Tip Of AI
Discover what you can do with AI
GET3D is a generative model that is capable of generating high-quality 3D textured shapes. It is designed to meet the growing demand for content creation tools that can scale in terms of quantity, quality, and diversity of 3D content for use in virtual worlds and other applications. The model directly generates explicit textured 3D meshes with complex topology, rich geometric details, and high-fidelity textures. One of the key features of GET3D is its ability to generate diverse shapes with arbitrary topology. It is capable of generating 3D meshes for various objects, including cars, chairs, animals, motorbikes, human characters, and buildings. The generated shapes exhibit high quality in terms of geometry and texture. GET3D utilizes a rasterization-based differentiable renderer to obtain RGB images and silhouettes. It uses two 2D discriminators for classifying whether the inputs are real or fake. The model is trained with adversarial losses defined on 2D images. In addition to generating high-quality 3D meshes, GET3D also demonstrates other capabilities. It achieves a good disentanglement between geometry and texture, allowing for meaningful interpolation between different shapes. It also supports unsupervised material generation, text-guided shape generation, and can generate novel shapes by perturbing the latent code. The paper provides qualitative results and highlights the diversity and quality of the generated 3D meshes. It also includes information on the authors, citation details, and references to related works.
