Kerala retro AI photo prompts turn one ordinary phone photo into a 1968 hand-tinted studio portrait, a 1984 monsoon frame or a 1987 hand-painted film poster. You upload the photo to ChatGPT, paste one prompt, and wait about a minute. No editing skill, no app, no subscription beyond ChatGPT itself.

This is the trend running through Kerala Instagram right now — nostalgia photos of parents rendered young, families staged like an old studio sitting, couples under a monsoon verandah. Most attempts come out looking like a modern photo wearing a costume. The twelve prompts below are written to avoid that, and this page explains exactly why they work.

Everything here is copy-paste. Nothing needs editing except the reference blocks for couple and family shots.

On this page

  1. What the Kerala retro trend actually is
  2. What you need before you start
  3. The three-step method
  4. Pick your look
  5. The one rule that makes or breaks it
  6. Prompts for men
  7. Prompts for women
  8. Prompts for couples
  9. Prompts for family
  10. How to make yours look different
  11. Troubleshooting
  12. FAQ

What the Kerala retro trend actually is

People are feeding a normal photo into ChatGPT and asking for it back as if it had been made in Kerala between 1968 and 1990 — the hand-tinted studio print, the newspaper clipping, the Onam greeting card, the painted film poster outside a single-screen theatre.

It spreads because it is not a filter. It rebuilds the picture: the clothes change, the hair changes, the light changes, the paper the print sits on changes. A filter cannot do that. A well-written prompt can.

Most attempts fail for one reason. People describe a mood — “vintage, nostalgic, old Kerala vibes” — and the model returns a 2026 photo with a sepia wash. What works is describing the process: the film stock, the lens, the printing method, the damage on the paper. Say how the picture was physically made and the period arrives on its own.

What you need before you start

  • ChatGPT. The free tier works; the paid tier is faster and refuses less often. These prompts are written for GPT Image, the image model inside ChatGPT.
  • One clear photo. Face lit, eyes visible, not a group shot cropped down. A blurry source photo stays blurry in the output.
  • Two minutes. That is the whole job.

The three-step method

Step 1 — upload the photo. One photo for a single person. Two photos for a couple, or one photo containing both. Up to four photos for a family.

Step 2 — paste the prompt. For men and women, paste the concept prompt on its own, nothing before it. For couples and families, paste the matching reference block first, then the concept prompt underneath it in the same message.

Step 3 — ask for a change instead of starting over. If the face drifted or the mood is off, reply in the same chat with one short instruction — “keep everything, make the rain heavier” — rather than pasting the prompt again. Re-pasting resets the likeness.

Pick your look

Six concepts, and what each is good for. The men’s and women’s versions of the same number are the same shoot, so M1 and W1 sit together as a pair.

Look Best for Prompt
1968 hand-tinted studio Parents, grandparents, anything that should feel like a real family heirloom. Holds the face better than any other concept. M1 · W1 · C1 · F1
1984 monsoon Cinematic, moody, one strong image. The best performer on Instagram. M5 · W5
1987 film poster The loudest. Painted, oversaturated, invented Malayalam title. Also the most copied. M8 · W8
1988 studio glamour Coloured gels and a soft-focus filter. Almost nobody is running this one. M9 · W9
1990 Onam card Festival posts, greetings, anything warm and family-facing. C5 · F5
1968 family sitting Group photos. Formal, seated elders, nobody smiling — exactly as it was done. F1

Not sure? Start with M1 or W1. The monochrome base absorbs small mistakes in the face, so it is the most forgiving prompt in the set.

The one rule that makes or breaks it

When you upload two or more photos, ChatGPT will average the faces into one invented person unless you stop it. This is the single most common failure in the whole trend, and one paragraph fixes it.

Paste the matching block above the concept prompt, in the same message. Single-person prompts need no block at all.

Block A — two separate photos, one couple

Snippet
Use the two uploaded photographs as identity references only. The first image is the man, the second is the woman. Each person keeps the facial structure, skin tone, age and proportions of their own reference image alone. Do not blend the two faces, do not transfer features between them, and do not average them into a single appearance. He stands or sits on the left of the frame, she on the right, their faces clearly separated with no overlap.

