AI Photo Prompts: 30 Ready-to-Use Prompts for an AI Photoshoot
AI photo prompts that work: how a strong photo prompt is built, plus 30 ready examples for photoshoots, portraits, avatars, photo editing and products.
AI photo prompts are text descriptions of a frame that a model uses to create a photo from scratch or rework your own picture: place you in a new location, change your outfit, the light or the era, or turn a selfie into a studio portrait. A good prompt reads like a brief for a photographer — who is in the frame, what they wear, where they stand, what the light is, which lens and what style — and when you work from your own photo, it gets one more line saying what to keep. In the iishki studio these prompts work without a VPN: Nano Banana Pro is the go-to for a photoshoot from a selfie, GPT Image 2.6 Sunburst handles a series of edits without losing the face, and Seedream 5.0 Pro and FLUX 3 build scenes from many references. Below are the prompt formula, 30 ready-to-use examples grouped by scenario and a table of common mistakes.
What makes a good photo prompt
A strong prompt is not long, it is ordered. Models read best when the main thing comes first and details follow. A working six-part formula:
- Subject Who is in the frame and what they are doing: "a woman around thirty laughing, covering her eyes with her hand".
- Outfit Clothing, fabric, accessories, hair: "oversized linen shirt, thin gold chain".
- Scene Location and the details around it: "wooden pier, gently rippling water, boats in the background".
- Light Direction, softness and temperature: "low sun behind her, rim light in the hair".
- Camera Shot size, angle, lens, depth of field: "waist-up, 85 mm, blurred background".
- Style Photorealism, film, editorial, black and white: "like a frame shot on colour film, soft grain".
If you upload your own photo, the prompt starts with a seventh part — the likeness anchor: "Use the person from the uploaded photo as the only reference for appearance; keep facial features, age, skin and eye colour". Without it the model happily "improves" the face beyond recognition.
What you don't need in a prompt: pixel sizes, aspect ratios and words like "4K, masterpiece, ultra-realistic". Aspect ratio and resolution are chosen in the settings under the prompt field, and empty adjectives are either ignored or turned into plastic skin.
Which model to choose for the job
| Task | Model | Why |
|---|---|---|
| Photoshoot and portrait from your own selfie | Nano Banana Pro | Strong photorealism, holds the face, up to 8 references, up to 4K |
| A series of edits to one frame without losing the face | GPT Image 2.6 Sunburst | Keeps the face and product steadier from edit to edit |
| A scene from several photos: person + outfit + background | Seedream 5.0 Pro | Up to 10 references, confident with art and cartoon styles |
| A photo from scratch, with no selfie, plus text in the frame | FLUX 3 | Photorealism, accurate text, up to 10 references, up to 4K |
| Lots of quick variations of one character | Nano Banana 2.1 | Follows the prompt more closely, stable character, up to 4K |
| A poster or product card with text | GPT Image 2.6 Flare | Clean layouts, headlines and feature callouts |
| A frame in the style of your own samples | Grok Image | Up to 3 references for style and composition |
| Raise the resolution of a finished shot | Topaz Photo Upscale | Up to 4× enlargement without redrawing the content |
If you don't know where to start, start with Nano Banana Pro: most prompts below were written for it and work in the other models unchanged. A detailed comparison is on the GPT Image 2 vs Nano Banana Pro page, and a roundup of photoshoot models is in the AI photoshoot tools guide.
How to make an AI photoshoot from your photo: step by step
1. Pick the source photo
You need a sharp portrait in daylight: no sunglasses or filters, the face large in the frame, head not tilted too much. A compressed copy from a messenger gives a soft, smeared result, so use the original from your gallery. If the picture is small or noisy, run it through Topaz Photo Upscale first.
2. Open the model and upload the photo
On the Nano Banana Pro card, add your photo as a reference. With several references, their order is the numbering: "the first photo is the face, the second is the dress".
3. Build the prompt with the formula
Likeness anchor, then subject, outfit, scene, light, camera, style. One scene per generation: don't ask for a beach, a party and an office in the same prompt.
4. Choose aspect ratio and resolution
For stories and messenger avatars use 9:16 or 1:1, for a profile portrait 4:5 or 3:4, for a cover 16:9. Try in 2K, then repeat the variant you like in 4K.
5. Fix one detail at a time
Don't rewrite the whole prompt because of one thing you dislike. Add a clarification ("softer light, no shine on the forehead") or upload the result as a new reference and fix only that.
30 ready-to-use AI photo prompts
Prompts marked "from photo" need your own picture; the rest work from scratch. Swap colours, places and clothes freely — the structure holds.
Portrait and avatar
1. Messenger avatar (from photo)
Use the person from the photo as the only reference for appearance; keep facial features and age. Close-up, slight smile, looking into the camera. Plain warm background the colour of baked milk. Soft light from a large window on the left. 85 mm lens, background slightly blurred. Natural skin, no retouching.
2. Business headshot for a CV (from photo)
Keep the face, hairstyle and age of the person in the photo. Waist-up portrait in a navy blazer and white T-shirt, arms relaxed. Light grey studio backdrop. Even diffused light, a faint shadow under the chin. Focus on the eyes, calm confident expression.
