8 ChatGPT Image Codes for Any Image

One image can become eight very different visuals. Upload the same image each time, choose a look below, and use the full prompt under its slash label.

What this helps you do

Make up to eight visual versions of a photo, drawing, or other image while keeping its main subject recognizable. For a photo of a person, that means the same face in every version. Each slash word is the short name for one of the prompts below.

Before you start

• A clear image that you own or have permission to edit.
• Access to ChatGPT Images.
• Three details you want to keep. For a person, that could be their hair, outfit, and pose. For an object, it could be its shape, color, and label.

Terms to know

• Reference image: the image you upload so ChatGPT has a subject and composition to work from.
• Style: the visual treatment, such as paper cut or comic art.
• Slash label: the short name for a full prompt, such as /CLAYWORLD. Paste the full prompt below after uploading your photo.
• Drone view: a shot taken from high above, looking straight down.
• Fish-eye lens: an extreme wide-angle lens that curves the edges of the picture.

Step-by-step

1. Upload your image to ChatGPT Images.
2. Pick one of the eight prompts below. Paste the full text into the chat with the image attached.
3. Replace [THREE DETAILS TO KEEP] with the features that make your subject recognizable. If the photo shows a person, make sure the prompt says "keep the person recognizable." The /STUDIOQUALITY, /DRONEVIEW, /FISHEYE, /COMICCOVER, and /CLAYWORLD prompts already cover it. Add it to the other three. Leaving it out is the most common reason a face drifts.
4. Save the result. Start a new image request with the original image for the next style, rather than editing the previous result.
5. Place every version beside the original. Check the subject and your three details in every image. Regenerate any version that changes them too much.

Copy this

Run each one as a separate image request. Use the same original image every time.

A good five-code set for a portrait is /STUDIOQUALITY, /DRONEVIEW, /FISHEYE, /COMICCOVER, and /CLAYWORLD.

1. /PAPERCUT

Use my uploaded image as the reference. Recreate its main subject and scene as handmade paper-cut art. Use layers of colored paper, visible cut edges, and soft shadows between the layers. Keep the subject in the same position and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

2. /CLAYWORLD

Use my uploaded image as the reference. Recreate its main subject and scene as a miniature clay set photographed on a tabletop. Give the clay rounded edges, subtle handmade texture, and soft studio lighting. Keep the subject in the same position, keep the person or main subject recognizable, and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

3. /COMICCOVER

Use my uploaded image as the reference. Turn its main subject into a vertical comic-cover illustration with bold black ink lines, deep shadows, rich color, and expressive shading. Keep the person or main subject recognizable and preserve [THREE DETAILS TO KEEP]. Leave clear space at the top for a title I can add later. Do not add text.

4. /BLUEPRINT

Use my uploaded image as the reference. Redraw only what is visible as clean white contour lines on a deep blue background, like a design concept drawing. Keep the main subject in the same position and preserve [THREE DETAILS TO KEEP]. Do not invent hidden parts, measurements, or labels.

5. /NIGHTSCENE

Use my uploaded image as the reference. Reimagine the same subject and composition after dark. Add warm light, deep shadows, and reflections where they fit the scene. Keep the main subject recognizable and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

6. /STUDIOQUALITY

Use my uploaded photo as the reference. Turn it into a studio-quality portrait. Brighten the face and clean up the light with soft, even studio lighting. Keep the background simple and uncluttered. Keep the person recognizable, with the same face, pose, and expression, and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

7. /DRONEVIEW

Use my uploaded photo as the reference. Rebuild the same scene from directly above, as if a drone took the photo looking straight down. Show the person and the space around them from overhead. Keep the person recognizable and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

8. /FISHEYE

Use my uploaded photo as the reference. Reshoot the same scene through a dramatic wide-angle fish-eye lens, close to the subject, with curved edges. Keep the person recognizable and preserve [THREE DETAILS TO KEEP]. Do not add words or logos.

The one line for photos of people

Add this line to any prompt above when your photo shows a person.

Keep the person recognizable: same face, same hairstyle, same expression.

Choose the details to keep

For a portrait, name the person's hairstyle, clothes, and pose. For a product, name its shape, color, and visible label. For a room, name the furniture layout, wall color, and main light source. Use the details that make your image recognizable, then run every style from the same original.

Things to know

Changing the visual style can change small details. Compare each result with the original before using it. If you need an exact product logo or readable text, add that later in an image editor instead of relying on generation to reproduce it.

If something goes wrong

• The subject changes: rerun the prompt and name the features that must stay recognizable.
• An important detail disappears: ask ChatGPT to restore that detail while keeping the rest of the image.
• The face disappears in /DRONEVIEW: ask for a slightly angled view from above so the face stays visible.
• The face stretches too much in /FISHEYE: keep the fish-eye effect but ask for the face to stay near the center of the frame.
• One style looks too similar to the original: strengthen the material or drawing instructions for that style, then rerun it from the original image.

Final check

You have five or more visibly different images from the same original, and each one still shows the main subject with the three details you chose to preserve. For a portrait, the person's face is recognizable in every version.