TRELLIS.2: Microsoft Releases 3D Generative Model with O-Voxel Structure

Repositorymicrosoft/TRELLIS.2Tool

TRELLIS.2 is an open-source 4B-parameter 3D generative model by Microsoft designed for high-fidelity image-to-3D generation. Using a field-free sparse voxel structure named O-Voxel and Diffusion Transformers, it handles arbitrary topologies, sharp geometry, and complete PBR materials.

Key points

  1. Image-to-3D generation powered by a 4-billion-parameter model using vanilla Diffusion Transformers (DiTs).
  2. Novel O-Voxel representation handles open surfaces, non-manifold geometry, and internal enclosed structures.
  3. Full PBR material modeling, capturing base colour, roughness, metallic values, and opacity.
  4. High generation speed: 512³ resolution in around 3 seconds on an NVIDIA H100 GPU.
  5. Fully optimization- and rendering-free conversions: O-Voxel converts to a textured mesh in under 100 ms via CUDA.
  6. Prerequisites: Linux-only environment with an NVIDIA GPU holding at least 24 GB of VRAM (tested on A100 and H100).

Why it matters

Traditional 3D generative models struggle with open surfaces like fabric or foliage and rely on computationally expensive post-processing fields. TRELLIS.2 solves these constraints by combining compact latent representations with direct mesh conversion, accelerating asset pipelines for 3D graphics and games.

Analysis

TRELLIS.2 exhibits strong technical maturity for an open research release. Microsoft provides pretrained checkpoints, training and inference code under the MIT license, an arXiv paper, and a Hugging Face Space. However, the hardware barrier—requiring a Linux setup and a GPU with at least 24 GB of VRAM—restricts local execution mainly to research teams and production environments with data-centre hardware.

Mentioned links

Found in

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:Superman

    Microsoft TRELLIS.2: 4B Image-to-3D Generative Model Using O-Voxels

    Microsoft has released the repository for TRELLIS.2, a 4-billion parameter generative model designed for high-fidelity image-to-3D synthesis. It introduces a field-free sparse voxel structure called O-Voxel to generate complex 3D meshes with full PBR materials.

    292bookmarks12.3kviews

    AI & LLMs· Tool

  • Repository:lightningpixel/modly

    Modly: Local Desktop App for AI-Powered 3D Mesh Generation

    Modly is an open-source desktop application for Windows, Linux, and Apple Silicon macOS that creates 3D models from images or text prompts directly on local GPUs.

    7.4kstarsTypeScript

    AI & LLMs· Tool

  • Repository:img2threejs/img2threejs

    img2threejs: Procedural Code-Only Image-to-3D for Three.js

    img2threejs turns reference images into procedural, animation-ready Three.js models generated purely as code, avoiding traditional mesh extraction or photogrammetry.

    15.1kstarsPython

    AI & LLMs· Tool

  • Repository:ashhart/TensorFold

    TensorFold: LLM Inference Engine for Apple Silicon and Modern NVIDIA GPUs

    TensorFold is an open-source inference engine designed to serve large language models on Apple Silicon (via MLX/Metal) and newer NVIDIA GPUs (Compute Capability 8.9+). It provides OpenAI- and Anthropic-compatible endpoints with model-tailored kernels and speculative draft verification.

    1.1kstarsPython

    AI & LLMs· Tool

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.