Nano Banana Pro + GPT Image 2.5 Sunburst - a fixed cast of 8 AI models for a fashion series
Build 8 recurring AI models once, then drop them into a new outfit and scene for every post
Tested by Nouamane BenbrahimDisclaimer: This is an independent creative test made with AI, for demonstration only. It isn't an ad for any brand shown or named, and isn't affiliated with, sponsored or endorsed by them. All trademarks belong to their owners
Why do AI fashion pages feel like a different stranger in every post?
I wanted a page where the same faces show up day after day, so people actually start to recognize them. So I locked in a cast of 8 models, a woman and a man from four backgrounds: white, Black, East Asian and South Asian. Then I started dropping them into weird scenes in runway looks: a man on an ice floe in a dark sea, another balancing on a needle of rock. This post is how I built the cast and everything that broke once I started reusing it.
Here's my exact process.
1. Build every model the same way
First I make one reference image per person, from a prompt alone. Every one of them gets the exact same setup: a plain white tee, black trousers and white sneakers with no logos, a light-grey paper backdrop, soft light, and the whole body in frame with lots of grey space above the head and below the feet. That way the only thing that changes between the 8 is the person, and the full figure survives a crop to 4:5.
My first round looked like this:
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly beautiful mid-20s white woman with fair skin and a warm-neutral undertone, a light dusting of freckles across the nose and cheeks, long honey-blonde hair falling past the shoulders in soft natural waves, light green-hazel almond-shaped eyes, full natural brows, defined high cheekbones, softly full lips, bare minimal makeup. THE VIBE: A clean model-agency character reference — calm, confident, effortlessly gorgeous. This is the identity anchor for a recurring fashion character, so she must read clearly, completely and attractively, head to toe. IDENTITY LOCK: The woman with long honey-blonde soft waves, fair warm-neutral skin with light freckles, green-hazel almond eyes and high cheekbones. Slim, tall, model proportions. THE SUBJECT: She stands full-length, weight shifted onto her right leg, left knee softly relaxed, one hand loosely resting at her thigh and the other relaxed at her side, shoulders turned a few degrees off square, chin level, gaze just off the lens with a faint relaxed half-smile. She wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real skin with subtle tonal variation and a natural fair-skin flush at the cheeks and nose, soft organic light; hair rendered as individual strands with a few flyaways. THE SETTING: A seamless light-grey paper studio backdrop curving into the floor, clean and uncluttered; no lighting equipment, stands or softboxes visible anywhere in frame. THE LIGHT: Large soft overhead-front softbox key slightly to camera-left with gentle fill, soft even lookbook light, a subtle soft contact shadow under her sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no text, no logos anywhere. THE MOOD: A high-end model agency comp card — polished, approachable, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
Pretty, but you'd forget her in two seconds. I said "redo the white guy and white girl they look too generic".
One more thing broke here. One of the first renders had a big softbox hanging at the top of the frame.
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly beautiful mid-20s Black woman with deep brown skin and a warm red-copper undertone, high sculpted cheekbones, a defined jawline, large dark-brown eyes, full lips, full natural brows, short natural tightly coiled black hair cropped close to the head, bare minimal makeup. THE VIBE: A clean model-agency character reference — calm, confident, effortlessly gorgeous. This is the identity anchor for a recurring fashion character, so she must read clearly, completely and attractively, head to toe. IDENTITY LOCK: The woman with deep brown skin and warm copper undertone, close-cropped natural black coils, high sculpted cheekbones and large dark-brown eyes. Slim, tall, model proportions with a long neck. THE SUBJECT: She stands full-length, weight shifted onto her left leg, right knee softly relaxed, one hand lightly touching the hem of her shirt and the other relaxed at her side, shoulders turned a few degrees off square, chin slightly lifted, gaze just off the lens with a calm, assured expression. She wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin; luminous deep skin with warm copper undertone depth, natural sheen and soft specular highlights on the cheekbones, nose bridge and brow, warmth in the shadows, never ashy or grey; soft organic light. THE SETTING: A seamless light-grey paper studio backdrop curving into the floor, clean and uncluttered. THE LIGHT: Large soft overhead-front softbox key slightly to camera-left with gentle fill, soft even lookbook light, a subtle soft contact shadow under her sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no text, no logos anywhere. THE MOOD: A high-end model agency comp card — polished, approachable, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
My prompt said the light came from "a softbox", and that's exactly what showed up. I can't prove that's why, but once I rewrote it as soft light "from a source far out of frame", plus "no lighting equipment anywhere in frame", it never came back.
2. Give every face one or two things you'd recognize anywhere
The fix for generic was asking for "a distinctive high-fashion editorial face, not a generic catalogue face", and then giving each person one or two marks that survive any outfit or scene. For her it's long copper waves and dense freckles.
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly beautiful mid-20s white woman with a distinctive high-fashion editorial face, not a generic catalogue face — pale porcelain skin with a cool-neutral undertone and a dense scatter of natural freckles across the nose, cheeks and shoulders; long thick deep copper-auburn hair falling past the shoulders in loose, slightly undone waves; wide-set pale grey-blue eyes with heavy natural lids; strong straight dark-auburn brows; sharp angular cheekbones with soft hollows beneath; a strong straight nose; a full mouth with a pronounced cupid's bow; a defined angular jaw. Bare minimal makeup. THE VIBE: A runway-agency character reference — the kind of face casting directors fight over: unusual, unforgettable, magnetic. This is the identity anchor for a recurring fashion character, so she must read clearly, completely and strikingly, head to toe. IDENTITY LOCK: The woman with long deep copper-auburn waves, pale porcelain skin with dense freckles, wide-set pale grey-blue eyes, strong straight brows and sharp angular cheekbones. Tall, slim, long-limbed runway proportions. THE SUBJECT: She stands full-length, weight shifted onto her right leg, left knee softly relaxed, one hand loosely resting at her thigh and the other relaxed at her side, shoulders turned a few degrees off square, chin level, gaze just off the lens with a cool, self-possessed expression, lips relaxed. She wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real freckled skin with subtle tonal variation and a faint natural flush at the nose and cheeks, soft organic light; hair rendered as individual strands with a few flyaways. THE SETTING: An empty seamless light-grey paper backdrop that fills the entire frame edge to edge and curves smoothly into the floor; only grey paper surrounds her, top to bottom. THE LIGHT: Soft diffused light from a source far out of frame above and slightly to camera-left, gentle fill, soft even lookbook light, a subtle soft contact shadow under her sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no equipment, no text, no logos anywhere. THE MOOD: A top-tier model agency comp card — polished, editorial, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
The rest of the cast got theirs too: a buzz cut and a small gap between her front teeth, a tiny gold nose stud, a cleft chin and a nose with a slight crook, a small silver hoop in one ear. I also swapped the smiles for a cool, self-possessed expression.
