One Node · Z-Image

One node, everything you need for Z-Image Turbo.
No node graph, no spaghetti — prompt, generate, done.

---

Installation

cd ComfyUI/custom_nodes/
git clone https://github.com/Adeliox/ComfyUI-OneNode-Z-Image.git

Restart ComfyUI. The node is in the menu under right click → Add Node → ZImageOneNode.

Required models

Download and place in ComfyUI/models/:

Model	Folder	Link
Z-Image Turbo UNET	models/diffusion_models/	HuggingFace
Qwen 3.4B text encoder	models/text_encoders/	HuggingFace
ae.safetensors (VAE)	models/vae/	HuggingFace

Usage

• ZIMG — Text-to-Image. Enter a prompt, pick a resolution, generate.
• Z-I2I — Image-to-Image. Load an image, adjust denoise strength, generate.

LoRA support

Up to 3 LoRAs supported. Place your LoRA files in:

Platform	Path
Standard	ComfyUI/models/loras/
Windows (ComfyUI Desktop)	%AppData%\Local\Comfy-Desktop\ComfyUI-Shared\models\loras

Trigger words for each LoRA are loaded from config.json and can be customized from the node settings.

---

Original author: yanokusnir-ai — creator of the FLUX.2 [klein] custom node from which this project was adapted.

Z-Image adaptation & simplification: Adeliox