This is to help fix txt. file captions made with auto captioners that uses a lot of the own subject body position, that can be tricky for inference and training with models like Flux.
- cd into this directory after cloning the repo
- start the app
py SR.py
It will ask for the input folder (will scan on the subfolders as well) on the cmd/powershell command. Press enter. It will output a report.
I also adjusted it to add after each substitution a string to help visualize manually each substitution made. You should remove this <!###-----------###> with another SR program like Notepad++ or Taggui.