This LoRA was trained on highly detailed images with pastel colors and full of details leaving almost no blank spaces. It tends to generate simplistic body shapes with round heads and long legs. Works best with a specific color pallette.