This extension for AUTOMATIC1111's StableDiffusion web-ui enables face-swapping in generated images, targeting artists and creators for tasks like character animation or model visualization. It aims to provide a productive tool while incorporating preventative measures against misuse.

How It Works

The extension leverages the insightface library for face detection and swapping. Users select a source face image and enable the feature within the web-ui. The core mechanism involves identifying faces in the generated output and replacing them with the selected source face, offering control over which faces are swapped via a comma-separated input.

Quick Start & Requirements

• Install via AUTOMATIC1111 web-ui's "Extensions" tab using the URL: https://github.com/s0md3v/sd-webui-roop
• Requires insightface==0.7.3.
• For Windows, Visual Studio with Python and C++ packages is recommended.
• Download inswapper_128.onnx model and place it in <webui_dir>/models/roop/ if encountering specific errors.

Highlighted Details

• Face-swapping functionality integrated directly into the StableDiffusion web-ui.
• Supports selecting specific faces for swapping when multiple are present.
• Offers guidance on achieving high-quality results using built-in options like "Restore Face" and "Upscaler," or img2img.
• Includes a built-in check for inappropriate media and may implement watermarks or shut down if legally required.

Maintenance & Community

• Developed by s0md3v.
• No specific community links (Discord/Slack) or roadmap are provided in the README.

Licensing & Compatibility

• The README does not explicitly state a license. The repository itself is under the MIT license.
• Compatible with AUTOMATIC1111's StableDiffusion web-ui.

Limitations & Caveats

The extension may fail to swap faces if the input image is detected as NSFW or if no face can be detected. Users are advised to consult external resources for troubleshooting installation errors.