Imagine a world where the bullet bra never went out of style. When this LoRA is active, bullet bras are just what every girl wears. For example, if your prompt contains the phrase: "she is wearing a green dress. She has medium breasts." You'll get the green dress with a medium bust size but now the silhouette has prominent conical points. If you prompt for a subject wearing a bra, the bra will be a bullet bra. The keyword is "bulletbra" but this doesn't need to be in your prompt, As long as the lora is active and weighted high enough (1 - 1.2 seems to work well), you'll get the effect.

This is my first LoRA and it is version 1. It doesn't work well if prompted for small breasts. I think this is a result of training data the skewed too heavy to larger bust sizes. All training images were in color. Hair styles of the training images were mostly modern hairstyles, though a few images were left with hairstyles popular in the 1940s.

All training images were photorealistic AI images but the LoRA seems to be flexible enough to handle anime and CGI prompts. Occasionally, images are generated where one breast is conical and the other is globular. This seems to be a consequence of having some training images where the conical form of one breast was not well defined relative to the other. Try a different seed if this happens.