I want to LoRa of a scene where my favorite person is handcuffed. It seems that handcuffs are difficult to generate in Stable Diffusion. I initially trained with 30 images, but they didn't appear. Now, I'm using 50 images with a strength of 0.7, and I'm finally getting decent images. My dataset includes scenes where police officers are apprehending individuals, so I thought using keywords like 'police,' 'policewoman,' and 'holding another's wrist' would work, but it only generates scenes where someone is handcuffed.