ComfyUI-Janus-Pro Documentation

Overview

The ComfyUI-Janus-Pro repository provides a set of custom nodes for ComfyUI, known as Janus-Pro nodes. These nodes are designed to support a unified multimodal understanding and generation framework, which can be integrated into ComfyUI workflows to enhance capabilities in image understanding and generation.

Installation

Method 1: Install via ComfyUI Manager (Recommended)

1. Install ComfyUI-Manager.
2. Search for "Janus-Pro" within the manager interface.
3. Click to install the package.

Method 2: Manual Installation

1. Clone the repository into your ComfyUI custom nodes directory:

   cd ComfyUI/custom_nodes
   git clone https://github.com/CY-CHENYUE/ComfyUI-Janus-Pro
   

2. Install the required dependencies:

   For Windows:

   • ComfyUI Portable Version:

     cd ComfyUI-Janus-Pro
     ..\..\..\python_embeded\python.exe -m pip install -r requirements.txt
     

   • Custom Python Environment:

     cd ComfyUI-Janus-Pro
     path\to\your\python.exe -m pip install -r requirements.txt
     

   For Linux/Mac:

   • Using ComfyUI's Python Environment:

     cd ComfyUI-Janus-Pro
     ../../python_embeded/bin/python -m pip install -r requirements.txt
     

   • Using Your Environment:

     cd ComfyUI-Janus-Pro
     python -m pip install -r requirements.txt
     

   Notes for Installation Challenges:

   • Ensure git is installed.
   • Update pip if necessary: python -m pip install --upgrade pip.
   • Configure any proxies to allow git access to GitHub.
   • Use the same Python environment as ComfyUI.

Model Download

To get the Janus-Pro models:

1. Create a directory ComfyUI/models/Janus-Pro within your ComfyUI models folder.

2. Download models from Hugging Face:

   • Janus-Pro-1B
   • Janus-Pro-7B

3. Extract the downloaded models to:

   ComfyUI/models/Janus-Pro/Janus-Pro-1B/
   ComfyUI/models/Janus-Pro/Janus-Pro-7B/
   

Node Mappings

The repository provides the following nodes:

• JanusModelLoader - Load and manage models for multimodal capabilities.
• JanusImageUnderstanding - Facilitate image comprehension within workflows.
• JanusImageGeneration - Enable image generation based on multimodal inputs.

Features and Capabilities

• Unified Multimodal Framework: Combines capabilities for both understanding and generating images, integrating this seamlessly into ComfyUI workflows.
• Scalable Models: Offers support for different model sizes (1B and 7B), allowing flexibility depending on the task complexity and available compute resources.

Use in ComfyUI Workflows

The Janus-Pro nodes are designed to significantly enhance the capability of ComfyUI by bringing in advanced multimodal processing. They can be effectively used in cases where both image understanding and image creation are required, offering a more comprehensive approach compared to traditional single-modal nodes.

Contact and Support

For inquiries or support, you can reach out to the repository author through:

• Twitter: @cychenyue
• TikTok: @cychenyue
• YouTube: @CY-CHENYUE
• BiliBili: @CY-CHENYUE
• 小红书: @CY-CHENYUE

If this project assists you, please consider giving it a Star on GitHub. ⭐️