GET3D
NVIDIA's 2022 generative model that outputs explicit textured meshes ready for game engines.
About
GET3D learns to generate textured 3D meshes directly from 2D image collections, with no 3D ground truth required. It pairs a geometry generator based on deep marching tetrahedra with a texture generator, producing meshes with complex topology and high-fidelity textures that drop straight into renderers like Blender, Unity, or Unreal. It was one of the first models to make GAN-generated 3D assets genuinely usable downstream.
Presented at NeurIPS 2022 by NVIDIA Research, with code and pretrained category models for cars, chairs, animals, motorbikes, characters, and buildings. The catch is the NVIDIA Source Code License, which is not an open-source license and restricts commercial use. Diffusion-era models have since passed its quality, but it remains a well-documented baseline for GAN-based 3D generation.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
FID/COV/MMD vs prior 3D GANs in the paper (author-reported); no independent evals
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: NeurIPS 2022 peer-reviewed paper; official NVIDIA repo with code, weights, and eval scripts. Strengths: landmark explicit-mesh generation, engine-ready outputs, thorough documentation. Docked for: (1) NVIDIA Source Code License is not OSI-approved and restricts commercial use; (2) the 2022 GAN approach has been superseded in quality by diffusion and feed-forward methods. License: NVIDIA Source Code License.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · NVIDIA Research; NeurIPS 2022; explicit textured meshes from 2D image collections
- · Pretrained category models (cars, chairs, animals, characters, buildings); engine-ready outputs
- · No independent evals; GAN approach superseded in quality by diffusion methods
- · Repo license confirmed via README