The model was trained using a large RevAnimated model, but the picture was not run with this big model. Using other 2.5D models can also have good effects, and the weight will be more natural between 0.4-0.8.