Same for the Korean woman. My note was "asian girl also looks a bit too generic make her more distinctive too".
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly beautiful mid-20s Korean woman with light skin and a golden-neutral undertone, a clear glassy complexion, sleek straight jet-black hair falling to mid-back with a centre part, dark-brown softly hooded almond eyes, soft straight brows, a delicate nose, naturally full rosy-neutral lips, bare minimal makeup. THE VIBE: A clean model-agency character reference — calm, confident, effortlessly gorgeous. This is the identity anchor for a recurring fashion character, so she must read clearly, completely and attractively, head to toe. IDENTITY LOCK: The woman with sleek straight jet-black mid-back hair with a centre part, light golden-neutral skin, dark-brown hooded almond eyes and soft straight brows. Slim, tall, model proportions. THE SUBJECT: She stands full-length, weight shifted onto her right leg, left foot slightly forward, one hand tucking a strand of hair behind her ear and the other relaxed at her side, shoulders turned a few degrees off square, gaze just off the lens with a serene, quietly confident expression. She wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real skin with subtle golden tonal variation and little rosy flush, soft organic light; hair rendered as individual glossy strands with a few flyaways, not a moulded helmet. THE SETTING: A seamless light-grey paper studio backdrop curving into the floor, clean and uncluttered; no lighting equipment, stands or softboxes visible anywhere in frame. THE LIGHT: Large soft overhead-front softbox key slightly to camera-left with gentle fill, soft even lookbook light, a subtle soft contact shadow under her sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no text, no logos anywhere. THE MOOD: A high-end model agency comp card — polished, approachable, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
She got a sharp blunt bob with a heavy fringe and a small beauty mark under her left eye.
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly beautiful mid-20s Korean woman with a distinctive high-fashion editorial face, not a generic catalogue face — luminous light skin with a cool-neutral undertone and a healthy glow; a sharp blunt jet-black bob cut precisely at the jawline with a heavy straight-across blunt fringe sitting just above the brows; long, narrow, upturned monolid eyes, dark brown and intense; strong straight dark brows; high, prominent cheekbones; a small refined nose; full, softly defined lips with a natural deep-rose tone; a delicate pointed chin; a small dark beauty mark just below the outer corner of the left eye. Minimal makeup. THE VIBE: A runway-agency character reference — the kind of face casting directors fight over: graphic, cool, unforgettable, and healthy and radiant. This is the identity anchor for a recurring fashion character, so she must read clearly, completely and strikingly, head to toe. IDENTITY LOCK: The woman with the sharp jaw-length jet-black blunt bob and heavy blunt fringe, long narrow upturned monolid eyes, high prominent cheekbones, a small beauty mark under the outer left eye, and luminous cool-neutral skin. Tall, slim, long-limbed runway proportions. THE SUBJECT: She stands full-length, weight shifted onto her left leg, right foot slightly forward and turned out, one hand lightly hooked at the waistband of her trousers and the other relaxed at her side, shoulders turned a few degrees off square, chin slightly lifted, gaze just off the lens with a cool, self-assured expression. She wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real luminous skin with subtle tonal variation and little rosy flush, soft organic light; the bob rendered as individual glossy strands with a crisp but natural edge, not a moulded helmet. THE SETTING: An empty seamless light-grey paper backdrop that fills the entire frame edge to edge and curves smoothly into the floor; only grey paper surrounds her, top to bottom. THE LIGHT: Soft diffused light from a source far out of frame above and slightly to camera-left, gentle fill, soft even lookbook light, a subtle soft contact shadow under her sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no equipment, no text, no logos anywhere. THE MOOD: A top-tier model agency comp card — polished, editorial, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
The white guy took three tries. The second one came back gaunt and I just said "the white guy you made is no good. another". The third one asked for a memorable face that's still healthy and handsome.
Exact prompt
Create a new photorealistic image. Preserve the identity of the subject exactly as described: a strikingly handsome late-20s white man with a distinctive, memorable editorial face — healthy, rested and attractive, not gaunt, not generic. Fair skin with a warm-neutral undertone and a light sun-kissed glow across the nose and cheekbones; chin-length tousled dark-blond hair with lighter sun-bleached ends, tucked loosely behind one ear; striking ice-blue eyes, slightly hooded; strong, slightly arched dark-blond brows; broad high cheekbones; a nose with a slight crook from an old break; a wide mouth with a full lower lip; a strong square jaw with a cleft chin and light golden stubble. THE VIBE: A runway-agency character reference — rugged-meets-refined, warm, magnetic, instantly recognisable. This is the identity anchor for a recurring menswear character, so he must read clearly, completely and attractively, head to toe. IDENTITY LOCK: The man with chin-length tousled dark-blond hair with sun-bleached ends, ice-blue hooded eyes, a slightly crooked nose, a square jaw with a cleft chin and light golden stubble. Tall, athletic swimmer's build with broad shoulders. THE SUBJECT: He stands full-length, weight shifted onto his right leg, left knee relaxed, left hand resting in his trouser pocket and right arm relaxed at his side, shoulders turned a few degrees off square, gaze just off the lens with a relaxed, faintly amused half-smile. He wears the plain white fitted crew-neck cotton t-shirt, the black straight-leg tailored trousers, and the plain white leather low-top sneakers with no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real healthy skin with subtle tonal variation and a natural warm flush, bright clear eyes with no dark under-eye circles, soft organic light; stubble rendered as individual short golden hairs with a soft irregular line; hair as individual strands with flyaways. THE SETTING: An empty seamless light-grey paper backdrop that fills the entire frame edge to edge and curves smoothly into the floor with no visible seam or horizon line; only grey paper surrounds him, top to bottom. THE LIGHT: Soft diffused light from a source far out of frame above and slightly to camera-left, gentle fill, soft even lookbook light, a subtle soft contact shadow under his sneakers on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no equipment, no text, no logos anywhere. THE MOOD: A top-tier model agency comp card — polished, editorial, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the sneakers sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
Once all 8 looked like real editorial models I said "perfect. save these and give them names". Those 8 images are the cast, and most of the scenes after this start from one of them.