Block B — one group photo, same group restyled

Snippet
Use the uploaded group photograph as the identity reference for every person in it. Each face keeps its own facial structure, skin tone, age and proportions. Preserve the exact number of people and their relative ages — add no one, remove no one, and do not merge any two faces. Keep each person in roughly the same position within the group as in the reference.

Block C — several individual photos, one family

Snippet
Use the uploaded photographs as identity references only, one person per image, in this order: Image 1 is the father, Image 2 is the mother, Image 3 is the son, Image 4 is the daughter. Each face is sourced strictly from its own image and no other. Do not blend features across images. Arrange them in the order listed from left to right, faces clearly separated with no overlap.

Edit those roles to match what you actually uploaded. Four people is the reliable ceiling. Five is a coin flip. Six or more will not hold — generate in two groups and join them afterwards.

Block E — make them young (optional)

The one most people actually want: parents’ current photos, rendered at twenty-five.

Snippet
Render the subjects at approximately 25 years of age while keeping their facial structure and identity unchanged — smooth the skin to that age, restore full dark hair, and remove age lines, while the bone structure, features and skin tone stay exactly as in the reference.

Change the number to suit. It works reliably down to about twenty, and it works on single-person prompts too.

Prompts for men

Upload one photo. Paste the prompt alone. Add Block E underneath if you want the age changed.

M1 · 1968 hand-tinted studio portrait

Snippet
Create a hand-tinted black-and-white studio portrait from 1968, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so he stays clearly recognisable. Restyle his hair short, oiled and neatly side-parted with sideburns to mid-ear, and give him a thin trimmed moustache. Dress him in a white full-sleeve shirt buttoned to the collar, worn over a folded white mundu with a fine gold border, and leather sandals. Pose him seated on a carved wooden chair, both hands resting flat on his knees, upper body square to camera, expression still and unsmiling. Set a painted studio backdrop behind — a soft grey wash suggesting clouds and a distant balustrade. Shoot as a large-format black-and-white negative contact-printed onto matte fibre paper, then hand-coloured with translucent photographic oils: the base image stays monochrome and finely detailed while colour is brushed on selectively — pink flushed into the cheeks and lips, the mundu border filled gold, the chair washed a warm brown, a green touch on the painted foliage of the backdrop. The tinting sits slightly outside the lines in places and pools unevenly, leaving the greys visible beneath. Compose vertically at 2:3. The print shows a white deckled border, foxing spots at one corner, a soft crease across the lower edge and the warm ivory tone of aged paper.

M5 · 1984 monsoon

Snippet
Render this as a 1984 Kerala monsoon portrait, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so he stays clearly recognisable. Restyle his hair long and soaked, pushed back off the forehead with water running down his temples, and give him a full moustache beaded with rain. Dress him in a white cotton shirt gone translucent and clinging at the shoulders, sleeves rolled and dripping, worn over a folded mundu, a cloth bag held over his head against the downpour. Pose him mid-stride crossing a flooded courtyard, one foot in standing water throwing a splash, head down and turned slightly from the rain. Behind him a tiled-roof house with water sheeting off the eaves, a laterite wall, coconut palms thrashing in the wind, everything past ten metres lost in grey. Shoot on 35mm colour negative pushed one stop, 35mm lens at f/2, storm-cloud afternoon with a warm bulb glowing from the house doorway: rain streaks lit as bright lines against the dark treeline, his wet shoulders rim-lit, the frame cool blue-grey with one tungsten warm patch, coarse pushed grain, lifted blacks, motion blur in the falling water. Compose vertically at 2:3 at chest height. Skin keeps pores, running water and the strain of moving fast.

