seed images and point clouds for paper banner images, and seed images used for COCO CLIP R-Precision evaluations are available for download. Core advantages: achieves 3D model synthesis via point cloud diffusion, supports both text and image modalities, and offers open-source code and pretrained models for research. The project is under the MIT license and completely free to use.