This is the model used when making lora.
https://civitai.com/models/232545/badbro31
Trained with 768,768
I confirmed that it works great in a 512*1024 resolution full body shot.
weight : 0.5~0.8
trigger : Basically, entering
'negligee'
is stable.
And depending on your taste,
'black dress', 'brown dress', 'red dress', 'see-through', 'blue dress', 'green dress', and 'pink dress' are useful.