M8 · 1987 film poster

Snippet
Paint this as a hand-painted Malayalam film poster from 1987, using the uploaded photo strictly as an identity reference for facial structure, skin tone, age and proportions so he stays clearly recognisable beneath the brushwork. Restyle his hair thick, blow-dried and side-swept with volume at the crown, and give him a heavy shaped moustache. Dress him in an open-collar cream shirt with the top three buttons undone, sleeves rolled above the elbow, a gold chain against the chest, dark flared trousers. Pose him mid-turn with his chin lifted and jaw set, one hand pushing back his hair, shoulders angled to camera — the hero stance of the period. Behind him render a painted composite background: a burning orange sky bleeding into a coconut treeline, a smaller inset portrait of his face floating in the upper corner, and a wash of deep blue at the base. Execute it as gouache and enamel on board, then reproduced by four-colour screen printing on cheap poster paper: visible halftone rosettes, colour plates misregistered by a millimetre so cyan ghosts along his shoulder, flat oversaturated reds and yellows, brush texture surviving in the skin tones, hard black outlining around the figure. Place a bold hand-lettered Malayalam title in thick red brush script across the lower third with a yellow drop shadow, a strip of smaller cast lettering beneath it, and a small rectangular censor certificate mark in the bottom corner. Compose vertically at 2:3. Paper shows fold creases and a torn upper edge.

M9 · 1988 studio glamour

Snippet
Produce a 1988 photo-studio portrait in the glamour style, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so he stays clearly recognisable. Restyle his hair thick and blow-dried with height at the front, side-parted and feathered at the ears, and give him a full groomed moustache. Dress him in a black polo-neck under a grey blazer with wide lapels and padded shoulders, sleeves pushed slightly up, a gold-toned watch on the visible wrist. Pose him seated on a stool leaning forward, elbows on knees, hands loosely clasped, head turned to camera with a level unsmiling expression. Set him against a plain dark grey studio backdrop lit with a soft key from camera-left, a blue gel skimming the right side of his face and an amber gel behind him separating his shoulder from the background. Shoot on medium format, 105mm lens at f/4, ISO 100 colour negative, with a diffusion filter fitted: highlights bloom softly, blacks lift into haze, overall softness with focus held only at the eyes, fine grain, strong vignette. Compose vertically at 2:3 from mid-thigh up, camera slightly below eye level so he reads as tall. Skin retains pores, stubble shadow and natural shine beneath the diffusion. A small embossed studio name and a hand-written negative number sit in the lower corner of the print.

Prompts for women

Same rules. One photo, prompt alone, Block E underneath if you want the age changed.

W1 · 1968 hand-tinted studio portrait

Snippet
Create a hand-tinted black-and-white studio portrait from 1968, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so she stays clearly recognisable. Restyle her hair oiled and centre-parted, drawn back into a low bun with a jasmine strand pinned at the nape, no loose face-framing pieces. Dress her in a cream kasavu saree with a fine gold border worn over a plain elbow-sleeve blouse, pallu draped across the chest and tucked at the waist, a thin gold chain, small studs, glass bangles. Pose her seated on a carved wooden chair, hands folded in her lap, upper body square to camera, expression still and unsmiling. Set a painted studio backdrop behind — a soft grey wash suggesting clouds and a distant balustrade. Shoot as a large-format black-and-white negative contact-printed onto matte fibre paper, then hand-coloured with translucent photographic oils: the base image stays monochrome and finely detailed while colour is brushed on selectively — pink flushed into the cheeks and lips, the saree border filled gold, the blouse washed maroon, a green touch on the painted foliage of the backdrop. The tinting sits slightly outside the lines in places and pools unevenly, leaving the greys visible beneath. Compose vertically at 2:3. The print shows a white deckled border, foxing spots at one corner, a soft crease across the lower edge and the warm ivory tone of aged paper.

W5 · 1984 monsoon

Snippet
Photograph this as a 1984 Kerala monsoon portrait, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so she remains clearly the same person. Restyle her hair long, centre-parted and soaked through, clinging in dark ropes at the temples and shoulders with water running from the ends. Dress her in a cotton saree in deep indigo with a thin gold border, the fabric wet and heavy against her, pallu pulled over her head and held closed at the throat with one hand, a black umbrella collapsed in the other. Pose her standing under the overhang of a tiled-roof verandah, half in the rain, looking out to the left of frame rather than at the camera. Behind her the red laterite wall, a wooden pillar, a courtyard dissolving into grey rain, banana leaves bending under the weight of water. Shoot on 35mm colour negative pushed one stop, 50mm lens at f/2, late afternoon under storm cloud with a single warm bulb burning inside the house behind her: rain streaks catch the backlight as bright lines, her wet skin and hair rim-lit, the foreground cool blue-grey against that one pocket of tungsten warmth, heavy grain from the push, lifted blacks, soft focus falloff. Compose vertically at 2:3 at eye level. Skin shows real texture, water beading at the collarbone, no smoothing.