3. Attach the outfit, don't describe it
Next I picked a runway look for one of the men and described it in words: the cropped black blazer, the fur, the necklace.
Exact prompt
Create a new photorealistic image. Preserve the man with shoulder-length straight jet-black centre-parted hair tucked behind his ears, narrow upturned monolid eyes, very sharp high cheekbones, a narrow V-shaped jaw, clean-shaven luminous skin and a small silver cartilage hoop on his right ear, from the first reference — his face, hair, skin and lean runway build exactly. Do NOT invent a different person. Do NOT carry over his white t-shirt, trousers or sneakers from the reference — he wears only the new outfit below. THE VIBE: A high-fashion menswear runway look — sculptural, luxurious, a little provocative. IDENTITY LOCK: Preserve the man from the first reference. Do NOT source face, hair, or skin from anywhere else. THE SUBJECT: He stands full-length, weight shifted onto his right leg, left knee relaxed, arms hanging loosely at his sides, shoulders squared to camera, chin level, gaze straight just past the lens with a cool, impassive runway expression. THE OUTFIT: No shirt underneath. A black wool single-breasted blazer, cropped at the hip, with a flap pocket, worn pushed off the shoulders and open at the chest so his collarbones and upper chest show. Wrapped around his shoulders and crossing over his chest is a huge, thick faux-fur stole in shaggy blond, cream and caramel-tipped long-pile fur, sitting off the shoulders like an oversized collar. A fine silver chain necklace of small irregular links with a long thin silver drop pendant hanging down the centre of his chest. Slim straight-leg black wool trousers with a sharp centre crease. Black polished leather ankle boots with a thin curved white piping line across each boot, no logos. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real luminous skin with subtle tonal variation; fur rendered as individual long glossy fibres with real volume; hair as individual strands. THE SETTING: An empty seamless light-grey paper backdrop that fills the entire frame edge to edge and curves smoothly into the floor with no visible seam or horizon line; only grey paper surrounds him, top to bottom. No lighting equipment anywhere in frame. THE LIGHT: Soft diffused light from a source far out of frame above and slightly to camera-left, gentle fill, soft even lookbook light, a subtle soft contact shadow under his boots on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no text, no logos anywhere. THE MOOD: A top-tier fashion-house lookbook — polished, editorial, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the boots sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
Close, but the fur sits like a scarf and the necklace is way simpler. My reaction: "nah i want the exact outfit".
So I attached the runway photo as a second reference and told it to take only the clothing, jewellery and shoes from it, and never the face, hair or body of the model wearing it. His face comes only from the cast image.
Exact prompt
Create a new photorealistic image. Preserve the man from the first reference — his face, shoulder-length straight jet-black centre-parted hair tucked behind his ears, narrow upturned monolid eyes, sharp cheekbones, clean-shaven luminous skin, the small silver cartilage hoop on his right ear, and his lean runway build — exactly. Dress him in the exact outfit from the second reference, copied faithfully piece for piece. Take ONLY the clothing, jewellery and footwear from the second reference — do NOT take the face, hair, skin or body of the model in the second reference. Do NOT carry over the white t-shirt, trousers or sneakers from the first reference. THE VIBE: A high-fashion menswear lookbook — sculptural, luxurious, a little provocative. IDENTITY LOCK: The face, hair and skin come only from the first reference. Do NOT source them from the second reference or anywhere else. THE OUTFIT: Exactly as in the second reference — the thick shaggy blond, cream and caramel-tipped faux-fur stole wrapped tightly off the shoulders like a cowl, crossing over the upper chest and leaving the collarbones and upper chest bare with no shirt underneath; the black wool single-breasted blazer sitting below the fur, cropped at the hip with a flap pocket; the silver necklace of irregular links with the long thin drop pendant down the centre of the chest; the slim black wool trousers with a sharp centre crease; the black leather ankle boots with the thin curved white piping line. Match the proportions, colours, textures and the way the fur wraps exactly. THE SUBJECT: He stands full-length, weight shifted onto his right leg, left knee relaxed, arms hanging loosely at his sides, shoulders squared to camera, chin level, gaze just past the lens with a cool, impassive runway expression. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, real luminous skin with subtle tonal variation; fur rendered as individual long glossy fibres with real volume; hair as individual strands. THE SETTING: An empty seamless light-grey paper backdrop that fills the entire frame edge to edge and curves smoothly into the floor with no visible seam or horizon line; only grey paper surrounds him, top to bottom. No lighting equipment anywhere in frame. THE LIGHT: Soft diffused light from a source far out of frame above and slightly to camera-left, gentle fill, soft even lookbook light, a subtle soft contact shadow under his boots on the grey floor; neutral colour grade. THE KEY DETAILS: No props, no text, no logos anywhere. THE MOOD: A top-tier fashion-house lookbook — polished, editorial, aspirational. THE COMPOSITION: Vertical 9:16 frame; the full body from the top of the head to the boots sits inside the central portion of the frame with generous grey space above the head and below the feet, so the whole figure stays intact when cropped to 4:5. Figure centred horizontally. THE CAMERA: Clean studio capture at eye level, full-length, deep focus, everything sharp, no background blur.
Now the fur wraps off the shoulder, the necklace has its drop, and the boots have the white piping.
4. Let the outfit pick the place
For the scene, I look at the outfit and put it somewhere it has no business being. My rule of thumb: if it was a green poofy outfit, I'd put him on a lily pad in a slow river. This one is black tailoring and white fur, so I went with an ice floe in a dark open sea.
