SKIT_ComfyUI-ZImage-Generate

ComfyUI Z-Image Generate

A single ComfyUI custom node for generating images with Z-Image through Hugging Face Diffusers.

Node

• Z-Image Generate

Inputs:

• prompt: text prompt
• negative_prompt: optional negative prompt
• model_id: Hugging Face model id, default Tongyi-MAI/Z-Image
• width, height: output image size
• steps: diffusion steps
• guidance_scale: prompt guidance strength
• seed: deterministic seed
• dtype: auto, bfloat16, float16, or float32
• clear_cache: unload the cached pipeline after generation

Output:

• IMAGE

Install

Clone this repository into your ComfyUI custom nodes folder:

cd ComfyUI/custom_nodes
git clone https://github.com/YOUR_USERNAME/ComfyUI-ZImage-Generate.git

Install dependencies in the same Python environment that runs ComfyUI:

pip install -r ComfyUI-ZImage-Generate/requirements.txt

Then restart ComfyUI.

Model Download

On the first run, Diffusers downloads the model from Hugging Face. Large models can take time and disk space.

The default model id is:

Tongyi-MAI/Z-Image

You can also try another compatible Z-Image model id in the node's model_id field.

Notes

• CUDA is used automatically when available.
• auto dtype uses bfloat16 on CUDA and float32 on CPU.
• This node keeps one pipeline cached for faster repeated generations.
• Enable clear_cache if you want to release VRAM after a generation.

Repository Structure

ComfyUI-ZImage-Generate/
|-- __init__.py
|-- nodes.py
|-- requirements.txt
|-- README.md
`-- .gitignore