I've seen dall-e being able to imagine additional parts to an existing image along with a prompt, I'm sure it's possible. But probably not with this amount of creativity.