Skip to content
AI Tools Finder

TRELLIS

Microsoft Research's structured 3D representation model.

Open Source 16–24 GB VRAMRuns locally
Actually FreeNo SignupOpen SourceWatermark-Free
Visit TRELLISUpdated 2026-04-02 · Direct link

Hardware requirements

Runs locally · High-end GPU (16–24 GB)

16–24 GB VRAM
Min VRAM
16 GB
Rec. VRAM
24 GB
Min RAM
32 GB
Rec. RAM
64 GB
Disk
30 GB
GPU class
High-end GPU
12.1+No Apple SiliconGPU RequiredQuant: FP16

Image-to-3D fits 16 GB; text-to-3D wants 24 GB.

Screenshot placeholder · TRELLIS

What is TRELLIS?

TRELLIS uses a novel 'structured latent' representation — joint sparse-voxel + feature-grid — to generate high-quality 3D assets (mesh + Gaussian splat + radiance field) from images or text. MIT license, image-to-3D and text-to-3D pipelines.

Pros & cons

Pros

  • Outputs mesh + Gaussian splat + radiance field in one shot
  • MIT licensed
  • Microsoft Research backing — solid engineering

Cons

  • Heavier than Hunyuan3D-2 at equivalent quality
  • Texture quality below dedicated PBR pipelines

What's actually free?

MIT.

✓ Actually FreeNo SignupOpen SourceWatermark-Free

Alternatives

Hunyuan3D-2

Tencent's open 3D generator — multi-view, PBR, ready-to-use meshes.

Open Source 12–16 GB VRAM
Min VRAM
12 GB
GPU class
High-end GPU
Quant
FP16
Actually FreeOpen SourceWatermark-FreeAPI

TripoSR

Single-image to 3D mesh in under a second on a 4090.

Open Source 6–8 GB VRAM
Min VRAM
6 GB
GPU class
Entry GPU
Quant
FP16
Actually FreeNo SignupOpen SourceWatermark-Free

Stable Zero123

Novel-view synthesis — generate any angle from a single image.

Open Source 6–8 GB VRAM
Min VRAM
6 GB
GPU class
Entry GPU
Quant
FP16
Actually FreeOpen SourceWatermark-FreeHobbyist-Friendly