Logo of Trellis2
Trellis 2
FeaturesPricing
BlogFAQ
Logo of Trellis2
Trellis 2

Transform images into stunning 3D models with AI-powered technology

Product

  • Features
  • Pricing
  • Blog
  • FAQ

Company

  • About
  • Contact

Legal

  • Privacy Policy
  • Terms of Service

© 2026 Trellis 2. All rights reserved.

TRELLIS 2 ComfyUI: Complete Installation & Workflow Guide (2026)
2026/04/21
7 min read

TRELLIS 2 ComfyUI: Complete Installation & Workflow Guide (2026)

Step-by-step guide to installing and using Microsoft TRELLIS 2 in ComfyUI. Covers custom node setup, workflow configuration, low VRAM mode, GLB export, and troubleshooting common issues.

Last updated: April 21, 2026

Microsoft TRELLIS 2 is a 4-billion parameter open-source model that generates high-fidelity 3D assets from a single image. Running it inside ComfyUI gives you a visual node-based pipeline where you can chain image generation, 3D reconstruction, texture baking, and format export into one automated workflow.

This guide covers everything you need — from installing the custom nodes to building your first ComfyUI workflow — whether you're on Windows or Linux, with an RTX 3060 or an H100.

If you'd rather skip the setup entirely, you can use TRELLIS 2 online with no installation required.

Why Use TRELLIS 2 in ComfyUI?

ComfyUI integration unlocks several advantages over the standalone CLI:

FeatureStandalone CLIComfyUI
Visual workflowCommand-line onlyNode-based drag & drop
Batch processingScript requiredAutomatic with Queue
Pipeline chainingManualConnect nodes visually
Community workflowsN/AImport JSON workflows
Live previewNoYes
Integration with SD/XLSeparate processSame pipeline

Prerequisites

Hardware Requirements

ComponentMinimumRecommended
GPUNVIDIA 8GB VRAM (RTX 3060)NVIDIA 16GB+ VRAM (RTX 4070+)
RAM16 GB32 GB
Storage15 GB free (SSD)30 GB free (NVMe SSD)
OSWindows 10/11 or LinuxWindows 11 or Ubuntu 22.04+

Software Requirements

  • Python 3.10 or 3.11
  • PyTorch 2.1+ with CUDA 12.1+
  • ComfyUI (latest version)
  • Git for cloning repositories
  • NVIDIA Driver 525+

Installation

There are two main ComfyUI custom nodes for TRELLIS 2. This guide covers both options.

Option A: PozzettiAndrea/ComfyUI-TRELLIS (Recommended)

This is the most actively maintained node with regular updates.

Step 1: Install ComfyUI (if you haven't)

git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI
pip install -r requirements.txt

Step 2: Install the TRELLIS 2 custom node

cd ComfyUI/custom_nodes/
git clone https://github.com/PozzettiAndrea/ComfyUI-TRELLIS.git
cd ComfyUI-TRELLIS
pip install -r requirements.txt

Step 3: Download model weights

Download the TRELLIS 2 model weights from HuggingFace and place them in the correct directory:

# Create the models directory
mkdir -p ComfyUI/models/trellis/

# Download from HuggingFace (requires huggingface-cli)
pip install huggingface-hub
huggingface-cli download microsoft/TRELLIS.2-4B --local-dir ComfyUI/models/trellis/

Alternatively, download manually from HuggingFace TRELLIS.2-4B and extract to ComfyUI/models/trellis/.

Step 4: Verify installation

cd ComfyUI
python main.py

Open http://localhost:8188 in your browser. Search for "TRELLIS" in the node browser — you should see the TRELLIS nodes listed.

Option B: Official Microsoft Integration

Microsoft provides a reference integration in the TRELLIS.2 repository:

cd ComfyUI/custom_nodes/
git clone https://github.com/microsoft/TRELLIS.2.git
cd TRELLIS.2
pip install -r requirements.txt

Windows-Specific Notes

Windows users may encounter additional setup steps:

  1. Install Visual Studio Build Tools with "Desktop development with C++" selected
  2. If using a portable ComfyUI installation, use the embedded Python: 
    python_embeded\python.exe -m pip install torch==2.8.0 torchvision==0.23.0 torchaudio==2.8.0 --index-url https://download.pytorch.org/whl/cu121
  3. Some dependencies may require compiling from source — ensure cuda.dll is in your PATH

