Qwen ControlNet + Style Reference Workflow in ComfyUI

About This Workflow

Qwen ControlNet + Style Reference Workflow

This Qwen ControlNet workflow in ComfyUI lets you generate AI images using both a ControlNet reference image (for pose/structure) and a style reference image (for look and design).

Unlike normal text-to-image workflows, this setup uses ControlNet + Style Reference together. That means the AI doesn’t generate random compositions. Instead, it follows your ControlNet input image for accuracy, and applies the style of your reference style image.

This workflow is ideal for:

Character design with consistent style. Fashion and portrait generation. Recreating cinematic lighting from a reference photo. Anime or illustration style transfer. AI photoshoots using real pose references. Creating a full image set with matching style.

Required Models

Model links

Diffusion model

• Qwen-Image-Edit-2509_bf16.safetensors

LoRA

• Qwen-Image-Edit-Lightning-8steps-V1.0.safetensors

Text encoder

• qwen_2.5_vl_7b_fp8_scaled.safetensors

VAE

• qwen_image_vae.safetensors

ControlNet

• diffusion_pytorch_model.safetensors

Model Storage Location

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 diffusion_models/
│   │   ├── Qwen-Image-Edit-2509_bf16.safetensors
│   ├── 📂 loras/
│   │   └── Qwen-Image-Edit-Lightning-8steps-V1.0.safetensors
│   ├── 📂 controlnet/ 
│   │   └── diffusion_pytorch_model.safetensors
│   ├── 📂 vae/
│   │   └── qwen_image_vae.safetensors
│   └── 📂 text_encoders/
│       └── qwen_2.5_vl_7b_fp8_scaled.safetensors

Workflow Graph