Hunyuan3D ComfyUI Workflow

Creating stunning 3D assets has never been easier, thanks to the integration of ComfyUI and Hunyuan3D. Whether you’re new to 3D modeling or an experienced artist looking for streamlined solutions, this guide will help you get started with generating textured models and animations in no time. Let’s embark on a journey into the world of effortless 3D asset creation!

Hunyuan3D is a cutting-edge 3D synthesis system capable of generating high-resolution textured models. It has two main components:

1. Hunyuan3D-DiT: A shape-generation model that creates detailed meshes based on input conditions.
2. Hunyuan3D-Paint: A texture-synthesis model that enhances the meshes with vibrant and high-quality textures.

Additionally, Hunyuan3D-Studio simplifies tweaking and animating these creations, making professional-grade results accessible to everyone. The system’s 2GP mode is optimized to work on GPUs with as little as 6GB of VRAM, ensuring inclusivity for users with modest hardware.

System Requirements

To ensure smooth operation, your system should meet these requirements:

• RAM: Minimum 24GB (dependent on profile).
• VRAM: At least 6GB (lower profiles accommodate this).
• Python: Version 3.10 or higher.
• Torch: Version 2.5.1 with CUDA support.

Installation Steps

Model Files

• Original Model (.ckpt): Hunyuan3D Original Model
• Converted Model (.safetensors): Hunyuan3D Safetensors

Place the models into:

ComfyUI/models/diffusion_models/

First Install the ComfyUI-Hunyuan3DWrapper in ComfyUi Custom Nodes Folder

git clone https://github.com/kijai/ComfyUI-Hunyuan3DWrapper.git

Dependencies Installation

With a Python Environment
Run the following command to install dependencies:

pip install -r requirements.txt

Using Portable Python

python_embeded\python.exe -m pip install -r ComfyUI\custom_nodes\ComfyUI-Hunyuan3DWrapper\requirements.txt

Texture Generation (Compilation Required)

The texture generation module requires compilation. Precompiled components have been included, compatible with:

• Windows 11
• Python 3.12
• CUDA 12.6 (compatible with Torch built for CUDA 12.4)

Installing the Precompiled Rasterizer

Run one of the following:

custom_rasterizer-0.1-cp312-cp312-win_amd64.whl

Standard Environment

pip install wheels\custom_rasterizer-0.1-cp312-cp312-win_amd64.whl

Portable Installation

python_embeded\python.exe -m pip install ComfyUI\custom_nodes\ComfyUI-Hunyuan3DWrapper\wheels\custom_rasterizer-0.1-cp312-cp312-win_amd64.whl

Manual Compilation

If the precompiled files are incompatible with your environment, compile the necessary components manually:

Navigate to the rasterizer directory and install:

cd hy3dgen/texgen/custom_rasterizer
python setup.py install

Navigate to the differentiable renderer directory and build in place:

cd hy3dgen/texgen/differentiable_renderer
python setup.py build_ext --inplace