Control-LoRA utilizes low-rank parameter optimization added to ControlNet, providing a more efficient and compact model control method for consumer-grade GPUs. This product includes multiple Control-LoRA models, featuring capabilities such as MiDaS and ClipDrop depth estimation, Canny edge detection, photo and sketch coloring, and Revision. The Control-LoRA models are trained to generate high-quality images across diverse image concepts and aspect ratios.