Exact prompt
Create a new photorealistic image. Preserve the man from the first reference exactly — his face, shoulder-length straight jet-black centre-parted hair tucked behind his ears, narrow upturned monolid eyes, sharp cheekbones, clean-shaven luminous skin, the small silver cartilage hoop on his right ear, and his lean runway build. Preserve his exact outfit from the first reference, matched to the second reference: the thick shaggy cream and caramel-tipped faux-fur stole wrapped off the shoulders like a cowl, crossing the bare upper chest with no shirt; the cropped black wool single-breasted blazer with flap pocket beneath it; the silver irregular-link necklace with the long thin drop pendant; slim black creased wool trousers; black leather ankle boots with a thin curved white piping line. Take ONLY clothing details from the second reference — never its model's face, hair or skin. Do NOT place him in a studio. THE VIBE: A surreal high-fashion campaign still — impossible, calm, cinematic. A man dressed for a gala stands alone on a piece of ice that is far too small to hold him, adrift in the open ocean. IDENTITY LOCK: Face, hair and skin from the first reference only. THE SUBJECT: He stands full-length and perfectly composed on a tiny, flat, irregular ice floe barely wider than his two boots, weight shifted onto his right leg, arms relaxed at his sides, shoulders squared to camera, chin level, gaze just past the lens with a cool, impassive expression. The floe sits only a few centimetres above the waterline and does not sink. THE SETTING: Endless dark open sea in every direction, deep slate-blue and near-black water with low, slow rolling swells and small dark ripples lapping at the edges of the ice; a thin wet sheen and a few tiny ice shards around the floe. A distant flat horizon line two-thirds up the frame, an overcast heavy charcoal sky with soft cloud layers. No land, no boats, no other objects anywhere. THE LIGHT: Soft, cold, diffused overcast daylight from slightly camera-left; muted, desaturated cool palette with deep blacks; the cream and caramel fur is the warmest thing in the frame and glows softly against the dark water; true neutral skin tone, never blue-tinted; faint cool reflections of him on the water surface. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin; fur rendered as individual long glossy fibres; hair as individual strands with a few flyaways lifted by the sea breeze. THE KEY DETAILS: Translucent blue-white ice with real texture, air bubbles and a frosty top surface; realistic ocean water physics. No text, no logos anywhere. THE MOOD: A luxury fashion-house campaign — quiet, uncanny, unforgettable. THE COMPOSITION: Vertical 9:16 frame, figure centred; his full body from head to boots and the whole ice floe sit inside the central portion of the frame, with dark sky above his head and dark water below the floe, so everything stays intact when cropped to 4:5. Shot from slightly above water level at a gentle distance so the floe's smallness reads instantly. THE CAMERA: Shot on a full-frame camera with a 50mm lens, deep focus, everything sharp, cinematic film grain.
He fills almost the whole frame, so a feed crop would cut off his head or the ice. I wanted the framing to be the same on every post, so I stopped saying "further away" and gave it numbers: his whole figure, head to the bottom of the ice, takes about 40 percent of the frame height, with his head about 30 percent down from the top.
Exact prompt
Create a new photorealistic image. Recreate the scene from the first reference — the same man, the same outfit, the same tiny ice floe, the same dark open sea and overcast charcoal sky — but with the camera pulled much further back. Preserve the man exactly as in the first and second references: his face, shoulder-length straight jet-black centre-parted hair tucked behind his ears, narrow upturned monolid eyes, sharp cheekbones, clean-shaven skin, the small silver cartilage hoop on his right ear, his lean build. Preserve his exact outfit: the shaggy cream and caramel-tipped faux-fur stole wrapped off the shoulders like a cowl over the bare upper chest, the cropped black single-breasted blazer beneath, the silver irregular-link drop necklace, slim creased black trousers, black ankle boots with the curved white piping line. THE VIBE: A surreal high-fashion campaign still — impossible, calm, cinematic. A man dressed for a gala stands alone on a piece of ice far too small to hold him, adrift in open ocean. IDENTITY LOCK: Face, hair and skin from the references only. THE SUBJECT: He stands full-length and composed on the tiny flat irregular ice floe, barely wider than his two boots, weight on his right leg, arms relaxed at his sides, shoulders squared to camera, gaze just past the lens with a cool, impassive expression. The floe sits just above the waterline and does not sink. THE COMPOSITION — STRICT: Vertical 9:16 frame. The man is small in the frame: his whole figure from the top of his head to the bottom of the ice floe fills only about 40 percent of the frame height, centred horizontally. The top of his head sits about 30 percent of the way down from the top edge, with open overcast sky above him; the bottom edge of the ice floe sits about 30 percent of the way up from the bottom edge, with open dark water below it. Wide empty margins of sky and sea above and below, so the full figure and the whole floe stay comfortably inside a centred 4:5 crop with room to spare. The horizon line passes behind his chest. THE SETTING: Endless dark open sea in every direction, deep slate-blue and near-black water with low slow rolling swells; small dark ripples lap at the edges of the ice, with a wet sheen and a few tiny ice shards around the floe. A heavy charcoal overcast sky with soft cloud layers. No land, no boats, no other objects. THE LIGHT: The same soft, cold, diffused overcast daylight falls on him as on the sea — no studio light, no spotlight on the figure; he is lit by the sky alone, slightly brighter from camera-left. Muted desaturated cool palette, deep blacks; the cream and caramel fur is the warmest note in the frame; true neutral skin tone, never blue-tinted; a faint dark reflection of him on the water. THE FACE: Natural skin texture, no airbrushing; fur as individual long fibres; a few hair strands lifted by the sea breeze. THE KEY DETAILS: Translucent blue-white ice with real texture and a frosty top surface; realistic ocean water. No text, no logos anywhere. THE MOOD: A luxury fashion-house campaign — quiet, uncanny, unforgettable. THE CAMERA: Shot from a boat slightly above water level on a full-frame camera with a 35mm lens at a distance, deep focus, everything sharp, subtle cinematic film grain.
