I figured that much, but if you do an image prompt, sometimes it returns something that sorta resembles the person. Or there are services where you can feed it multiple images, it trains on them and then outputs other variations.