Text-to-Image + CN_Depth + LoRA: Choose between 2D and 3D models.