W8 · 1987 film poster

Snippet
Render this as a hand-painted Malayalam film poster from 1987, treating the uploaded photo purely as an identity reference for facial structure, skin tone, age and proportions so she remains clearly recognisable under the paint. Restyle her hair into a large voluminous blow-out with soft curls falling past the shoulders, centre-parted, a single jasmine strand at one side. Dress her in a chiffon saree in bright turquoise with a gold border, worn over a fitted short-sleeve blouse, gold jhumkas and stacked bangles. Pose her looking back over one shoulder with her pallu lifted by wind, lips parted, eyes wide and directed at the viewer. Behind her paint a composite backdrop: a hard magenta-to-orange gradient sky, a rain-streaked window motif at one edge, and a smaller inset of her face in profile floating upper-left. Execute as gouache and enamel on board reproduced through four-colour screen printing on cheap poster stock: coarse halftone rosettes, plates misregistered so magenta ghosts past her jawline, flat oversaturated colour, visible brush texture through the skin, heavy black outlining, eyes and lips overworked with extra detail as poster painters did. Lay a bold hand-lettered Malayalam title in yellow brush script across the lower third with a red drop shadow, a band of smaller cast lettering beneath, and a small censor certificate rectangle in the corner. Compose vertically at 2:3. The paper carries fold creases, a scuffed corner and paste residue at the edges.

W9 · 1988 studio glamour

Snippet
Create a 1988 photo-studio glamour portrait, using the uploaded image strictly as an identity reference for facial structure, skin tone, age and proportions so she is clearly the same person. Restyle her hair into a big teased blow-out with heavy crown volume and loose curls, one side swept back, lifted at the edges by a studio fan. Dress her in a satin blouse in deep emerald with padded shoulders and a wide collar, worn under a draped chiffon saree pallu pinned high, gold jhumkas, a thin chain. Pose her at three-quarters to camera, chin down, one hand raised to touch her jaw, looking directly into the lens. Light her against a plain black studio backdrop with a single softened key from the left, a magenta gel raking her right cheek and a blue gel behind her shoulder throwing a coloured rim through the hair. Shoot on a medium-format body, 105mm portrait lens at f/4, ISO 100 daylight-balanced colour negative, with a diffusion filter over the lens: highlights bloom and smear outward, blacks lift into a milky haze, the whole frame slightly soft with only her eyes holding focus, gentle grain, heavy vignette pulling the corners to black. Compose vertically at 2:3, tight from the chest up. Skin keeps real pores and a natural sheen under the diffusion rather than being smoothed. A small embossed studio name sits in the lower right corner of the print.

Prompts for couples

Two separate photos: paste Block A first. One photo with both people in it: paste Block B first. Add Block E after the block if you want them younger.

C1 · 1968 hand-tinted studio — best for parents

Snippet
Create a hand-tinted black-and-white studio portrait of a married couple from 1968, both faces staying clearly recognisable. Restyle his hair short, oiled and side-parted with a thin moustache; restyle hers oiled and centre-parted, drawn back into a low bun with a jasmine strand. Dress him in a white full-sleeve shirt with a folded mundu; dress her in a cream kasavu saree with a fine gold border over a plain elbow-sleeve blouse, thin gold chain, small studs, glass bangles. Pose them in the formal manner of the period: he seated on a carved wooden chair with both hands resting on his knees, she standing at his right shoulder with one hand placed lightly on the chair back, neither touching the other, both square to camera with still unsmiling expressions. Set a painted studio backdrop behind — a soft grey wash suggesting clouds and a distant balustrade. Shoot as a large-format black-and-white negative contact-printed onto matte fibre paper, then hand-coloured with translucent photographic oils: the base stays monochrome and finely detailed while colour is brushed on selectively — pink flushed into both faces, her saree border filled gold, her blouse washed maroon, a green touch on the painted foliage. The tinting sits slightly outside the lines and pools unevenly, greys visible beneath. Compose vertically at 2:3. The print has a white deckled border, foxing at one corner and the warm ivory tone of aged paper.

