A LoRA I trained for fun to test a workflow, using a dataset of curvier body types and underboob T-shirts.

Best suited for fuller, softer physiques. If your prompt relies on extreme muscle definition or bodybuilder anatomy, expect the LoRA to soften those features.

Recommended strengths

• 0.7–0.8: Natural fuller proportions while preserving prompt fidelity (some prompts may work better around 0.6).

• 0.9–1.0: Stronger effect with more pronounced curves.

• Above 1.0: Experimental; may overpower prompt details or significantly alter body proportions.

This was primarily created as a workflow experiment, but if you enjoy fuller body types, you might find it useful for a variety of fantasy, sci-fi, historical, and modern prompts.