LoRA for flat chest as text-2-image or as image edit.
Works really well on Flux2-base-9b at strength 1.0 but for distilled versions of the model you may need to push the strength up to 1.5 or 1.6.
it was trained without a trigger word but captioned with "flat chested" so that works well. e.g. "image of a flat chested woman..." or for edit "make her flat chested"
Feedback appreciated and have fun with it.