🚀 Fast, seamless head swaps for character images — preserve lighting, expression, and style for photoreal results.
▶️ Run Directly in Cloud:
https://www.runcomfy.com/comfyui-workflows/z-image-turbo-headswap-for-characters-in-comfyui-seamless-face-replacer?utm_source=civitai
💡 Overview
Z Image Turbo Headswap helps you replace heads on character images in seconds while keeping lighting, natural expressions, and overall style coherent. Built for designers and creators who need high-quality, stylistically consistent face replacement without heavy manual masking.
✨ Key Features
Z Image Turbo backbone: Fast, detailed generations.
Aligned swaps: Precise head alignment and blending.
Photoreal-friendly: Keeps skin tone and scene lighting consistent.
🚀 Getting Started
Provide source and target face / character inputs per the graph.
Adjust strength and region controls if exposed.
Generate and iterate.
Click the "Run Directly" link above to bypass local setup and test this workflow immediately in your browser.
Description
Initial release.
FAQ
Comments (1)
Hi! I downloaded the workflow and checked the JSON locally. I can only find one actual image input node: LoadImage titled Source Image. I do not see a second LoadImage / Target Image / Reference Image node anywhere in the graph.
From the graph structure, Image Comparer seems to use:
image_a = final output from Inpaint Stitch
image_b = resized version of the same Source Image
So at the moment it looks like the workflow is doing a single-image inpaint/restyle pass with SAM3 face/hair masking, prompt conditioning, and character LoRAs, rather than a true source-to-target headswap with two separate image inputs.
Could you clarify where the second target input is supposed to be, or whether this uploaded JSON is missing that part of the graph?
Thanks.