>>14818450The prompts and model are pretty standard stuff. You don't need much more than "girl, standing, leotard". What makes the difference is doing a shitty photoshop of the picture beforehand. Just draw a skin color over the clothes, give it some shading if you can, and set Denoising Strength to something around 0.3-0.7 (you will have to mess around with this setting often depending on the picture). When you're done paste the result over the original picture and see if it makes sense, otherwise keep tweaking it.