3. Black-and-white studio portrait (from photo)
Facial features from the photo unchanged. Black-and-white shoulder portrait, head turned slightly. One hard light source on the right, deep shadow across half the face. Dark background, film grain, high contrast.
4. Window portrait (from photo)
Keep the person's appearance. Sitting on a windowsill in a chunky grey sweater, holding a mug with both hands, looking out. Overcast daylight, soft shadows. A lifestyle-magazine frame with a warm grade.
5. Fine-art portrait from scratch
Portrait of an elderly craftsman in a carpentry workshop: grey beard, leather apron, a hand plane in his hands. A sunbeam from a small window, dust in the air. 50 mm lens, deep shadows, detailed hands and wrinkles. Documentary photography.
6. Avatar set in one grid (from photo)
Face from the photo unchanged. A 2×2 grid of four portraits: in a business suit on grey, in a hoodie on the street, in a sweater by a fireplace, in a shirt on a seaside terrace. Different light and expression in each frame, same appearance.
Location photoshoot
7. City walk (from photo)
Keep the face and build of the person in the photo. Walking down an old cobbled street in a beige trench coat with a paper cup of coffee. Early morning, soft side light, light haze. Full-length, eye level, blurred background.
8. Sea at sunset (from photo)
The photo is the only reference for appearance. Standing ankle-deep in the water in a light white dress, the wind lifting the hem. Sunset behind, warm rim light in the hair. Medium shot, 85 mm lens, warm tones.
9. Summer field (from photo)
Keep facial features and age. Sitting in tall grass in a linen shirt, holding a bunch of wildflowers. Sun low, golden hour, lens flare. Shot from grass level, soft focus in the foreground.
10. Winter forest (from photo)
Face from the photo unchanged. Standing among snowy fir trees in a long cream puffer coat and a knitted hat, snowflakes in the air. Soft cold light of an overcast day, visible breath. Medium shot, slightly blurred background. More winter scenes are in the AI winter photoshoot guide.
11. Café with rain outside (from photo)
Keep the person's appearance. Sitting by a café window, raindrops running down the glass, a cup and an open notebook on the table. Warm lamp light inside, cold blue outside. 35 mm lens, cinematic grade.
12. Night city rooftop (from photo)
Facial features and age as in the photo. Standing by a rooftop parapet in a leather jacket, city lights behind. Soft side light from a neon sign, bokeh. Waist-up, the mood of a movie still.
Style and era
13. 90s film snapshot (from photo)
Keep the person's face. A flash snapshot on a 90s point-and-shoot: standing by a rug hung on the wall in a denim jacket, slight red-eye, an orange date stamp in the corner. Warm faded colours, coarse grain.
14. 1950s Hollywood (from photo)
Appearance from the photo unchanged. Studio portrait in the style of 1950s film-star photography: finger waves, satin dress, pearls. Soft modelling light from above with a butterfly shadow under the nose. Black and white, velvety midtones.
15. 1970s postcard (from photo)
Keep facial features. The person by a fountain in a city park, 1970s clothes, a checked jacket. Colours like an old colour postcard: soft blue, warm yellow, slight blur at the edges.
16. Fashion magazine (from photo)
Face and build as in the photo. A fashion-magazine cover: full-length portrait in a voluminous red coat on white, a confident pose, a look over the shoulder. Hard studio light, a crisp shadow on the floor. No lettering.
17. Oil painting (from photo)
Keep the face recognisable. A Renaissance-style oil portrait: dark background, velvet cloak, a hand resting on a book. Soft light from the left, visible brushstrokes, craquelure on the canvas.
18. Anime character (from photo)
Redraw the person from the photo in Japanese animation style, keeping the hairstyle, hair colour and the features that make them recognisable. An evening street with lanterns, warm tones, clean lines, soft shadows.
Editing and retouching your own photo
19. Replace the background (from photo)
Replace the background with a bright Scandinavian living room with a large window on the right. Face, pose, clothing and hair unchanged. Cold daylight from the window. Add a soft shadow on the floor.
20. Change the outfit (from photo)
Dress the person in a dark green corduroy suit, fabric texture visible, relaxed fit. Don't touch the face, hair, pose or background; light on the suit matches the light on the face.
21. Evening instead of day (from photo)
Turn this daytime frame into an evening one: sunset, long shadows, warm light from behind, lights on in the windows. People, buildings and composition stay the same; only the lighting and sky change.
22. Remove distractions (from photo)
Remove the people in the background and the bin by the wall. Rebuild the background with no trace of editing; everything else as in the original.
23. Restore an old photo (from photo)
Restore the old photograph: remove scratches, stains and creases, bring back sharpness to the faces. Don't change features, clothing or composition; keep a light film grain. Natural colour, no oversaturation.
24. Studio-style relighting (from photo)
Relight the portrait: soft key light front-left, a weak fill on the right, a thin rim light along the hair. Face, expression and clothing unchanged, background half a stop darker.
If the finished shot lacks sharpness, raise the resolution in Topaz Photo Upscale; more on that in how to improve photo quality with AI. For editing-only prompts, see the Nano Banana Pro photo editing guide.
