PromptVibe
All free tools
Free tool · works in your browser

AI Photo Prompt Builder

Choose what you want in the picture and get a clear, ready-to-paste prompt for ChatGPT or Gemini, with a short note on what every part of it does. The prompt is assembled on your phone from a fixed list of options. Nothing is uploaded here and no image is made on this page.

Build your prompt

Pick one option in each step. The prompt below updates as you go.

1. Goal

You upload a photo in the AI chat together with the prompt.

2. Who is in your photo?
3. Outfit
4. Scene
5. Light
6. Camera feel
7. Mood

The romantic mood appears only for couples.

8. Aspect ratio

Your prompt

Use the uploaded photo. Keep the face, facial features, skin tone, hairstyle and body shape of the woman exactly the same as in the photo. Do not change anyone's age, and do not smooth, slim or beautify. Outfit: change the clothing to a deep maroon silk saree with a thin gold border, neatly draped, fitted naturally to the body in the photo. Setting: place the woman on a home balcony at dusk, lined with lit clay diyas and marigold strings on the railing. Lighting: warm diya glow as the main light, gentle orange highlights on the face, blending naturally across skin and clothes. Camera: 85mm portrait lens look, shallow depth of field, softly blurred background, eyes in sharp focus. Mood: joyful and candid, natural smiles. Finish: photorealistic, natural skin texture, realistic hands and fingers, no text, no watermark, no logos. Format: vertical 4:5 aspect ratio.

Open ChatGPT(opens in a new tab)Open Gemini(opens in a new tab)

In the AI chat, attach your photo first, then paste the prompt. Only use photos of yourself or of people who agreed.

What each part does

  • Identity lock (Edit my own photo): The identity lock comes first so the tool changes the clothes and background, not the person. Without it, faces often drift towards a generic 'model' look.
  • Subject (Woman): Naming who is in the picture, including 'Indian', steers skin tone and features. Left out, many tools fall back on a generic default face.
  • Outfit (Saree): Fabric (silk), colour (maroon) and border detail give the tool a clear texture and palette. 'Neatly draped' cuts down on messy, melted-looking folds.
  • Scene (Diya-lit balcony): Clay diyas and marigolds make the festival setting clear through props alone, and 'dusk' leaves the sky dark enough for the lamps to glow.
  • Light (Warm diya glow): Naming the diyas as the main light ties the lighting to the scene, so the face is lit by the same warm lamps you see in the picture.
  • Camera feel (85mm portrait): This lens description gives a close, flattering crop and a blurred background, so the person stands out clearly.
  • Mood (Joyful): 'Candid' and 'natural smiles' push the tool away from stiff, posed expressions.
  • Finish (Realistic, no text): Asks for a realistic finish and names common problems (stray text, odd fingers) so the tool is less likely to produce them. It reduces these mistakes; it cannot rule them out.
  • Aspect ratio (4:5 portrait): 4:5 is a common portrait size for feed posts and gives more room above and below the person than a square.

How the parts of a photo prompt work

An image tool reads your prompt as a list of instructions, and each instruction controls one part of the picture. The subject decides who appears. The outfit and scene set the colours and the place. Light changes the mood more than most people expect: the same face looks warm and soft in golden-hour sun and sharp and modern under neon. The camera feel decides how close the frame is and how blurred the background looks, and the aspect ratio decides the shape of the final picture.

The builder writes each part on its own line with a label such as “Lighting:” or “Camera:”. Labels make it easier for you to find and change one part later, and they help the tool keep separate ideas apart. Specific words do more than general ones. “Maroon silk saree with a thin gold border” gives the tool a fabric, a colour and a detail, while “nice saree” leaves all three to chance.

Change one thing at a time

When a result is close but not right, resist the urge to rewrite the whole prompt. Change one line, run it again, and compare. If you change the light, the scene and the mood together and the new picture is better, you will not know which change helped, and you cannot repeat it next time.

A simple order works well: fix the face first (for photo edits, check the identity line is still at the top), then the outfit, then the scene, and leave light and mood for last. AI tools also give a slightly different picture every time, even with the same prompt, so try a prompt two or three times before you decide it does not work.

Editing your own photo versus creating a new one

“Edit my own photo” is for when you attach a picture in the AI chat. The prompt then opens with an instruction to keep the face, skin tone, hairstyle and body shape the same, because tools tend to smooth or reshape faces when they are asked to change the background. Even with this line, results vary, so look closely at the eyes and skin tone before you post. “Create a new image” needs no upload and describes an imaginary person instead.

Safety and consent

Only upload photos of yourself, or of people who have clearly agreed to have their picture edited. That includes friends in a group photo: ask each of them first. Do not use this kind of prompt to make pictures of real public figures, and keep photos of children to simple family scenes.

Before you upload, read the privacy settings of the AI tool you use, because each service handles uploaded photos under its own rules. When you share an edited picture, it is honest to say it was made with AI.

Want ready-made examples first? Browse the prompt guides or write a caption for your result with the caption tool.