It didn't hit 40 percent exactly, he still fills more like half the frame. But there's room top and bottom now.
5. Relight him so he belongs there
Then I noticed it: "he kinda looks photoshopped into the scene why isn't the lighting more natural?" He's crisp and evenly lit like he's still in the studio, and the sea behind him is flat and grey.
So I rewrote only the light. He's lit by the same heavy overcast sky as the sea, with very low contrast and no crisp studio light. A cool grey fill bounces up off the water onto his jaw and the fur. Then I added everything the scene would do to him: a thin layer of sea haze, wind lifting his hair and the fur fibres the same way as the ripples, damp on his boots, water washing over the edge of the floe, and his reflection on the water.
Exact prompt
Create a new photorealistic image. Recreate the first reference exactly — the same man, same face, hair, cartilage hoop, same outfit (shaggy cream and caramel-tipped faux-fur cowl off the shoulders over the bare upper chest, cropped black blazer, silver drop necklace, creased black trousers, black boots with white piping), the same tiny ice floe, the same dark open sea, the same framing, camera distance, and position in the frame — but fix the lighting so he truly belongs in the scene. It must look like ONE photograph taken on location, never a studio portrait composited onto a sea background. THE VIBE: A real on-location fashion campaign photo shot on a cold, overcast day at sea — raw, atmospheric, believable. IDENTITY LOCK: The man from the first reference, unchanged. THE LIGHT — the key change: He is lit only by the same flat, heavy overcast sky that lights the sea. Soft, broad top-light from the cloud cover, very low contrast on his face and body, no crisp studio key light, no sharp edges, no glossy highlights. A cool blue-grey ambient fill wraps around him from the sky and is bounced up from the dark water onto the underside of his jaw, chest and the fur. Skin keeps its true natural tone but is slightly muted and cool in the shadows, with natural cold-weather flush on his cheeks, nose and knuckles. The fur's caramel tones are slightly desaturated by the grey day. His contrast, sharpness, black levels and colour grade exactly match the sea and sky around him. THE ATMOSPHERE: A thin layer of cold sea haze and mist sits over the water, softening the horizon and very slightly veiling him, consistent with the distance to the camera. Sea wind lifts loose strands of his hair across his face and ruffles the long fur fibres in the same direction as the ripples. Fine droplets and a damp sheen on his boots and the bottom of his trousers; small splashes of water washing over the edges of the floe around his boots. His soft, broken reflection and a dark shadow on the water are consistent with the overcast light. THE KEY DETAILS: Same lens perspective and depth for the man and the sea; matching film grain across the entire image. No text, no logos anywhere. THE MOOD: A luxury fashion-house campaign shot for real at sea — quiet, cold, uncanny, unforgettable. THE COMPOSITION: Exactly as in the first reference: vertical 9:16, figure centred, open overcast sky above his head and dark water below the floe, everything intact inside a centred 4:5 crop. THE CAMERA: Shot on location from a boat on a full-frame camera with a 35mm lens, natural light only, subtle cinematic film grain.
That one I called "looks great." The trade-off is the haze lifted his black suit to more of a charcoal, which I was fine with.
6. Skip the studio step
I tried the same relight trick on another guy standing on a river of lava, and it didn't work. I attached the frame, asked it to keep his framing and only rebuild the light, and it handed back almost the same picture.
Exact prompt
Create a new photorealistic image. This is a photograph taken on a river of molten lava at dusk from the start — not a composite. Rebuild the entire scene as if shot natively here, with unified lighting across subject and environment, shot from further back. THE CHARACTER ANCHOR: Preserve the man from the first reference — his face, chin-length tousled dark-blond hair with sun-bleached ends, ice-blue eyes, slightly crooked nose, cleft chin, light golden stubble, broad build — and his exact outfit: the burgundy satin robe-jacket with black and dusty-rose floral brocade, black brocade sleeve panels, quilted burgundy shawl collar, black satin sash, the brocade scarf ending in black fringe, white top at the neckline, layered silver necklaces, silver rings, a silver drop earring, wide flared burgundy satin trousers and chunky white sneakers. Do NOT source face, hair or skin from anywhere else. He is relit entirely by this scene's light; the reference's studio lighting is discarded. THE ITERATION ANCHOR: Carry forward from the second reference — the lava scene, the floating black basalt slab, the glowing orange underlight, the smoky dusk sky, the embers, the colour grade. THE NEW CHANGE: He is further from the camera and smaller in the frame, and the lava reads clearly as one wide slow-moving RIVER flowing toward and past the camera, with dark rock banks on both sides, the slab floating in the middle of the current. THE VIBE: A man in a decadent burgundy lounge robe stands perfectly at ease on a lone slab of black rock riding a slow river of glowing orange lava — impossible, serene, a little arrogant; a real photograph somebody risked their life to take. THE SUBJECT: He stands full-length on the slab, weight on his right leg, left knee relaxed, arms loose at his sides, shoulders squared to camera, chin level, gaze just past the lens with a cool, impassive, unbothered expression — mouth relaxed and closed, no smile. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin, a light sheen of heat on his skin. THE SETTING: A wide river of molten lava, bright orange-yellow with a crust of cooling black plates cracking and folding, glowing seams between them, flowing slowly through a dark volcanic valley. The flat solid slab of black basalt, barely larger than his two feet, floats in the middle of the current, its edges glowing faint red. Black jagged lava-rock banks on both sides, low smoke drifting across an ash-grey dusk sky. No people, no buildings, no other objects. THE LIGHT: The lava is THE key light — it lights him from below and in front: warm orange up the front of his trousers, the underside of the scarf fringe and sash, his jaw, chin and under his brows; deeper shadow above under the dusk sky, with a dim cool blue-grey sky ambient on his shoulders and hair. Orange specular highlights on the satin matching the lava; the white sneakers glow orange. His contrast, black level and grade match the scene. Heat shimmer just above the lava around the slab; the hot updraft lifts his hair and stirs the scarf fringe. No studio key. THE KEY DETAILS: Drifting embers and fine ash in the air, smoke wisps curling off the lava. No text, no logos anywhere. THE MOOD: A luxury fashion-house campaign — decadent, dangerous, completely calm; rich orange against near-black, the burgundy deep and saturated. THE COMPOSITION — STRICT: Vertical 9:16 frame, figure centred horizontally. He is small in the frame: the full figure from the top of his head to the bottom edge of the rock slab fills only about 40 percent of the frame height. The top of his head sits about 30 percent of the way down from the top edge, with dark smoky sky above; the bottom edge of the slab sits about 30 percent of the way up from the bottom edge, with a wide band of glowing lava river below it in the foreground. Everything stays inside a centred 4:5 crop with margin. THE CAMERA: Shot on location from slightly above the lava surface at a distance, 35mm lens, natural light only, deep focus, subtle film grain.
