The Navratri look that photographs best is not the loudest one. It is the traditional Gujarati chaniya choli in deep black and rust red, covered in mirror work and thread embroidery, with an oxidised coin belt at the waist and a tasselled dupatta falling from one shoulder. Against old stone pillars and marigold garlands, it reads as rich and rooted instead of costume-like — which is exactly why it keeps coming back every year. You do not need to own the outfit or find the location. Upload one clear selfie, paste a prompt below, and ChatGPT or Google Gemini rebuilds the whole scene around your own face.
About this look
Three things make this style work, and every prompt below keeps all three. First, the colour pair: black with rust or brick red, never bright pink or neon, because the darker base lets the mirror work catch the light. Second, the detail load sits on the top half — a flared kediyu-style choli with dense embroidery, oxidised jhumkas, stacked bangles and a coin belt — so a waist-up photo still looks complete on a phone screen. Third, the background is a warm stone courtyard with marigolds and brass lamps, kept fully out of focus so nothing competes with the outfit. Skip any of the three and the result drifts into a generic lehenga photo.
There are five versions below: the classic traditional portrait, a kediyu look for boys, a couple frame, a full garba night with dandiya, and a softer daylight one. Each is a complete prompt — copy it as it is, the face-lock line and the framing rules are already inside.
Prompt 1: Traditional black and rust chaniya choli for girls
The main look — mirror work, coin belt and a heritage courtyard behind you. Start here if you only make one photo this Navratri.
✦ AI PROMPTCopy → paste in ChatGPT / Gemini
Using my uploaded photo, keep my face, facial features, skin tone and hairstyle exactly the same — 100% identity match, no change to the face at all. Create a waist-up three-quarter portrait of me in a traditional Navratri chaniya choli in deep black and rust red, with dense mirror work and colourful thread embroidery on a flared kediyu-style choli, an embroidered dupatta with tassels draped from one shoulder, an oxidised silver coin belt at the waist, oxidised jhumkas and stacked bangles, hair tied back in a braid. Background: an old stone heritage courtyard with carved pillars, hanging marigold garlands and glowing brass lamps, kept fully out of focus. Lighting: warm golden lamp light with a soft rim light on the hair. Face and body turned toward the camera, whole head visible with space above the hair, no side profile. Finish it like a professional model shoot: 85mm lens look, shallow depth of field, magazine-quality retouch, polished natural skin. Vertical 4:5 format. No text, no watermark.
Results vary a little every time — run it 2–3 times and keep the best one.
Prompt 2: Kediyu and dhoti look for boys
The matching male version — mirror-work kediyu, printed dhoti and a bandhani turban, shot in the same courtyard light so a couple’s photos look like one set.
✦ AI PROMPTCopy → paste in ChatGPT / Gemini
Using my uploaded photo, keep my face, facial features, skin tone and hairstyle exactly the same — 100% identity match, no change to the face at all. Create a waist-up three-quarter portrait of me in a traditional Navratri kediyu — a short flared mirror-work jacket in black and rust red over a white kurta, with a printed dhoti, a bandhani turban with a tassel, a beaded mala and an oxidised kada. Background: an old stone heritage courtyard with carved pillars, marigold garlands and brass lamps, kept fully out of focus. Lighting: warm golden lamp light with a soft rim light. Face and body turned toward the camera, calm confident expression, whole head visible, no side profile. Finish it like a professional model shoot: 85mm lens look, shallow depth of field, magazine-quality retouch. Vertical 4:5 format. No text, no watermark.
Results vary a little every time — run it 2–3 times and keep the best one.
Prompt 3: Couple garba portrait
Upload one photo where both faces are clearly visible. The two outfits are colour-matched in the prompt, so the pair looks styled rather than accidental.
✦ AI PROMPTCopy → paste in ChatGPT / Gemini
Using my uploaded photo, keep both of our faces, facial features, skin tones and hairstyles exactly the same — 100% identity match, no change to either face. Create a waist-up three-quarter portrait of us as a couple in matching traditional Navratri outfits: she wears a black and rust chaniya choli with mirror work, a tasselled dupatta and an oxidised coin belt; he wears a mirror-work kediyu in the same colours with a printed dhoti and a bandhani turban. We stand close together facing the camera with warm relaxed smiles. Background: a stone heritage courtyard with marigold garlands and brass lamps, fully out of focus. Lighting: warm golden lamp light with a soft rim light. Whole heads visible, no side profile. Magazine-quality retouch, 85mm lens look, shallow depth of field. Vertical 4:5 format. No text, no watermark.
