Create vision and DALL·E feedback loop. Start by uploading a photo or an image, then keep re-uploading the result image. No text prompt required for image uploads.