Exact prompt
Create a new photorealistic image. This is a photograph taken at dusk on a river of molten lava from the start — not a composite. Rebuild the entire scene as if shot natively here, with one unified light across the man and the lava. It must look like a single real on-location photograph, never a lit studio portrait placed into a lava scene. THE VIBE: A man in a decadent burgundy lounge robe rides a lone slab of black rock down the middle of a slow river of molten lava at dusk — impossible, serene, a little arrogant. Dark, hot, dangerous, and completely calm. IDENTITY LOCK: Preserve the man from the first reference exactly — his face, chin-length tousled dark-blond hair with sun-bleached ends, ice-blue eyes, slightly crooked nose, cleft chin, light golden stubble, broad build — and his exact outfit: the burgundy satin robe-jacket with black and dusty-rose floral brocade, black brocade sleeve panels, quilted burgundy shawl collar, black satin sash, the brocade scarf ending in black fringe, white top at the neckline, layered silver necklaces, silver rings, silver drop earring, wide flared burgundy satin trousers, chunky white sneakers. Keep his framing and size in frame. Rebuild ONLY the light on him and the lava river. THE SUBJECT: He stands full-length on the slab, weight on his right leg, arms loose at his sides, shoulders squared, chin level, gaze just past the lens with a cool, impassive expression, mouth relaxed and closed. THE SETTING: One single, continuous river of molten lava flowing slowly toward and past the camera, unbroken from bank to bank and filling the whole lower part of the frame and the foreground — a glowing orange-yellow molten surface with thin dark cooling skin wrinkling and folding on top of it, flowing like a thick slow river with ripples and folds around the slab. No rock islands, no separate streams, no solid ground anywhere near him except his one small flat slab of black basalt, which floats in the middle of the current, its underside and edges glowing red-orange. Dark jagged lava-rock banks only at the far left and right edges and in the distance. Thick smoke and steam rising off the river; a dark ash-grey dusk sky with low smoke. THE LIGHT: Exposed for the glowing lava. The lava river is the one key light and the brightest thing in the frame; there is no other light on him. He sits about two stops under the lava: lit only from below by the lava's hard orange glow — strong on the fronts and insides of his trousers, the white sneakers glowing orange, the underside of the sash and the scarf fringe, the underside of his jaw and chin, his nostrils and under his brows. Above that the light falls off fast: his chest, shoulders and the upper robe sit in deep warm shadow, his face mostly dim, his hair a dark silhouette against the grey sky catching only a thin orange rim on the stray strands. The dusk sky gives only a faint cool blue-grey on the tops of his shoulders and head. Orange light bleeds into the burgundy satin as hot specular streaks along the folds; blacks are crushed. His contrast, exposure and grade are exactly those of the lava scene around him. THE ATMOSPHERE: Heat shimmer rising from the river visibly distorts the air around his lower legs and slightly warps the background behind him; drifts of smoke and steam pass between him and the camera, partly veiling his legs; embers and fine ash float through the frame, some closer to the camera than he is, out of focus. The hot updraft lifts his hair and swings the scarf fringe. THE KEY DETAILS: Thick viscous lava folding around the edges of the slab, the slab sitting low in the current. No text, no logos anywhere. THE MOOD: A luxury fashion-house campaign shot for real — low-key, near-black and glowing orange, the burgundy deep and smouldering. THE COMPOSITION: Exactly as in the first reference: vertical 9:16, figure centred, the full figure from head to the bottom of the slab filling about 40 percent of the frame height, top of his head about 30 percent down from the top edge with dark smoky sky above, the slab about 30 percent up from the bottom edge with glowing lava river below — everything inside a centred 4:5 crop with margin. THE CAMERA: Shot on location from slightly above the lava surface at a distance, 35mm lens, natural light only, deep focus, subtle film grain.
I kept the first one anyway ("honestly v2 isn't bad"), but it changed how I work: "instead of going to the studio then scene we go scene direct". Now the cast image and the outfit photo go straight into the scene prompt, with no studio outfit shot in between. My guess is the studio shot was where that flat light kept coming from.
The first scene-direct one had a funny miss. I asked for it "way sharper" and got a crisper photo of him resting his foot on what looks like an invisible stool.
Exact prompt
Create a new photorealistic image, but far sharper and crisper. Preserve everything from the first reference exactly — the man balancing on one leg on the needle-thin rock spire, his face, deep ebony skin, short two-strand twists, silver ear hoop, round sunglasses, his exact pose, his exact outfit (the asymmetric cream blazer with black brushstroke swirls, frog closures, pocket square, black turtleneck, wide black trousers with the ink-brush swirl, hanging straps, silver chain, chunky black boots), the sea of cloud, the misty peaks, the composition and his size and position in the frame. THE NEW CHANGE: Render the whole photograph with high clarity and micro-contrast, like a high-end digital medium-format capture — razor-sharp, high-resolution detail everywhere, no film grain, no softness, no blur, no glow. THE VIBE: The same serene, vertiginous, impossible ink-painting scene — now hyper-crisp and luxurious. THE SUBJECT: Tack-sharp detail: individual twists and flyaway hairs, the texture of his skin and pores, the reflections in the sunglasses lenses, the wool weave of the cream blazer, the crisp edges of the painted brushstrokes, the frog closures, every link of the silver chain, the stitching and grain of the leather boots, the fabric folds of the trousers. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin; luminous deep skin with warm undertone depth and natural sheen on the cheekbones, never ashy or grey. THE SETTING: Sharp, defined granite texture on the spire — cracks, lichen, grit, crisp edges; clearly defined cloud billows with crisp tops; the nearer peaks and pines sharp and detailed, the farthest peaks still fading softly into mist for depth. THE LIGHT: The same soft overcast daylight from slightly camera-left, the ONLY light on him, with clean defined shadows; his light, contrast and grade still match the scene. Only a thin wisp of mist now, low around the spire below his boot, so his legs read clearly. THE MOOD: A luxury fashion-house campaign as a living ink painting — near-monochrome grey, white and black, crisp and pristine. THE COMPOSITION: Exactly as in the first reference — vertical 9:16, figure centred, top of his head about a quarter of the way down, the needle continuing down into the clouds, everything inside a centred 4:5 crop. THE CAMERA: High-resolution medium-format digital capture, deep focus, everything tack-sharp from the spire to the near peaks, clean noise-free image, no grain.