Troubleshooting Installation Issues

ProblemSolution
CUDA out of memory on startupEnable Low VRAM mode (see below)
ModuleNotFoundError: trellisReinstall requirements: pip install -r requirements.txt --force-reinstall
Node not appearing in browserCheck custom_nodes/ folder structure — ensure no nested directories
xformers import errorInstall matching version: pip install xformers==0.0.23 --index-url https://download.pytorch.org/whl/cu121
Windows path errorsMove ComfyUI to a path without spaces (e.g., C:\ComfyUI)

Basic Workflow: Image to 3D

Here's the simplest ComfyUI workflow to generate a 3D model from an image.

Nodes Required

  1. Load Image — Your input photo
  2. TRELLIS Image to 3D — Core generation node
  3. TRELLIS Export — Save as GLB/OBJ
  4. Save Image (optional) — Preview the result

Step-by-Step

  1. Add a Load Image node — Right-click canvas → Add Node → image → Load Image
  2. Add TRELLIS Image to 3D — Right-click → Add Node → TRELLIS → Image to 3D
  3. Add TRELLIS Export — Right-click → Add Node → TRELLIS → Export
  4. Connect the nodes:
    • Load Image → IMAGE output → TRELLIS Image to 3D → image input
    • TRELLIS Image to 3D → MESH output → TRELLIS Export → mesh input
  5. Configure the Export node:
    • Format: GLB (recommended for web/game engines)
    • Resolution: 512 for speed, 1024 for quality, 1536 for production
  6. Click Queue Prompt to run

Output Quality Settings

ResolutionTime (H100)Time (RTX 3060)Best For
512³~3s~30sQuick previews
1024³~17s~120sStandard quality
1536³~60s~300s+Production assets

Advanced Workflow: Text to 3D with Textures

For generating 3D models from text prompts with full PBR materials:

  1. CLIP Text Encode — Your text prompt (e.g., "a medieval sword with golden handle")
  2. TRELLIS Text to 3D — Generate from text description
  3. TRELLIS Texture Bake — Apply PBR materials (Base Color, Roughness, Metallic)
  4. TRELLIS Export — Export with embedded textures

Key Parameters

ParameterOptionsRecommendation
Shape Qualitydraft / standard / highstandard for most use cases
Texture Resolution512 / 1024 / 20481024 balances quality and speed
PBR Channelscolor / fullfull for game engines
Mesh Simplificationnone / low / medium / highlow for real-time applications
UV Unwrapauto / smartsmart for better texture packing

Low VRAM Mode

If you have 8GB VRAM or less, enable low VRAM mode to avoid out-of-memory errors:

Configuration

In the TRELLIS node settings:

  • Set Resolution to 512
  • Enable Low VRAM Mode checkbox
  • Set Batch Size to 1
  • Disable Texture Baking if not needed

Performance Reference (RTX 3060 8GB)

SettingVRAM UsageTime per Model
512³ Low VRAM~6.5 GB~60s
512³ Normal~7.5 GB~30s
1024³ Low VRAM~7.8 GB~180s

Tip: Close other GPU-consuming applications (browser tabs with WebGL, other AI tools) before running TRELLIS 2 on low VRAM GPUs.

Export Formats

TRELLIS 2 supports multiple export formats through ComfyUI:

FormatExtensionBest ForTexture Support
GLB.glbWeb, games, AR/VRFull PBR
OBJ.obj3D printing, BlenderBasic color
FBX.fbxUnity, Unreal EngineFull PBR
STL.stl3D printingNone
3D Gaussians.plyNeRF renderingColor only

For game development, GLB is recommended — it's compact, widely supported, and includes PBR textures in a single file.

Recommended Community Workflows

The ComfyUI community has shared several production-ready workflows:

  • RunComfy Trellis 2 Advanced Workflow — Image to structured voxel mesh with UV unwrapping and GLB export
  • Patreon One-Click Workflow — Auto-downloads models and configures dependencies (Windows only)
  • Multi-View Pipeline — Combines multiple camera angles for higher quality outputs

To import a workflow JSON in ComfyUI:

  1. Drag the .json file onto the ComfyUI canvas
  2. Or go to Load → select the workflow file
  3. Adjust input images and parameters
  4. Click Queue Prompt

Common Issues and Solutions

"TRELLIS node produces empty output"

This usually means the model weights are not in the correct directory. Verify:

ComfyUI/models/trellis/  should contain the model files from HuggingFace

"Generation is extremely slow"

  • Check GPU utilization with nvidia-smi — if it's low, the model may be running on CPU
  • Ensure CUDA is properly installed: python -c "import torch; print(torch.cuda.is_available())"
  • Close other GPU-heavy applications

"Output mesh has artifacts"

  • Use a higher-resolution input image (512x512+)
  • Ensure the subject is centered with a clean background
  • Try increasing the Shape Quality parameter
  • For complex topologies (open surfaces, thin structures), use 1024³ or higher

What's Next?

