Enter a text prompt and watch a pair of models take turns making the prompt into an image and trying to guess the image prompt.