Midjourney’s /describe command flips the usual process. Instead of you describing a scene for the AI to generate, you upload an image—and the AI tells you what it sees. Just type /describe in the Discord chat, upload your image, and hit enter. Midjourney will return four different text prompts that it thinks could have produced that image.
Each result is clickable, meaning you can instantly re-generate the image or explore variations based on those prompts. It’s great for reverse-engineering styles, studying visual language, or finding inspiration when you don’t know how to describe something yourself. Whether you want to remix an old photo or learn how Midjourney “thinks,” /describe gives you a peek behind the curtain.
