joy-caption-alpha-two

Currently the best visual language large-scale model on the market, suitable for image reverse prompting words and the labeling work required for alchemy.

This tool is based on the newly released Joy2. A foreigner made a GUI which I find quite good, so I translated it into Chinese and made it into an integrated package. I hope it can help little friends who have demand in this area.

Installation Instructions

After downloading and extracting the files, you should see the contents as shown in the picture (except for venv). Please check the "Running Instructions" for detailed information first.

setup.ps1 is an automation script that will automatically create a virtual environment (venv file) to avoid conflicts with system dependencies and install the dependencies required by the tool.

run.bat is the startup file. After successfully installing the dependencies, run it (just click it each time you want to run the program).

Instructions for Use

Area A: Select multiple images/single image for labeling.

Area B: You can choose the type of prompting words and do some filtering (ComfyUI users should be familiar with this part).

Area C: If it's about training characters, you can name the characters here or customize trigger words.

The bottom set of buttons are the execute buttons, and the text on them is already very clear. It's worth noting that when you run it for the first time, you should first select "Load Model". The program will automatically download the required models. Please ensure smooth network throughout, once the loading is complete, you can label as needed.

Feedback is welcome if there are any issues with this integrated package for the first time.