After extracting the depth map of the image, various production styles can be obtained by using models trained with specific rules, such as Stable Diffusion. These models are trained with certain rules or constraints during the learning phase, and then generate new images based on these rules or constraints.