TRELLIS Structured-3D-Latents Pipeline

Properties

family

3d

modality

text/image -> 3D (mesh, radiance field, 3DGS)

whyMultiModel

A single representation cannot serve every downstream use; the structured-latent decoder lets one model emit interchangeable mesh, voxel, and 3DGS outputs that each need different post-processing.

steps

[object Object], [object Object], [object Object]

controls

Input image or text; output format (mesh/3DGS/volume); guidance strength; seed.

exampleStack

TRELLIS (or TRELLIS.2) -> texture refine -> retopo.

useCases

[Game and VFX assets][AR product previews][Rapid concept modeling]

pitfalls

Thin structures and interiors are under-represented; topology needs retopology before production use.

source

TRELLIS: Structured 3D Latents for Scalable and Versatile 3D Generation (arXiv:2412.01506)