FaceSwap with the Inswapper Model

One of the most fun applications of AI-powered digital art is the faceswap. It allows you to imagine yourself or your friends as characters in any story – from famous movie scenes to video games to anything else that you may have created with Stable Diffusion.

That's why we have added a faceswapping workflow on InstaSD with one of the most powerful faceswap models: inswapper_128.

For a simple demonstration of how this faceswapping workflow using ComfyUI works, I took a photo from phone as the source and tried faceswaps with iconic scenes from movies and videogames. Here is the original photo (I had the background removed with the Photos app on iOS):

And here are some of the results.

How to Use the Faceswapper Workflow on InstaSD

First, launch an instance with the FaceSwap workflow.

Inside the FaceSwap workflow, we have two inputs:

• The source image provides the face that you want pasted
• The input image is the photo to which you want to paste the face.

Next, we're using a custom node, ReActor Node for ComfyUI with our swap model inswapper_128.

This already gives us a pretty good result. But to make things even better, we pipe this output to a face restore model:

And with just such a simple workflow, we have a powerful and fast workflow.

‍