This AI model generates new images while preserving the original image characteristics by applying Lineart ControlNet to the input image and performing Stable Diffusion's Image2Image generation. Combined with IP-Adapter, it enables generation that better