The AI Photo Prompt Writing Formula: A Complete Guide

AI PHOTO PROMPTSPROMPT WRITING FORMULA8 COPY-PASTE PROMPTS

Most people type one lazy line like 'make me look professional' and then wonder why the result looks plastic. The problem is rarely the tool. It is the missing detail. Once you learn a repeatable ai photo prompt formula, your results become far more predictable, and you stop wasting free image credits on random guesses.

Here the formula breaks into seven labelled lines: subject, outfit, setting, light, camera, style and avoid. Each line answers one question the AI would otherwise guess on its own. There are also eight complete examples, from a studio headshot to a rainy street portrait, so you can see the how to write ai photo prompts idea in action.

Everything here works in Gemini and ChatGPT. You attach your own photo, ask the tool to keep your face unchanged, and then let the formula handle the rest. Results do vary from try to try, so think of every prompt as a first draft that you adjust.

Copy-Paste Prompts: Ai Photo Prompt Formula (8 Ready Ideas)

SUBJECT: Use my uploaded photo, keep my face, skin tone and hairstyle exactly unchanged, [age range] [man/woman], head and shoulders
OUTFIT: [colour] [garment], simple and well fitted
SETTING: [place], softly blurred background
LIGHT: [light type] from the [direction]
CAMERA: 85mm lens, f/2, eye-level
STYLE: natural, realistic photograph, true-to-life colours
AVOID: extra fingers, warped face, text, watermark

SUBJECT: Attach my photo, keep my face unchanged, shoulders-up portrait, calm confident expression
OUTFIT: charcoal blazer over a plain white shirt
SETTING: seamless light grey studio backdrop
LIGHT: large soft window-style key light from the left, gentle fill from the right
CAMERA: 85mm lens, f/2.8, sharp eyes
STYLE: clean corporate portrait, neutral colour grade
AVOID: heavy skin smoothing, glossy forehead, extra shadows on neck

SUBJECT: Use my uploaded photo, keep my face and features exactly the same, half-body, relaxed natural smile
OUTFIT: mustard cotton kurta or a light linen shirt
SETTING: open park path with tall trees and far-away blurred greenery
LIGHT: low golden sunlight from behind, warm rim light on hair
CAMERA: 50mm lens, f/1.8, slight upward angle
STYLE: warm film-like colours, gentle grain
AVOID: over-bright sky, plastic skin, distorted hands

SUBJECT: Keep my face from the attached photo unchanged, walking pose, three-quarter body, holding a transparent umbrella
OUTFIT: dark olive raincoat, jeans, canvas shoes
SETTING: wet city lane at dusk, shop signs glowing, reflections on the road
LIGHT: cool blue ambient light mixed with warm shop-window glow
CAMERA: 35mm lens, f/2, shallow depth of field
STYLE: cinematic but realistic, muted teal and amber tones
AVOID: readable shop names, fake logos, crowd of faces

SUBJECT: Attach my photo, keep my face unchanged, seated at a desk, looking slightly toward the camera with a friendly smile
OUTFIT: rust-coloured sweater
SETTING: tidy wooden desk, laptop, small green plant, bookshelf softly out of focus
LIGHT: daylight from a side window, soft shadows
CAMERA: 50mm lens, f/2.2, chest-level angle
STYLE: warm lifestyle photograph, natural skin texture
AVOID: messy cables, extra screens, floating objects

SUBJECT: Use my uploaded photo, face and hairstyle unchanged, standing portrait, hands relaxed, gentle smile
OUTFIT: cream silk kurta with a thin gold border, or a pastel saree with a simple border
SETTING: courtyard with marigold garlands and a few clay lamps
LIGHT: warm lamp glow mixed with soft evening sky light
CAMERA: 70mm lens, f/2.5, centred composition
STYLE: rich warm colours, realistic fabric texture
AVOID: extra jewellery, blurry hands, text overlays

SUBJECT: Keep my face unchanged from the attached photo, tight portrait, serious thoughtful expression
OUTFIT: black turtleneck
SETTING: plain dark wall with a faint gradient
LIGHT: single hard side light creating a clear half-shadow on the face
CAMERA: 100mm lens, f/4, very sharp eyes
STYLE: black and white, high contrast, fine film grain
AVOID: colour tint, blown highlights, skin blur

SUBJECT: Close shot of a hand holding a [product, e.g. ceramic coffee cup], natural nails, relaxed fingers
OUTFIT: plain cream sleeve visible at the edge
SETTING: bright kitchen counter with a folded cloth and a small plant
LIGHT: soft morning window light from the right
CAMERA: 60mm macro style, f/3.5, top-down at 45 degrees
STYLE: clean lifestyle product photo, natural colours
AVOID: extra fingers, brand names, text on cup

Why a fixed structure beats a long paragraph

Image models read your words as a pile of clues. When those clues sit in one long sentence, some get ignored. When each clue has its own labelled line, nothing is skipped, and you can see at a glance which part is weak. Think of how a tailor takes measurements: shoulder, chest, length, each on its own line. Nobody hands the tailor a paragraph and hopes.

That is the real strength of a chatgpt photo prompt structure like this one. If the background looks wrong, you only edit SETTING. If the shadows look harsh, you only edit LIGHT. You are debugging, not rewriting from zero.

What goes inside each of the seven lines?

SUBJECT tells the tool who or what is in the frame, the pose, and the face-preservation instruction. OUTFIT and SETTING give context. LIGHT and CAMERA control the photographic feel, and STYLE sets the final colour mood. AVOID is your safety net.

