My first attempt at a LoRA. This one is meant for standard modern scuba gear, and is trained on images of scuba divers in a variety of poses and angles.

Using a phrase like "scuba diving" will usually be most likely to get good results, and I lean towards a model weight of around 0.8

Using the keyword "blurry" may obviously make the picture look blurry, but also can create photos that look more candid or realistic.

Seems to work poorly with styles that aren't realistic, so I'm hoping to fix that soon.