ComfyUI PhotoMaker Custom Nodes

Introduction

The ComfyUI PhotoMaker repository is an unofficial implementation of the PhotoMaker project for ComfyUI. This repository offers a set of custom nodes designed to enhance the functionality of ComfyUI by integrating with various image processing models, such as PhotoMaker Adapters and local models, using advanced features like Lora models and style customization.

Installation

To install the ComfyUI PhotoMaker custom nodes, follow these steps:

1. Navigate to the custom_nodes directory of your ComfyUI installation:

   cd custom_nodes
   

2. Clone the repository:

   git clone https://github.com/ZHO-ZHO-ZHO/ComfyUI-PhotoMaker.git
   

3. Navigate to the cloned repository directory:

   cd custom_nodes/ComfyUI-PhotoMaker
   

4. Install the necessary Python packages:

   pip install -r requirements.txt
   

5. Restart ComfyUI to apply the changes.

Purpose

This repository serves as an extension to the ComfyUI platform, aimed at providing additional functionality to create photo-realistic images using pre-trained models. It supports customizations in terms of styling, model loading, and batch processing, making it a versatile tool for advanced image generation applications.

Node Offerings

The ComfyUI PhotoMaker repository includes the following nodes:

• BaseModel_Loader_fromhub: Loads base models from the Hugging Face hub.
• BaseModel_Loader_local: Loads base models from a local directory.
• PhotoMakerAdapter_Loader_fromhub: Loads PhotoMaker Adapter models from the Hugging Face hub.
• PhotoMakerAdapter_Loader_local: Loads PhotoMaker Adapter models from a local path.
• LoRALoader: Loads local Lora models and adjusts model weights.
• Ref_Image_Preprocessing: Processes reference images, supporting both direct input and path-based loading.
• Prompt_Styler: Integrates prompts with styling options, allowing weight adjustments and supporting various styles.
• NEW_PhotoMaker_Generation: Generates composite images using model pipelines and reference images.
• PhotoMaker_Generation: Original function for generating composite images.

Special Features

• Model Support: The nodes support downloading and loading models directly from the Hugging Face hub or from local paths.
• Lora Model Integration: Nodes for loading and adjusting Lora model weights enhance flexibility in image generation.
• Styling Options: Ability to apply a wide range of styles, such as Cinematic, Disney Character, Digital Art, and more, via the Prompt_Styler node.
• Batch Processing: Support for generating images in batches, which is useful for workflows requiring multiple outputs.
• Speed Improvements: Notable improvements in speed, reducing generation time significantly compared to previous versions.

Workflow Integration

These custom nodes are well-suited for enriching ComfyUI workflows by providing enhanced image processing capabilities. For example, integrating the NEW_PhotoMaker_Generation node with Prompt_Styler allows users to generate stylized composite images efficiently. The repository also includes pre-configured workflows like Disney-Character PhotoMaker and others, which can be directly incorporated into existing ComfyUI processes.

By leveraging these nodes, users can expand the creative possibilities within ComfyUI, making it an invaluable tool for artists, designers, and AI researchers.