GET3D is a generative model that synthesizes high-quality 3D textured shapes directly from images. It can generate diverse shapes with arbitrary topology, high-quality geometry, and texture. The model is trained using 2D image collections and utilizes differentiable surface modeling, differentiable rendering, and 2D Generative Adversarial Networks. The generated 3D meshes can be used in various downstream applications in industries such as gaming, virtual reality, and architecture.
Features
- Generative model for 3D textured shape synthesis
- Direct synthesis from 2D images
- Supports complex topology, rich geometric details, and high fidelity textures
Use Cases
- Content creation for gaming and virtual reality
- 3D modeling for animation and film production
- Architectural visualization and design
Suited For
- Artists
- Game developers
- Architects
FAQ
GET3D generates 3D shapes using two latent codes: one for geometry and one for texture. It extracts a 3D surface mesh from a 3D Signed Distance Function (SDF) and queries a texture field to get colors for the mesh.
GET3D can generate a wide range of shapes including cars, chairs, animals, motorbikes, human characters, and buildings. It is able to produce diverse shapes with arbitrary topology and high-quality geometry and texture.