I meant the rock. "i meant the rock tip it should be kinda like you know those super sharp rock tips? so he looks like a kung fu master balancing on the needle of the rock". So the prompt got very specific: a slender stone spike that tapers to a point narrower than his boot, him in a one-leg crane stance on it, and the camera much further back and level with him so you can see the whole needle under him.
Exact prompt
Create a new photorealistic image, but the original subject appears much further away. This is a photograph taken high in misty granite mountains above a sea of clouds, from the start — not a composite, with unified lighting across subject and environment. THE CHARACTER ANCHOR: Preserve the man from the first reference exactly — his face, deep ebony skin, short two-strand twists, silver ear hoop, round sunglasses, and his exact outfit: the asymmetric cream blazer with black calligraphic brushstroke swirls, frog closures, pocket square, black turtleneck, wide black trousers with the cream ink-brush swirl, hanging straps, silver chain, chunky black boots. Keep his raised-knee balancing pose on one leg. THE ITERATION ANCHOR: Carry forward from the first reference — the stone needle tapering to a sharp point under his boot, the misty granite peaks and twisted pines, the sea of cloud, the soft diffused overcast light and clean crisp grade. THE NEW CHANGE: The camera is much further back and level with him, not looking up. He is small in the frame and the full height of the needle below him is visible, so the sharp point reads clearly against the clouds. THE VIBE: A man balancing like a kung fu master on the razor-sharp tip of a stone needle above the clouds — serene, impossible, a living ink-wash painting. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin; luminous deep skin, never ashy or grey. THE LIGHT: Soft, bright, diffused overcast light slightly from camera-left, the only light on him, matching the scene; a thin wisp of mist below his boot; wind lifting the straps and blazer hem. THE MOOD: Near-monochrome grey, white and black, crisp and clean. THE COMPOSITION — STRICT: Vertical 9:16 frame, figure centred horizontally. The full figure from the top of his head to the needle point under his boot fills only about 35 to 40 percent of the frame height. The top of his head sits about 30 percent of the way down from the top edge, with open misty sky and peaks above and around him. The needle tapers away beneath his boot and widens down into the clouds through the lower 30 percent of the frame. Everything important stays inside a centred 4:5 crop with generous margin. No cropping of his head or boots. THE CAMERA: Shot from a neighbouring ridge at eye level with him, at a distance, 35mm lens, natural light only, deep focus, everything crisp, clean high-resolution detail.
7. Kill the plastic skin
Later I switched one of the women to a different image model, standing on a giant halved blood orange. Her face came out glossy: "her face looks too AI its like plastic".
Exact prompt
Create a new photorealistic image. This is a photograph taken in a Sicilian citrus grove in late-afternoon light, from the start — not a composite. Rebuild the entire scene as if shot natively here, with unified lighting across subject and environment. THE VIBE: An elegant woman in a romantic rose-print chiffon outfit stands serenely on top of a giant halved blood orange in a sun-drenched Sicilian orange grove — a surreal Sicilian still life at impossible scale. Calm, warm, luxurious, quietly absurd. IDENTITY LOCK: Preserve the woman from the first reference — her face, dusky deep-brown skin with a warm golden undertone, waist-length sleek straight jet-black centre-parted hair, heavy-lidded dark-brown doe eyes, thick straight low brows, a strong aquiline nose with a tiny gold stud in her left nostril, angular cheekbones, and her tall slim build. Do NOT carry over the white t-shirt, black trousers or white sneakers from the reference. She is relit entirely by this scene's light; the reference's studio lighting is discarded. THE SUBJECT: She wears a flowing silk chiffon blouse in an ivory base printed with large deep-red and blush-pink roses and green leaves, with long sleeves, gathered cuffs and a soft bow tied at the neck; a long, floor-skimming black silk chiffon skirt with a narrow black belt at the waist, the skirt flowing and layered. Gold hoop earrings. Strappy black sandals. She stands upright and relaxed on the flat cut face of the orange, weight on her left leg, one hand lightly holding a fold of her skirt, the other loose at her side, chin slightly lifted, gazing just past the lens with a calm, self-possessed expression. THE FACE: Natural skin texture, no airbrushing, no plastic or waxy skin; luminous golden-brown skin with natural sheen on her cheekbones and nose bridge, warmth in the shadows, never ashy. THE SETTING: A single giant, realistic blood orange cut in half, about two and a half metres across, resting on the grove floor with its cut face up: glistening deep crimson and ruby flesh in juicy segments radiating from a pale centre, a thin white pith ring and a thick orange-red peel with dimpled texture. Around it, rows of Sicilian orange trees with dark glossy leaves heavy with ripe oranges, fallen oranges and dry leaves on the reddish earth, an old dry-stone wall, and the distant hazy cone of Mount Etna against a clear warm sky. No people, no buildings, no other objects. THE LIGHT: Low late-afternoon Sicilian sun from camera-left is the ONLY light on her — warm, golden, directional, filtering through the orange trees in dappled patches. It lights the left side of her face and her black hair at the edges, glows through the thin chiffon sleeves, and sets the orange's juicy flesh glistening; her right side falls into soft warm shadow with green bounce from the leaves. Her contrast, sharpness and warm grade match the scene. A light breeze moves the chiffon sleeves, the skirt and a few strands of her hair. Her soft shadow falls across the cut face of the orange. THE KEY DETAILS: Droplets of juice glistening on the orange's cut segments; a couple of whole oranges on the ground nearby. No text, no logos anywhere. THE MOOD: A luxury Italian fashion campaign — sun-drenched Sicilian still life, ivory roses and black chiffon against crimson orange and green leaves, serene and witty. THE COMPOSITION — STRICT: Vertical 9:16 frame, figure centred horizontally. Her full figure from the top of her head to the orange under her sandals fills about 55 percent of the frame height. The top of her head sits about 12 percent of the way down from the top edge with sky and treetops above. The whole halved orange sits below her feet in the lower part of the frame, on the grove floor. THE CAMERA: Shot at about the height of her waist, at a gentle distance, 35mm lens, natural light only, deep focus, everything crisp.
