Z-Image ControlNet Workflow in ComfyUI

About This Workflow

Z-Image ControlNet Workflow

The Z-Image ControlNet workflow in ComfyUI is a powerful setup that helps you generate highly accurate AI images using a reference image.

This workflow takes a reference image as input and processes it through a selected ControlNet model (such as OpenPose, Canny, Depth, Lineart, or others)

Why Use This Workflow

Normally, when you generate images using only text prompts, the results can vary a lot. Even with the same prompt, the pose and composition may change every time.

With this workflow, ControlNet helps you lock important details such as:

Body pose Camera angle Composition Depth structure Edge outlines Facial alignment (depending on ControlNet type)

This makes the workflow extremely useful for creators who want stable and repeatable results.

Required Models

Model links

Diffusion model

• z_image_turbo_bf16.safetensors

Model Patches

• z-image-turbo-fun-controlnet-union.safetensors

Clip - Text Encoders

• qwen_3_4b.safetensors

VAE

• flux_vae.safetensors

Model Storage Location

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 diffusion_models/
│   │   ├── z_image_turbo_bf16.safetensors
│   ├── 📂 model_patches/
│   │   └── z-image-turbo-fun-controlnet-union.safetensors
│   ├── 📂 vae/
│   │   └── flux_vae.safetensors
│   └── 📂 text_encoders/
│       └── qwen_3_4b.safetensors

Workflow Graph