You may have seen "secret codes" like /drone camera or /photo booth that change a photo's style in ChatGPT. They are not hidden features. They are short labels that ChatGPT reads and expands into a full image instruction. Below is a setup that makes them work the same way every time, plus 20 ready codes.
What you need
A ChatGPT account that can upload and create images. chatgpt.com
A clear photo of yourself (or you and your partner) with your face well lit.
Optional: a ChatGPT Project, so you only paste the code list once.
Step by step
1. Know what the codes are. OpenAI has not built in photo codes. Typing a code like /droneview works only because ChatGPT understands the words, as Schieb.de explains. If you type "/" in the message box, ChatGPT may show its own shortcuts menu, such as image creation or search (SurePrompts). Those are different from these codes. To get the same result each time, give ChatGPT the meaning of each code first.
2. Paste the code list. Paste it at the start of a chat. Or create a Project, open its instructions, and paste it there so every chat in that Project knows the codes.
When I upload a photo and type one of these codes, edit my photo into that style. Keep every person's face, skin tone, hairstyle, body shape and clothing the same unless the code says otherwise. Output one photorealistic image unless the style is illustrated. Codes:
/drone = cinematic aerial drone shot looking down from about 30 metres, me small in a wide scenic landscape, golden hour light, soft long shadows
/photobooth = vertical 4-frame photo booth strip with white border, flash lighting, a different playful pose and expression in each frame
/convexmirror = me reflected in a round convex traffic safety mirror on a pole, fisheye curve, street background
/sidemirror = me seen in a car's side-view mirror, mirror sharp in the foreground, car door and road softly blurred
/goldenhour = relight to warm sunset light from one side with gentle lens flare
/polaroid = square instant-film print with white frame, slightly faded colours, lying on a wooden table
/vintagefilm = 1990s 35mm film look, grain, warm tones, slight light leaks
/noir = black and white film noir, hard side light, deep shadows
/cctv = grainy security-camera still from a high corner, timestamp in the corner
/magazine = fashion magazine cover with a clean masthead and two short cover lines
/closeup = tight portrait crop, shallow depth of field, soft background blur
/newbg = keep me unchanged, replace the background with [describe place]
/rain = night street in light rain, neon reflections on wet road
/mirrorselfie = me taking a mirror selfie in a stylish modern bathroom
/window = me seen through a rainy café window, reflections on the glass
/studio = professional studio portrait, plain grey backdrop, soft key light
/chibi = cute chibi cartoon version of me, big head, small body, clean outlines
/isometric = small 3D isometric diorama of me in my favourite room
/blueprint = architectural blueprint-style line drawing of the scene
/sketch = pencil sketch on textured paper
If I add words after a code, treat them as extra details. If I use two codes together, combine them. If I use more than two, ask me which matters most.3. Use a code. Upload your photo and type just the code, or add details:
/drone on a hill above Munnar tea gardens at sunrise/photobooth couple version, matching outfits, laughing in frame 44. Fix the result if needed. If the face drifts, reply:
Keep everything, but make the face match my original photo exactly. Don't change my features, skin tone or age.5. Make your own codes. Add a line to your list in the same format: /code = what it should look like. Short, visual descriptions work best. Name the camera angle, the light and the setting.
Good to know
Results vary. The same code can give different images each time. The model also invents details that weren't in your photo, like the landscape in a drone shot.
Faces may change slightly. Image models can alter faces. Check before posting.
Label AI images. Instagram and other platforms ask you to label realistic AI-edited or AI-generated images, especially of people. Use the AI label when you post.
Consent. Only edit photos of people who have agreed, and don't upload photos of children, IDs or documents.
Limits. How many images you can create depends on your ChatGPT plan.
Privacy. If you don't want your photos used for training, turn off Settings > Data controls > Improve the model for everyone.
Sources
Found this useful? Follow @kevinsingh.me on Instagram for a new AI tool every day, and subscribe so you don't miss the next one.