FaceSwap - AI-Powered Headshot Generator

FaceSwap is a flow designed to seamlessly swap faces between two images. This project is built on top of ComfyUI, a modular and flexible interface that enables easy integration and customization of AI workflows. FaceSwap utilizes models to detect key facial landmarks, map them to the target face, and generate a realistic face swap by considering aspects like lighting, expression, and pose. This flow is designed to provide an easy-to-use interface through ComfyUI, allowing users to swap faces effortlessly with minimal setup.

Requirements

To run FaceSwap, you'll need to install ComfyUI and set up a few dependencies. Follow the steps below for installation and setup.

Prerequisites

Python 3.9 or higher (Ensure Python 3.9+ is installed on your machine)
NVIDIA GPU (Recommended for faster processing, though CPU can also be used)

Step-by-Step Installation

Clone the ComfyUI Repository Start by cloning the main ComfyUI repository:

git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI

Install Dependencies Install all the required dependencies for ComfyUI:

pip install -r requirements.txt

Install ComfyUI-Manager To manage custom nodes, you'll need to install ComfyUI-Manager. Run the following commands inside the ComfyUI/custom_nodes directory:

cd ComfyUI/custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Manager.git

After cloning the ComfyUI-Manager, restart ComfyUI.

Install Reactor Node for ComfyUI by clicking on Manager -> Custom Nodes Manage

or you can clone the repository in the directory ComfyUI/custom_nodes

git clone https://github.com/Gourieff/comfyui-reactor-node.git

Download Pre-trained Models Download inswapper_128.onnx and place the downloaded model in the ComfyUI/models/insightface/ directory

Note: If directory don't exist, create them manually.

Start ComfyUI Run ComfyUI by executing the following command:

python3 FaceSwap.py

This will start ComfyUI, and you can access the interface by navigating to http://127.0.0.1:8188 in your browser.

Setting Up and Using the FaceSwap Workflow

Clone the Headshot_Generator-FaceSwap Repository Clone this repository that contains the FaceSwap workflow and assets:

git clone https://huggingface.co/foduucom/Headshot_Generator-FaceSwap
cd Headshot_Generator-FaceSwap

How to use Once ComfyUI is running, open the browser window (default: http://127.0.0.1:8188), and follow these steps:

Click on the Load button in the menu bar and select the workflow.json file from this repo you just cloned.
Upload the Workflow.json to start the face-swapping process.
Provide Input Images
- Source Image: The image containing the face you want to swap.
- Target Image: The image where the face will be swapped into.
Click Queue Prompt to initiate the face-swapping process. The AI model will process the images and generate the output.

or you can use by python script provided in this repository:

python3 FaceSwap.py

#Remember change the input paths in script here :
#prompt["1"]["inputs"]["image"] = "//put your input image"
#prompt["2"]["inputs"]["image"] = "//put your source face image"

Troubleshooting

ComfyUI Not Running: Ensure that all dependencies are installed correctly and that you’re using Python 3.9 or higher. If issues persist, check the ComfyUI GitHub repository for troubleshooting guides.
Missing Models: Ensure you’ve downloaded the required models (sam_vit_h_4b8939.pth and groundingdino_swint_ogc.pth) and placed them in the correct directories (ComfyUI/models/sams/ and ComfyUI/models/grounding-dino/).

Slow Performance:

Using a GPU is highly recommended for better performance. If you’re using a CPU, the processing time may be longer.

Compute Infrastructure

Hardware

NVIDIA GeForce RTX 3060 card

Model Card Contact

For inquiries and contributions, please contact us at [email protected].

@ModelCard{
  author    = {Nehul Agrawal and
               Priyal Mehta},
  title     = {FaceSwap - AI-Powered Face Swap / Headshot Generator},
  year      = {2024}
}