Standard image generation models have a very strong bias toward depicting women with arched backs.

This LoRA model supports the generation of postures with rounded backs rather than arched ones.

背中を丸める姿勢を出しやすくするLoRA