ComfyUI MagicClothing Custom Nodes Repository

Introduction

The ComfyUI MagicClothing repository provides custom nodes designed to enhance the functionality of ComfyUI by adding capabilities related to clothing image generation, animation, and inpainting. This repository is particularly useful for those looking to integrate advanced clothing synthesis and manipulation workflows into their ComfyUI platform.

Installation

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

1. Clone the Repository: Navigate to the custom_nodes directory of your ComfyUI installation and clone the repository:

   git clone https://github.com/frankchieng/ComfyUI_MagicClothing.git
   

2. Install Required Python Packages: Navigate into the cloned directory and install the dependencies:

   pip install -r requirements.txt
   

3. Download Required Models: Download the necessary models from the Huggingface MagicClothing repository. Place them in the checkpoints directory of the repository as follows:

   |-- ComfyUI_MagicClothing
       |-- checkpoints
           |-- cloth_segm.pth
           |-- magic_clothing_768_vitonhd_joint.safetensors
           |-- OMS_1024_VTHD+DressCode_200000.safetensors
           |-- stable_ckpt
               |-- garment_extractor.safetensors
               |-- ip_layer.pth
   

Repository Purpose

This repository is designed to provide advanced customization options for ComfyUI, focusing on generating and modifying cloth images. It supports a range of workflows that leverage face detection, openpose integration, and specialized lower and full-body models to produce high-quality output specific to clothing.

Custom Nodes

The following custom nodes are provided within this repository:

• MagicClothing_Generate: Used for generating cloth-based images based on a set of prompts and constraints.
• MagicClothing_Animatediff: Facilitates the creation of animated effects in clothing images.
• MagicClothing_Inpainting: Used for inpainting cloth images to adjust, enhance, or fill in missing areas.

Special Features and Capabilities

The repository includes several noteworthy capabilities:

• FaceID Integration: The repository can integrate IPAdapter FaceID technology with human face detection to synthesize realistic face and cloth images.
• OpenPose Support: Enhancements using controlnet openpose allow for detailed body positioning, improving cloth image synthesis precision.
• Model Variety: Includes models tailored for upper body, lower body, and full-body cloth synthesis, useful for a range of applications from simple garments to complete outfits.
• Clothing Inpainting: Allows for seamless integration and modification of existing cloth images.

Use in ComfyUI Workflows

The nodes and accompanying workflows provided by this repository are designed to be integrated into ComfyUI workflows, leveraging its advanced capabilities for artistic and commercial purposes. The nodes allow users to perform:

• Advanced clothing generation, suitable for digital fashion design and visualization.
• High-quality animations of clothing items, enhancing interactivity and engagement in user interfaces.
• Precise inpainting of clothing elements, allowing for modification and repair of cloth prototypes or concepts.

By providing these functionalities, the ComfyUI MagicClothing repository enables users to expand their creative capacities within ComfyUI environments significantly.