suggest weight: 0.4

I try to train a LORA that can do several views of cowgirl,as alpha vision, it works ,like 4 out of 10,get what you want.

add"from side","from above","from behind","from above"

prompts: straddling,cowgirl_position,girl_on_top,sex,(1girl,1boy),full body

negative prompts: hat (ps: you dont want a real cowgirl do you)

the data I used are young male & mature female, so the outcomes may be BIGGER female partner