Results vary a little every time — run it 2–3 times and keep the best one.
Prompt 4: Garba night with dandiya
The energetic one — dandiya sticks raised near the face, string lights melting into warm bokeh behind you. This is the version that performs best on Reels.
✦ AI PROMPTCopy → paste in ChatGPT / Gemini
Using my uploaded photo, keep my face, facial features, skin tone and hairstyle exactly the same — 100% identity match, no change to the face at all. Create a knee-up three-quarter shot of me in a black and rust mirror-work chaniya choli with a tasselled dupatta and an oxidised coin belt, in an elegant mid-dance garba pose with both arms raised holding decorated dandiya sticks, bright joyful expression. Background: an open garba ground at night with colourful string lights and marigold garlands turned into soft warm bokeh, no other people. Lighting: warm festive night light with a colourful rim light on the hair and shoulders. Face and body turned toward the camera, whole head visible with space above the hair, no side profile. Magazine-quality retouch, 85mm lens look, shallow depth of field, rich cinematic colour grade. Vertical 4:5 format. No text, no watermark.
Results vary a little every time — run it 2–3 times and keep the best one.
Made with Prompt 4 — your photo will look different
Prompt 5: Soft daylight version
Same outfit, morning light, no drama. This suits a calmer grid, and it is the safest choice if your selfie is a little soft or was taken indoors.
✦ AI PROMPTCopy → paste in ChatGPT / Gemini
Using my uploaded photo, keep my face, facial features, skin tone and hairstyle exactly the same — 100% identity match, no change to the face at all. Create a waist-up three-quarter portrait of me in a traditional black and rust chaniya choli with mirror work, a tasselled dupatta and oxidised jewellery, standing in a sunlit stone courtyard in the morning. Background: pale stone walls, a few marigold garlands and soft shadows, kept gently out of focus. Lighting: soft natural daylight from the side, low contrast, airy and clean, no harsh shadows. Face and body turned toward the camera with a gentle smile, whole head visible, no side profile. Magazine-quality retouch, 85mm lens look, shallow depth of field, natural colour grade. Vertical 4:5 format. No text, no watermark.
Results vary a little every time — run it 2–3 times and keep the best one.
How to create it
Open ChatGPT or Google Gemini on your phone. Both are free and both accept photo uploads.
Tap the plus or photo icon and upload one bright, sharp, front-facing selfie. Daylight near a window works best.
Paste your chosen prompt in the same message as the photo and send.
If something is off, reply with one short correction such as “make the choli darker” or “keep my face exactly like the uploaded photo”. Do not paste the whole prompt again.
Download the result and post it in 4:5. For a matching set, change only the lighting line and run it again.
Tips for a better result
Tie your hair back in the reference selfie if you want the jhumkas to show — hair over the ears hides the jewellery the prompt asks for.
Keep the shot at waist-up. In a full-body result the mirror work turns into grey speckles on a phone screen.
If the embroidery looks flat, reply with “heavier mirror work catching the light” instead of editing the prompt.
Run each prompt two or three times. The pose and the folds change every run, and the third one is usually the best.
Post before the ninth night. Navratri searches peak in the first four days and drop sharply after Dussehra.
FAQs
Will it keep my real face?
Yes, if your selfie is bright, sharp and front-facing, and the face-lock line stays at the start of the prompt. A dark or blurry upload is the reason the face changes in almost every complaint we see.
Can I change the colours?
You can, but keep one dark base colour. Swapping black and rust for two bright colours removes the contrast that makes the mirror work visible.
Does this work for a group photo?
Up to two or three people, yes, and prompt 3 is written for that. Beyond that the faces get small and the model starts mixing them up, so shoot separately and post a carousel instead.
Is it free?
Yes. Every prompt here is free to copy and both apps have free tiers that accept photos. The only limit is how many images the free tier lets you make in a day.
More Navratri looks
For the dance-floor version see the Navratri Garba prompt, and for the Bengali side of the season there is the Durga Puja look. To build your own version of any of these, use the Prompt Builder, and the prompt guide explains why each line in these prompts is there.
Want a different look?Build your own prompt — pick the outfit, background and lighting, then copy it in one tap.
AI Photo Editing
AI Photo Editing
AI Photo Editing
AI Photo Editing
AI Photo Editing
AI Photo Editing