Initial version. Recommended strength is around 0.8 - 0.9. All dataset images were captioned with JoyCaption, with her name insert where appropriate.