A practical guide
AI Photo Prompts: Build a Scene Someone Would Want to Share
A useful photo prompt explains who or what is in the picture, what is happening and what should remain unchanged. The strongest social ideas also give the viewer something recognizable: a habit, an era, a shared joke or an unexpected setting.
Explore the examples ↓
Start with the premise
'Make a funny portrait' leaves the main creative decision open. 'A very formal 1980s mall-studio couple portrait with matching sweaters and a dreamy double exposure' gives the image a specific joke. The people can still look good while the photography convention does the comic work.
Separate identity from treatment
When you upload a reference photo, name the features you want preserved before describing clothes, backdrop and lighting. Do not expect an era filter to preserve every detail automatically. Compare the generated result with the reference.
Use a small number of deliberate details
A blue studio backdrop, one prop and a particular expression can establish a yearbook role. A toy's accessories can tell a story about the person without a long caption. Choose the details that make the premise understandable and remove the rest.
Supply visible text exactly
For the Group Chat Annual, prepare each name, short award and one-line quote in advance. Long titles are harder to fit. The generated sample uses four portraits with matching hierarchy so each joke is equally easy to read.
Try these changes to the examples
Change the yearbook award to a friend's recognizable habit; replace the action figure's accessories with a camper's weekend kit; move the travel portrait from a city street to a windy coastline. Change one major idea at a time so you can see what improved.
Use optimization after you know the essentials
Choose the purpose, the reference role, the exact words and the visual direction first. Then use one-shot optimization to organize the brief, review it and adopt it before generating. The process should not invent missing dates, prices or personal facts.
See the idea take shape
Explore the image. Open the prompt. Change the details that make it yours.

Group Chat Annual — Most Likely To
Original AI-generated example.
See the exact prompt
Create ONE finished flat portrait 2:3 retro yearbook page titled exactly 'GROUP CHAT ANNUAL', with subtitle 'MOST LIKELY TO...'. A confident cream-and-navy editorial layout with four equal large photographic portraits in a clean 2 by 2 grid. All are fictional adult friends aged 28–34, uniformly lit against the same mottled blue 1990s studio backdrop, warm flash and light film grain. Give each a distinctive believable friendly face and understated retro clothing. Top left: Marcus, a Black man in a cream knit sweater with a relaxed deadpan smile. Top right: Elena, a Latina woman in a striped blouse and simple headband with a bright smile. Bottom left: Jamie, an East Asian man in a navy overshirt holding a small pour-over dripper with comically serious pride. Bottom right: Chloe, a white woman with large tortoiseshell glasses holding a tiny flip phone beside her face with a knowing smile. People look good; the joke is their familiar group-chat behavior, not race, bodies or appearance. Under every portrait place EXACT text with large, clear typography: top left 'MARCUS' / 'LEAVE BY 9:02' / 'My bed is calling.'; top right 'ELENA' / 'BE 5 MINS AWAY' / 'Still in the shower.'; bottom left 'JAMIE' / 'BUY ALL THE GEAR' / 'New hobby. Day two.'; bottom right 'CHLOE' / 'SEND AN 8-MIN VOICE NOTE' / 'This could have been a text.' Preserve all spelling, align names and titles in the same hierarchy, let the longest title wrap to two lines. Plenty of breathing space, portraits about half the total layout and text the rest, no tiny filler. No actual school, logo, watermark, AI label, paper mockup or surrounding desk. Show the complete flat page edge to edge.

Awkward 80s mall portrait
Original AI-generated example.
See the exact prompt
Create an authentic-looking 1986 department-store photo-studio couple portrait, one 4:5 vertical finished photo. Fictional adult couple: a 36-year-old South Asian man with warm brown skin and a neat mustache, and a 34-year-old white woman with shoulder-length auburn curls and a few freckles. They sit side by side in matching high-neck pastel-blue and cream cable-knit sweaters. Both have stiff, painfully formal posture and polite slightly frozen 'photographer counted to three' smiles. One of his hands rests very neatly flat on her shoulder, her hands folded too neatly in her lap. In the upper right background an iconic large faint double-exposure of the SAME couple's softly lit side profiles gazes wistfully upward into space. Mottled mauve and dusty-blue airbrushed backdrop, soft-focus portrait lens, warm direct flash, restrained 35mm film grain. The warm humorous absurdity comes entirely from the dated pose and double exposure; both people are portrayed kindly and attractively with natural human features. Realistic hands, coherent repeated faces in overlay. No text, labels, borders, brands, watermark, AI mark or mockup.

The Overthinker action figure
Original AI-generated example.
See the exact prompt
Create a witty photorealistic studio product photograph of an ORIGINAL personalized action figure in a sealed clear blister package, portrait 4:5, straight-on close-up. The toy is a fictional 52-year-old Black female creative professional with deep brown skin, short silver natural curls and a knowing, mildly overwhelmed expression. Clearly a well-crafted articulated vinyl toy, not a tiny real human. She wears a warm ochre utility vest, cream shirt, olive trousers and white sneakers. Three separate molded accessory pockets beside her contain a tiny iced coffee, a miniature laptop with an abstract cluster of red unread notification dots, and a hilariously tall stack of miniature sticky-note pads. A small neck pillow sits at her feet. Cardboard backer uses sophisticated cobalt blue, butter yellow and vermilion with bold nostalgic retail typography. Exact readable packaging text only: 'THE OVERTHINKER', '47 IDEAS. ONE AFTERNOON.', 'COLLECT YOURSELF.', 'AGES 25+'. Brand-new fictional product design, playful adulthood self-recognition. Real clear-plastic reflections, crisp cardboard print, a single hanging-hole punch, art-directed soft product light and a subtle neutral background. Keep all accessory compartments visually separate and every word legible. No real toy brand, no copied franchise, barcode, manufacturer claim, price, watermark or AI label.

Lisbon travel portrait
Original AI-generated example.
See the exact prompt
Create one highly natural 4:5 travel street portrait of a completely fictional 38-year-old Latina woman with medium olive skin, shoulder-length dark wavy hair and a relaxed expressive smile. She is standing on a sunlit Lisbon-style cobblestone street with a softly blurred vintage yellow tram in the background, wearing an unbuttoned cobalt linen shirt over a cream tank and carrying a small canvas bag. She turns back toward a friend holding the camera, caught mid-laugh, a breeze lifting a few strands of hair. Three-quarter-length framing, slightly imperfect candid vacation-camera feel, warm afternoon side light, subtle analog grain, real skin texture and believable proportions. No beauty-filter smoothing, no generic fashion pose, no readable signage, travel-company branding, text, watermark or AI label. It should feel like the favorite unexpected photograph from a trip.
A few useful answers
Do I need a long prompt?
No. Specific information is more useful than length. State the subject, action, visual treatment and important constraints.
Can I reuse the full example prompts?
Yes. The example sections include the actual prompts used. Replace fictional details with the ones you want, and supply any required reference photos.


