ReActorSaveFaceModel Node Documentation

Overview

The ReActorSaveFaceModel node is a part of the ComfyUI-ReActor extension. It is designed to save facial models derived from images or facial data. This node allows users to store facial embeddings as "safetensors" files within a specified directory. These stored models can be used later in face-swapping workflows or other operations involving face models.

What It Does

The ReActorSaveFaceModel node processes an input image or a pre-existing face model to create a facial model. This model captures the unique facial features and is saved in the form of a "safetensors" file. Users can specify a name for the saved model, allowing easy identification and retrieval for future use.

Inputs

This node accepts the following inputs:

• Save Mode (Boolean): This input determines whether the node is in saving mode. When enabled, the node actively saves the face model.

• Face Model Name (String): This input specifies the name under which the facial model will be saved. Users can provide a custom name to identify the saved model file.

• Select Face Index (Integer): This input allows users to specify the index of the face to be processed from an image. This is useful when the image contains multiple faces, and the user wants to save a specific one.

• Image (Optional): An optional input where users can provide an image containing the face(s) they want to save as a model. If an image is provided, it will be processed to extract and save facial features.

• Face Model (Optional): Users can optionally provide a pre-existing face model that they wish to save. This is useful when working with models created or processed in earlier steps of the workflow.

Outputs

The ReActorSaveFaceModel node does not produce any direct outputs. Instead, it performs the action of saving the face model to a file specified by the user.

Usage in Workflows

In ComfyUI workflows, the ReActorSaveFaceModel node is used primarily in scenarios where users need to:

1. Preserve Facial Models: When users wish to store a face model extracted from an image for future reference or use, this node facilitates the saving process.
2. Create Facial Databases: Users can build a library of face models by saving multiple models, each identified by a unique name. This library can be used in various facial recognition or swapping tasks.
3. Integration into Complex Workflows: This node can be integrated into broader workflows involving facial recognition, face-swapping, or model comparison tasks.

Special Features and Considerations

• Support for Multiple Faces: If an image contains multiple faces, users can specify which face to process and save using the "Select Face Index" input.
• Custom Naming: The ability to name saved models allows users to organize and manage face model files effectively.
• Mode of Operation: The node features a Save Mode toggle, allowing users to disable saving when necessary. This ensures users have control over when and what to save.
• Seamless Integration: Though it doesn't produce a direct output, the node's capability to save face models provides value in ComfyUI workflows by allowing reuse of models across different sessions or projects.

By incorporating the ReActorSaveFaceModel node into ComfyUI workflows, users gain the ability to efficiently save and catalog facial models, enhancing their face-swapping and recognition capabilities.