DeepSeek Janus Pro ComfyUI Workflow

Janus Pro is a multimodal model open-sourced by DeepSeek on January 27, 2025, capable of both image understanding and generation.

This tutorial is based on the https://github.com/CY-CHENYUE/ComfyUI-Janus-Pro plugin, providing a detailed guide on how to set up and use DeepSeek Janus Pro in ComfyUI.

Janus Pro ComfyUI Step-by-Step Tutorial

1. Install the ComfyUI-Janus-Pro Plugin

Plugin repository: https://github.com/CY-CHENYUE/ComfyUI-Janus-Pro Installation methods:

• Install via ComfyUI Manager “Janus-Pro” (recommended)
• The ComfyUI-Janus-Pro repository also contains detailed plugin installation instructions, especially for manual installation where you need to install related dependencies.
• For other installation methods, please refer to the ComfyUI Plugin Installation Tutorial.

Ensure your network can access GitHub properly.

2. Download DeepSeek Janus Pro Models

In this step, we need to download the DeepSeek Janus Pro 1B or 7B models.

DeepSeek Janus Pro Model Download and Installation

You need to download all json format files and .bin format model files. After installation, the folder structure should look like this:

📁ComfyUI
├── 📁models
│   ├── 📁 Janus-Pro
│ ...   ├── 📁 Janus-Pro-1B 
│       │   ├── config.json
│       │   ├── preprocessor_config.json
│       │   ├── processor_config.json
│       │   ├── pytorch_model.bin
│       │   ├── special_tokens_map.json
│       │   ├── tokenizer.json
│       │   └── tokenizer_config.json
│       └── 📁 Janus-Pro-7B   
│           ├── config.json
│           ├── preprocessor_config.json
│           ├── processor_config.json
│           ├── pytorch_model-00001-of-00002.bin
│           ├── pytorch_model-00002-of-00002.bin
│           ├── pytorch_model.bin.index.json
│           ├── special_tokens_map.json
│           ├── tokenizer.json
│           ├── tokenizer.json
│           └── tokenizer_config.json

3. Janus Pro Workflow File Download

This workflow node includes both image description and image generation.

1. At position 1, select either the 1B or 7B model.
2. At position 2, upload the image you want Janus Pro to interpret.
3. At position 3, you can adjust the prompt for image description. I have tested inputs in Chinese, English, and Japanese without issues, but if you need output in a specific language, please specify it in the prompt, such as Please output in Japanese.
4. Position 4 is a text display node. Here, I am using the Show Text node from the ComfyUI-Custom-Scripts plugin.
5. At position 5, enter the prompt for image generation.

Janus Pro Related Links

Janus Pro Huggingface Model Downloads

Janus Pro Github Repository:

https://github.com/deepseek-ai/Janus

Experience DeepSeek Janus Pro on Huggingface Space

Chat With Janus 1.3B: https://huggingface.co/spaces/deepseek-ai/Janus-1.3B Chat With Janus Pro 7B: https://huggingface.co/spaces/deepseek-ai/Janus-Pro-7B Chat With Janus: https://huggingface.co/spaces/deepseek-ai/JanusFlow-1.3B

DeepSeek Janus-Pro ComfyUI Node

ComfyUI-Janus-Pro Node: https://github.com/CY-CHENYUE/ComfyUI-Janus-Pro