C5 · 1990 Onam card — married couple

Snippet
Design this as a printed Onam greeting card from 1990 featuring a married couple, both faces clearly recognisable. Restyle his hair thick and side-swept with a full moustache; restyle hers into a soft blow-out with crown volume and a jasmine strand pinned at one side. Dress him in a cream kasavu shirt over a folded gold-bordered mundu; dress her in a cream kasavu set-mundu with a broad gold border, gold necklace, jhumkas and bangles. Pose them standing side by side beside a large circular pookkalam laid on a swept courtyard, he at the left and she at the right with a small gap between them, both with hands together in a namaskaram and warm open smiles directed at the camera, a brass lamp burning at the edge of the flowers. Behind them a tiled-roof verandah with wooden pillars and a hanging brass bell. Shoot on 35mm colour negative, 50mm at f/5.6, bright mid-morning: saturated flower colour, warm skin tones, mild grain. Then reproduce as offset printing on glossy card stock with visible halftone rosettes and slightly heavy magenta. Frame the photograph inside a printed gold foil border with floral corner motifs, and set a Malayalam Onam greeting in ornate gold script across the lower band with a small English line beneath. Compose vertically at 2:3. The card shows a soft centre-fold crease, a scuffed foil edge and slight yellowing at the margin.

Prompts for family

One group photo: paste Block B first. Separate individual photos: paste Block C first and edit the roles. Four people is the ceiling.

F1 · 1968 hand-tinted family sitting

Snippet
Create a hand-tinted black-and-white family portrait from 1968, every face staying clearly recognisable. Restyle each into period grooming: men with short oiled side-parted hair and thin moustaches, women with oiled centre-parted hair drawn into low buns with jasmine, children with oiled combed-flat hair. Dress the men in white full-sleeve shirts over folded mundus, the women in cream kasavu sarees with fine gold borders over plain elbow-sleeve blouses with thin gold chains, the children in stitched cotton shirts and frocks. Arrange them formally: the eldest seated on carved wooden chairs at the centre, the others standing behind in a single row, any small children seated cross-legged on the floor in front. Every person faces the camera squarely with hands at their sides or folded, no smiles, no one touching. Set a painted studio backdrop behind — a soft grey wash suggesting clouds and a balustrade. Shoot as a large-format black-and-white negative contact-printed onto matte fibre paper, then hand-coloured with translucent photographic oils: the base stays monochrome and finely detailed while colour is brushed selectively — pink flushed into each face, saree borders filled gold, blouses washed maroon and blue, a green touch on the painted foliage. Tinting sits slightly outside the lines and pools unevenly. Compose horizontally at 3:2, everyone in focus. The print has a white deckled border, foxing at one corner and the warm ivory tone of aged paper.

F5 · 1990 Onam courtyard

Snippet
Photograph this as a 1990 Onam family portrait in a Kerala courtyard, every face staying clearly recognisable. Restyle each into period styling: men with thick side-swept hair and full moustaches, women with soft blow-outs and crown volume with jasmine pinned at one side, children with combed-flat hair. Dress the men in cream kasavu shirts over folded gold-bordered mundus, the women in cream kasavu set-mundus with broad gold borders and gold necklaces, jhumkas and bangles, the children in small kasavu outfits. Arrange them in a loose row behind a large circular pookkalam laid on a swept courtyard, the eldest at the centre, children crouched at the front edge of the flowers, everyone facing camera with warm open smiles, a brass lamp burning at the near edge of the arrangement. Behind them a tiled-roof house verandah with wooden pillars, a hanging brass bell, coconut palms beyond. Shoot on 35mm colour negative, 35mm lens at f/8, bright mid-morning light with the courtyard in open sun: saturated flower colour, warm skin tones, deep shadow under the verandah roof, mild grain, slight highlight bloom where the sun hits the white kasavu. Compose horizontally at 3:2 from chest height, everyone in focus. Skin keeps texture and the shine of heat. The courtyard holds only brass vessels, a coir mat and the flowers.

