ComfyUI Extension: ComfyUI_JanusProCaption

ComfyUI-JanusProCaption

English | 简体中文

ComfyUI-JanusProCaption is a tool focused on generating image descriptions, powered by the Janus Pro model. It provides users with an intelligent and straightforward way to describe image content. Whether it's used in image-to-image workflows or for generating captions for LoRA model training, this tool meets diverse needs. <br>

Special thanks to https://github.com/deepseek-ai/Janus for releasing the powerful Janus Pro model, and to https://github.com/CY-CHENYUE/ComfyUI-Janus-Pro for implementing Janus Pro's methods for image generation and reverse prompting within ComfyUI. Based on their work, I have modified the functionality to suit my needs. This project focuses on leveraging Janus Pro for image understanding, including describing single images and generating caption files for LoRA training datasets. <br>

Current Features

1. JanusProDescribeImage: Describe a single image in detail using the Janus Pro model. This feature allows users to input an image and receive a detailed description based on a provided question. <br>

2. JanusProCaptionImageUnderDirectory: Automatically generate captions for all images in a specified directory using the Janus Pro model. This feature processes each image, generates a description, and saves the description in a corresponding .txt file. <br>

Contact Me

• B站: @黎黎原上咩
• YouTube: @SweetValberry