README.md

ComfyUI [with HunyuanImage2.1 support]

The most powerful and modular visual AI engine and application.

This repository is a fork of ComfyUI, which properly implements HunyuanImage2.1.

ComfyUI lets you design and execute advanced stable diffusion pipelines using a graph/nodes/flowchart based interface. Available on Windows, Linux, and macOS.

Get Started with HunyuanImage2.1

Clone this repository and installing dependencies:

git clone git@github.com:KimbingNg/ComfyUI-HunyuanImage2.1.git
pip3 install -r requirements.txt

Run ComfyUI:

python3 main.py --listen --port 443

Basic Workflow

Download the following models and place them in the appropriate ComfyUI directories:

Text Encoders

Download and put in your ComfyUI/models/text_encoders directory:

• byt5_small_glyphxl_fp16.safetensors
• qwen_2.5_vl_7b.safetensors

VAE Models

Download and put in your ComfyUI/models/vae directory:

• hunyuan_image_2.1_vae_fp16.safetensors
• Optional (for refiner): hunyuan_image_refiner_vae_fp16.safetensors

Diffusion Models

Download and put in your ComfyUI/models/diffusion_models directory:

• diffusion models

Workflow

You can then load up or drag the following workflow in ComfyUI:

• Non-distill model workflow (50-100 steps inference): hunyuan_image_2.1_workflow.json

• Distill model workflow (faster speed with 8-step inference but slightly lower quality): hunyuan_image_2.1_distill_workflow.json