There was no Z Image Base LoRA for big areolas yet so I whipped one up.

Trained on 113 manually-tagged images at 1024x1024 (was originally meant for SDXL) for 2000 steps. The captions were a mix of danbooru-style tags and natural language.

No trigger word, but each image had "large areolae" captioned so perhaps it's a good idea to include that in your prompt. Sagging breasts are also supported as they were part of the dataset.

0.75-1.00 strength should be fine.