This is a "light" version of the On/Off LoRA. It is intended to generate side-by-side pictures of individual women, with a fully dressed version on the left and the same subject wearing underwear on the right.

For best results, precede your prompt with "onoffunder, side-by-side photo" and then describe the clothing the subject should be wearing in the left (dressed) image. There is no need to describe the underwear or state of undress in the right image; the LoRA should take care of that.

"Side-by-side photo" was not in the training caption set, so it's not really a trigger word, but it is often helpful for guiding your base model toward initiating the shapes necessary for the composition to work.