SDXL HQ + ZiT Refine workflow

Hey, sharing a ComfyUI workflow I've been testing locally - still WIP, but results look promising on my end so far.

Pipeline: EasyLoader → base sampler → EasyUse HiresFix → hi-res refine → FaceDetailer → optional Z-Image polish pass.

Not claiming it's perfect yet - just sharing in case it's useful for others.

Feedback welcome if you try it.

Requirements: ComfyUI-Easy-Use, Impact Pack (+ face detector + SAM), RealESRGAN_x4plus or any upscaling models, SDXL checkpoint, qwen_3_4B or fp8, ae vae or flux vae/ultraflux improved.

Installation

Method 1: Using ComfyUI Manager (Recommended)

1. Open ComfyUI and click on Manager.

2. Click on Install via Git URL.

3. Paste the repository URL: https://github.com/thevesperhouse-hub/comfyui-ZiT-Upscale

4. Click Install. ComfyUI Manager will automatically download the nodes and install the required dependencies.

5. Restart ComfyUI.

Method 2: Manual Installation

1. Navigate to your ComfyUI custom_nodes directory in your terminal:

   cd ComfyUI/custom_nodes/

2. Clone this repository:

   git clone https://github.com/thevesperhouse-hub/comfyui-ZiT-Upscale.git

3. Navigate into the cloned folder:

   cd comfyui-ZiT-Upscale

4. Install the required Python packages:

   pip install -r requirements.txt

5. Restart ComfyUI.

Dependencies

This package requires mediapipe and transformers (which are automatically handled if installed via ComfyUI Manager or by running the pip install command).

Tested on:

• oneObsession_v22

• JuggernautXL_v8Rundiffusion

• realismByStableYogi_ponyV65

• babesByStableYogi_v65FP16

/!\ Note that some models will give you some garbage results

Before Z-Image-Turbo pass:

After Z-Image-Turbo pass: