Trellis 3D
ExternalTrellis 3D is a cutting-edge AI platform from Microsoft that generates high-quality, editable 3D assets—including meshes, radiance fields, and 3D Gaussians—from text or image prompts using the innovative Structured LATent (SLAT) representation. Trained on approximately 500K objects with models scaling up to 4B parameters, it enables rapid prototyping and variant creation for games, VR, 3D printing, and visualization workflows. Ideal for designers, developers, and hobbyists seeking free, local-run tools that deliver significant time savings without subscription limits, it excels in mechanical and simple assets while supporting production optimization.
Description
Trellis 3D is a cutting-edge AI platform from Microsoft that generates high-quality, editable 3D assets—including meshes, radiance fields, and 3D Gaussians—from text or image prompts using the innovative Structured LATent (SLAT) representation. Trained on approximately 500K objects with models scaling up to 4B parameters, it enables rapid prototyping and variant creation for games, VR, 3D printing, and visualization workflows. Ideal for designers, developers, and hobbyists seeking free, local-run tools that deliver significant time savings without subscription limits, it excels in mechanical and simple assets while supporting production optimization.
Key capabilities
- Generates editable 3D assets (meshes, radiance fields, 3D Gaussians) from text or image prompts using SLAT representation
- Supports text-to-3D and image-to-3D generation
- Outputs in multiple formats: GLB, OBJ, STL, GLTF, USDZ, PLY
- Open-source under MIT license for local editing and workflows
Core use cases
- 1.Rapid prototyping of mechanical and simple 3D assets
- 2.Creating static assets for VR and games
- 3.Generating prototypes for 3D printing
- 4.Variant creation and local refinement for visualization
Is Trellis 3D Right for You?
Best for
- Prototypers and designers working with mechanical/simple assets
- VR/game developers for background/static assets
- Hobbyists generating 3D printing prototypes
Not ideal for
- Animators and riggers due to topology issues
- High-end production studios needing minimal refinement
- Users with organic or complex photo inputs
Standout features
- Ultra-high quality with character consistency and multi-image referencing
- Large pre-trained models (up to 4B parameters) on ~500K objects
- Easy local editing and production workflow optimization
- Free trial and open-source installation on Linux with NVIDIA GPUs
Pricing
Base
Ultra
Pro
Free
Max
User Feedback Highlights
Most Praised
- Excels at accurate 3D models from mechanical/machinery images
- Up to 75% reduction in prototyping time
- High detail capture and crisp meshes for simple assets
- Free local runs outperform some paid alternatives
- Useful for VR assets and 3D printing after minor fixes
Common Complaints
- Poor mesh topology with triangulation, holes, and artifacts
- Inefficient polygon distribution wastes resources
- Limited performance on organic subjects and complex images
- Texture loss and suboptimal UV mapping at high resolutions
- Installation and workflow bugs, especially in integrations like ComfyUI