Indeed, but the only way the AI is going to learn to draw a 3d object from different perspectives is if it understands the object in a three dimensional rather than 2d sense. Simply training it on 2d images with depth perspective is not going to help it anymore than training it on 2d images, it will simply get better…