  • TRELLIS 2 Download Guide — Direct links to all model weights and dependencies
  • ComfyUI Workflow Tutorial — Detailed step-by-step workflow with downloadable JSON
  • How to Install TRELLIS 2 — Standalone installation guide
  • How to Use TRELLIS 2 — General usage guide

FAQ

Can I run TRELLIS 2 in ComfyUI on AMD GPUs?

Not officially. TRELLIS 2 requires CUDA. AMD users can try ROCm-based PyTorch, but expect compatibility issues. For AMD GPUs, the online version is recommended.

How much VRAM do I need?

Minimum 8GB (with low VRAM mode at 512³). Recommended 16GB+ for 1024³ and higher resolutions.

Can I use TRELLIS 2 with other ComfyUI nodes?

Yes. You can chain TRELLIS 2 with Stable Diffusion XL, ControlNet, or any image generation node. Generate an image first, then feed it into TRELLIS for 3D conversion.

Is TRELLIS 2 free to use?

Yes. TRELLIS 2 is open-source under Microsoft's license. You can use it locally at no cost. The HuggingFace model weights are also free to download.

All Posts

Author

avatar for Trellis2 Team
Trellis2 Team

3D technology specialists focused on AI-powered 3D model generation, format conversion, and browser-based 3D rendering. We test and review 3D tools so you don't have to.

Categories

    Why Use TRELLIS 2 in ComfyUI?PrerequisitesHardware RequirementsSoftware RequirementsInstallationOption A: PozzettiAndrea/ComfyUI-TRELLIS (Recommended)Option B: Official Microsoft IntegrationWindows-Specific NotesTroubleshooting Installation IssuesBasic Workflow: Image to 3DNodes RequiredStep-by-StepOutput Quality SettingsAdvanced Workflow: Text to 3D with TexturesKey ParametersLow VRAM ModeConfigurationPerformance Reference (RTX 3060 8GB)Export FormatsRecommended Community WorkflowsCommon Issues and Solutions"TRELLIS node produces empty output""Generation is extremely slow""Output mesh has artifacts"What's Next?FAQCan I run TRELLIS 2 in ComfyUI on AMD GPUs?How much VRAM do I need?Can I use TRELLIS 2 with other ComfyUI nodes?Is TRELLIS 2 free to use?

    More Posts

    How to Use TRELLIS 2 Online Free: No GPU Required (2026)

    How to Use TRELLIS 2 Online Free: No GPU Required (2026)

    Complete guide to using Microsoft TRELLIS 2 online for free. Covers Hugging Face Spaces, Google Colab, and other cloud platforms — no GPU or Python installation needed.

    avatar for Trellis2 Team
    Trellis2 Team
    2026/05/04
    Image to 3D Model for 3D Printing: Complete Workflow Guide (2026)

    Image to 3D Model for 3D Printing: Complete Workflow Guide (2026)

    Turn any photo into a 3D-printable model. Covers AI tool selection, mesh repair in Blender, STL export settings, slicer configuration, and troubleshooting common print failures. Works with TRELLIS 2, Meshy AI, and Tripo AI.

    avatar for Trellis2 Team
    Trellis2 Team
    2026/05/03
    TRELLIS 2 PBR Texture Generation in ComfyUI: Complete Guide (2026)

    TRELLIS 2 PBR Texture Generation in ComfyUI: Complete Guide (2026)

    How to generate production-quality PBR textures (Base Color, Roughness, Metallic, Normal maps) with TRELLIS 2 in ComfyUI. Covers texture baking nodes, parameter tuning, and export for Unity/Unreal.

    avatar for Trellis2 Team
    Trellis2 Team
    2026/04/21

    Newsletter

    Join the community

    Subscribe to our newsletter for the latest news and updates