Be specific with nouns and directions. 'Soft window light from the left' is much stronger than 'nice lighting'. 'Navy blazer' beats 'formal clothes'.

• SUBJECT: who, pose, expression, face unchanged
• OUTFIT: colour plus garment plus fit
• SETTING: place plus how blurred the background is
• LIGHT: type plus direction
• CAMERA: lens in millimetres, aperture, angle
• STYLE: one clear mood, not five
• AVOID: three or four real problems you have seen

Can I paste this ai photo prompt formula straight into Gemini or ChatGPT?

Open the tool, tap the attach icon and upload one clear, front-facing photo of yourself. Paste the prompt, fill the brackets, and send. If the first image is close but not perfect, reply with one change only, such as 'same image, but make the background lighter'.

With a gemini photo prompt template you may find it follows the AVOID line more strictly, while ChatGPT sometimes reads the CAMERA line more loosely. Test both with the same text and keep whichever suits you. Do not retype the whole prompt in the second tool; copy it from your notes so the comparison stays fair. And if the tool asks a question back, answer it briefly instead of starting a new chat, since the earlier image is still useful context.

How do I improve a prompt without starting over every time?

Change one line at a time. If you edit four lines together, you will never know which edit helped. Save the versions that work in a notes app, and name them like 'studio-grey-v2'.

Keep each prompt between forty-five and ninety words. Shorter prompts leave gaps, and much longer ones start to contradict themselves, for example 'soft light' in one line and 'dramatic shadows' in another. Keep a small notes file with three columns: the prompt, what you liked, and what you changed next. After a week you will have your own cheat sheet, built from your face and your phone camera, not someone else's.

• Use real lens numbers such as 35mm, 50mm, 85mm for believable perspective.
• Name one dominant colour for the outfit and one for the background.
• Say 'natural skin texture' if faces look too smooth.
• Ask for 'eyes sharp' when the face looks soft.

Three ways a beginner breaks the ai photo prompt formula

The first mistake is stacking styles: 'cinematic, vintage, anime, hyper-real'. The tool averages them and you get mush. Pick one. The second is forgetting the face line, so the person comes out looking like a stranger. A related slip is uploading a dark or side-angle selfie and then blaming the formula; the AI can only copy what it can clearly see.

The third is writing an AVOID line full of vague words such as 'bad quality'. Name the exact fault: 'extra fingers', 'warped teeth', 'text on clothing'.

How each line changes the picture: a small experiment

The best way to understand the formula is to run it as an experiment. Start with the Studio Headshot Example. Generate it once and save the result. Now change only the LIGHT line to 'hard direct light from the right, strong shadow on the left cheek' and generate again. The same person, outfit and background now look dramatic and serious. Change it to 'soft overcast light, no strong shadows' and the same picture feels gentle and flat.

Next, change only CAMERA. Replace 85mm with 24mm and add 'close to the subject'. Faces stretch slightly and the background gets bigger, which is why wide lenses are poor for portraits. Replace it with 135mm and the face looks compressed and calm. Then change only SETTING, from the grey backdrop to 'a bright office with a plant', and watch the mood move from studio to workplace.

After four or five runs you will know what each line controls in your favourite tool. This knowledge is worth more than a hundred copied prompts, because you can fix a bad picture in one precise edit.

Which line to fix when something looks wrong

A quick fault map saves a lot of time. If the face looks unlike you, fix SUBJECT: repeat the face-unchanged instruction and use a sharper source photo. If the clothes are the wrong colour or fit, fix OUTFIT with a colour and a garment name. If the background looks messy or has strange objects, fix SETTING with fewer items and 'softly blurred'.

If skin looks flat or shadows are harsh, fix LIGHT. If the head looks too big or the background too close, fix CAMERA with a longer lens. If the whole picture feels over-processed or cartoonish, fix STYLE and remove extra style words. If text, watermarks or extra fingers keep appearing, fix AVOID with those exact words.

This map also helps you ask for help. When you tell a friend or a chat tool 'the SETTING is fine but the LIGHT is wrong', everyone understands the problem immediately.

From 'make me look good' to a prompt that works

Suppose you start with 'Make me look good for my freelancing profile.' That tells the AI almost nothing. Walk it through the formula. SUBJECT: my photo, face unchanged, head and shoulders, friendly confident smile. OUTFIT: navy shirt with an open collar. SETTING: bright home office, blurred shelf behind. LIGHT: soft daylight from a window on the left. CAMERA: 85mm, f/2.2, eye level. STYLE: natural warm photograph. AVOID: plastic skin, text, watermark.

Notice that the new prompt is about forty words and each line answers a question a photographer would ask. If you also work as a consultant, swap the outfit to a blazer and the setting to a plain grey wall. If you teach, make it a cotton shirt and a window with soft light. The structure stays; only the details change.

One honest limit: even a perfect prompt will not produce a perfect picture every time. Models differ, versions change, and faces drift. Retry, compare against your original, and only post the picture that still looks like you.

Frequently Asked Questions

Final Thoughts

A good ai photo prompt formula is really just a habit: say who, what they wear, where they are, how it is lit, how it is shot, which mood you want, and what to avoid. Start with the template, test it with your own photo, and tweak one line at a time. Results will vary, and that is normal.

Pick two of the examples above, run the one-line-at-a-time experiment today, and notice which line changes the picture most. Once your pictures look clean, a few notes on negative prompts will help you build a sharper AVOID line.

Note: AI image and text results vary by tool, version and settings. Use only your own photos, avoid copying real people’s likeness, and follow the terms of the AI tool you use.