Couples and scenes from several photos
25. Couple portrait (from two photos)
The first photo is the man, the second is the woman; keep both faces unchanged. They walk side by side through an autumn park, holding hands and laughing. Soft diffused light, leaves underfoot, medium shot, 50 mm lens.
26. Person + outfit + place (from three photos)
The first photo is the person; keep the face and build. The second gives only the cut and colour of the coat. The third gives the place and lighting. Build a photorealistic frame: the person in that coat stands in that place, with light from the third photo falling on the face and fabric.
27. With a pet (from two photos)
The first photo is the person, the second is their dog; keep both looking like themselves. They sit together on the porch of a wooden house, the dog resting its head on their knees. Warm evening light, cosy mood, medium shot.
Products
28. Product on a clean background (from photo)
Place the sneakers from the photo on a light grey background with a soft shadow beneath. Shape, colour, laces and logo exactly as in the original, no new details. Even studio light.
29. Product in an interior (from photo)
The same candle in a glass jar as in the photo, standing on a wooden table by a window, with dried herbs and a linen napkin beside it. Morning light, soft shadows, label unchanged.
30. Ad banner with text
A coffee shop ad banner: a cappuccino on a light table, a croissant beside it. A large headline at the top, "COFFEE TO GO", and a small line at the bottom, "every day from 8 am". Clean grid, warm beige colours.
For banners with text, use GPT Image 2.6 Flare or FLUX 3 — they render lettering more cleanly. Marketplace product cards are covered separately in product photos for marketplaces with AI.
Common mistakes and how to fix them
| Mistake | Cause | What to do |
|---|---|---|
| The person doesn't look like you | No likeness anchor, small or filtered source | Start with "keep facial features, age, skin tone", upload a sharp original |
| Skin looks plastic, doll-like | The prompt says "perfect skin, ultra-realistic, 8K" | Drop the adjectives, write "natural skin, visible pores, no retouching" |
| The person looks pasted onto the background | Light and shadows aren't described | Say where the light comes from and what it is, ask for a shadow on the floor or wall |
| The model mixed several scenes | Three locations and five actions in one prompt | One scene and one main accent per generation |
| Hands and fingers are broken | A complex pose plus many edits at once | Simplify the pose, fix one detail at a time, finish with "natural hands, five fingers" |
| Wrong aspect ratio | Ratios in the text are ignored | Pick the ratio in the settings under the prompt |
| Text has typos | The model is weak at lettering | Put the text in quotes, use GPT Image 2.6 Flare or FLUX 3 |
| The outfit came with someone else's face | The reference has no assigned role | "From the second photo, only the cut and colour of the clothes" |
What's next
These 30 prompts cover an avatar, a photoshoot in any location, a retro look, photo edits and a product card. Start with Nano Banana Pro, and if you want the finished portrait to move, continue with how to animate a photo with AI. A roundup of editing models is in the AI photo editing guide. The generation price is shown on each model card, and you pay only for the image itself.
Don't want to fiddle with prompts?
Ready-made scenes: upload your photo and get the result in one click.
Models in this article
Frequently asked questions
How do I write a good prompt for an AI photo?
Describe the frame the way a photographer would plan it: who is in the shot, what they wear, where they stand, where the light comes from, which lens it was shot on and in what style. If you upload your own photo, add a separate line saying what to keep: face, features, age, skin tone. Set aspect ratio and resolution in the settings, not in the text.
Which AI model is best for a photoshoot from my own selfie?
For photorealism and precise edits from a selfie, start with Nano Banana Pro: it holds the face and accepts up to 8 references. If you edit the same frame several times and worry the face will drift, use GPT Image 2.6 Sunburst. For scenes built from many references, Seedream 5.0 Pro and FLUX 3 work well.
Can I write prompts in languages other than English?
Yes. In the iishki studio prompts work in Russian as well as English, so every example here can be used as is. Only text that must appear on the image should be written exactly as you want it, in quotation marks.
Why does the AI change my face?
The model redraws anything you did not ask it to keep. Add 'keep the face, features and age unchanged', upload a sharp photo where the face takes up a good part of the frame, and change one thing per generation: background, outfit or light.
Can I make an AI photoshoot for free?
Signing up, uploading photos and copying prompts is free; you only pay for the generation itself, in tokens. The price of an image is shown on each model's card, and no subscription is required.
Every model in this roundup is available in the IIshki studio at reduced club prices. Club members get tokens for generations every month.
Join the clubMore articles
- Nano Banana Pro reviews: what it actually doesNano Banana Pro reviews: what holds up — photorealism, text in images, face consistency, edits from a photo — and where the model still slips. Try it online.
- How to Use GPT Image 2.6: Text, Edits, GIF Stickers and SVGHow to use GPT Image 2.6 by OpenAI: which version to pick, text in images, edits from a reference, animated GIF stickers and SVG vectors, plus common mistakes.
- AI Greeting Card and Video Greeting: New Year and Birthday GuideHow to make an AI greeting card and a video greeting for New Year or a birthday: a card with readable text, a talking photo, a festive clip, prompts.