How to make yours look different from everyone else’s

Telling the model to “make it unique” does nothing — it has no memory of anyone else’s run. Change one concrete thing in the prompt before you generate, and the picture stops matching the hundred others using the same text.

  • 1968 hand-tinted — swap the tint palette (pink/gold/maroon for blue/green/ochre), stand instead of sit, ask for heavier foxing, add an oval vignette mask.
  • 1984 monsoon — umbrella open, collapsed or gone; verandah, flooded street, bus stop or jetty; a passing headlight instead of the doorway bulb; a night version lit only from behind.
  • 1987 film poster — change the title colour and placement, the number of inset portraits, the sky gradient; frame it as romance, action or family drama; tear the left edge instead of the top.
  • 1988 studio glamour — change the gel pair (magenta/blue, amber/cyan, red/green), fan on or off, black or mottled grey backdrop, a profile with half the face in shadow, a square 6×6 crop.
  • 1990 Onam — morning or late afternoon light, a different pookkalam colour scheme, verandah or open yard, a sadhya leaf in frame, gold foil border or plain white card edge.

Troubleshooting

It looks like a modern photo with period props. The most common failure of all. Move the film-and-camera sentence earlier in the prompt, and cut the adjectives about mood to make room for more about process. Describing how the print was made beats describing how it feels, every time.

Two faces came out looking like the same person. You skipped the block, or the two source photos are lit very similarly. Re-run with Block A and add: the two people look distinctly different from each other.

A face landed on the wrong body. The left-to-right ordering in Block C did not take. Re-run describing each person by a visible trait as well as a position — Image 1, the older man in glasses, stands at the left.

The outer people in a family shot came out as strangers. You are past the ceiling. Split into two groups of three, generate separately, and join them afterwards.

Everyone is smiling in a 1968 portrait. ChatGPT defaults to smiles. Add: no one is smiling; all mouths are closed.

Block E did not change the ages. Move it to the very end of the message instead of the top. The last instruction wins more often than the first.

The Malayalam text is broken glyphs. Expected — GPT Image cannot render Malayalam script reliably. Either accept it as part of the aged-print look, or crop the lettering out and set real type afterwards.

The likeness drifted on a heavily stylised concept. The poster and glamour looks push hardest on the face. The hand-tinted concepts hold it best, because a monochrome base absorbs small errors into the print texture instead of exposing them.

FAQ

Do I need ChatGPT Plus? No. The free tier generates these. Plus is faster, refuses less, and lets you run more images in a day.

Will it work on Gemini, Grok or Midjourney? Partly. These are written for GPT Image, which reads long descriptive paragraphs well. Midjourney needs the same ideas compressed into a shorter line with parameters. The reference blocks are specific to ChatGPT’s multi-image upload.

Why do the prompts never say “no modern cars”? ChatGPT has no negative channel, so a phrase like that reads as content and actually raises the chance a modern car appears. Every exclusion here is written as a positive description of what is in the frame.

Why is every title and masthead invented? Naming a real film or magazine gets you either a refusal or a mangled logo. Invented lettering looks more authentic and always generates.

Can I use these for a client or a brand campaign? Yes. The same method — describe the process, not the mood — is what makes AI product photography and ad creative usable commercially. That is the approach behind my social media and content services.

Why 2:3 and 3:2 instead of 9:16? GPT Image only outputs portrait, landscape or square. For a Reel or Story, generate at 2:3 and add the padding afterwards rather than fighting the model for a taller frame.

Want this done for your brand, not just your family album?

These prompts are the fun end of the same skill. The working end is using AI images inside a real marketing system — festival campaigns, product shoots that never needed a studio, ad creative tested at volume, and content that actually gets found in search.

That is what I do from Kannur: SEO, content, social media and Meta & Google Ads for brands that want the output without hiring a team for it. Have a look at the work, or start here.