The skin line in that prompt asked for "luminous" skin with "natural sheen". At full-body distance my read is that turns into a gloss over the whole face. So I took those words out, and added:
- Unretouched, up front: "Editorial fashion photography, unretouched, no beauty filter, no skin smoothing."
- What real skin looks like: fine visible texture, faint pores, slightly uneven tone, a few tiny imperfections, mostly matte with a little shine only on the tip of the nose and the forehead.
- A makeup level: bare skin, no foundation, no contour, no highlighter.
- Grain on the face too: fine natural film grain across the whole frame including her face.
Exact prompt
Create a new photorealistic image. This is a photograph taken in a Sicilian citrus grove in late-afternoon light, from the start — not a composite. Rebuild the entire scene as if shot natively here, with unified lighting across subject and environment. Editorial fashion photography, unretouched, no beauty filter, no skin smoothing. THE VIBE: An elegant woman in a romantic rose-print chiffon outfit stands serenely on top of a giant halved blood orange in a sun-drenched Sicilian orange grove — a surreal Sicilian still life at impossible scale. Calm, warm, luxurious, quietly absurd. IDENTITY LOCK: Preserve the woman from the first reference — her face, deep-brown skin with a warm golden undertone, waist-length sleek straight jet-black centre-parted hair, heavy-lidded dark-brown doe eyes, thick straight low brows, a strong aquiline nose with a tiny gold stud in her left nostril, angular cheekbones, and her tall slim build. Do NOT carry over the white t-shirt, black trousers or white sneakers from the reference. She is relit entirely by this scene's light; the reference's studio lighting is discarded. THE SUBJECT: She wears a flowing silk chiffon blouse in an ivory base printed with large deep-red and blush-pink roses and green leaves, with long sleeves, gathered cuffs and a soft bow tied at the neck; a long, floor-skimming black silk chiffon skirt with a narrow black belt at the waist, the skirt flowing and layered. Gold hoop earrings. Strappy black sandals. She stands upright and relaxed on the flat cut face of the orange, weight on her left leg, one hand lightly holding a fold of her skirt, the other loose at her side, chin slightly lifted, gazing just past the lens with a calm, self-possessed expression. THE FACE: Real, unretouched skin with fine visible texture, faint pores across her nose and cheeks, slightly uneven natural tone with a little darker pigmentation around the eyes and mouth corners, and a few tiny imperfections. Mostly matte skin — natural shine only faintly on the tip of her nose and the centre of her forehead. Minimal natural makeup: bare skin, no foundation, no contour, no highlighter, lightly groomed brows, a touch of colour on the lips. Deep warm undertone, warmth in the shadows, never ashy. Individual hair strands with a few flyaways at the crown. She looks like a real model photographed on location, not a render. THE SETTING: A single giant, realistic blood orange cut in half, about two and a half metres across, resting on the grove floor with its cut face up: glistening deep crimson and ruby flesh in juicy segments radiating from a pale centre, a thin white pith ring and a thick orange-red peel with dimpled texture. Around it, rows of Sicilian orange trees with dark glossy leaves heavy with ripe oranges, fallen oranges and dry leaves on the reddish earth, an old dry-stone wall, and the distant hazy cone of Mount Etna against a clear warm sky. No people, no buildings, no other objects. THE LIGHT: Low late-afternoon Sicilian sun from camera-left is the ONLY light on her — warm, golden, directional, filtering through the orange trees in dappled patches. It lights the left side of her face and her black hair at the edges and glows through the chiffon sleeves; her right side falls into soft warm shadow with green bounce from the leaves. Her contrast, sharpness and warm grade match the scene. A light breeze moves the chiffon sleeves, the skirt and a few strands of her hair. Her soft shadow falls across the cut face of the orange. THE KEY DETAILS: Droplets of juice glistening on the orange's cut segments; a couple of whole oranges on the ground nearby. No text, no logos anywhere. THE MOOD: A luxury Italian fashion campaign — sun-drenched Sicilian still life, ivory roses and black chiffon against crimson orange and green leaves, serene and witty. THE COMPOSITION — STRICT: Vertical 9:16 frame, figure centred horizontally. Her full figure from the top of her head to the orange under her sandals fills about 55 percent of the frame height. The top of her head sits about 12 percent of the way down from the top edge with sky and treetops above. The whole halved orange sits below her feet in the lower part of the frame, on the grove floor. THE CAMERA: Shot on location at about the height of her waist, at a gentle distance, 35mm lens, natural light only, deep focus, fine natural film grain across the whole frame including her face and skin.
"its definitely fixed". Honestly, my cast images were written with the same "luminous" wording, so if I rebuilt the cast today I'd use this skin block from the start.
8. Lock which way everyone faces
Last thing that kept biting me. I put one of the guys on a tiger and asked for him "facing the other way". The fix that worked was a block that names which side of the frame each end sits on: the tiger walks toward frame-left in side profile, its head and front paws on the left, its tail on the right, and nothing faces or walks toward frame-right.
Exact prompt
Create a new photorealistic image. This is a photograph taken in a central Indian sal forest from the start — not a composite. Rebuild the entire scene as if shot natively here, with unified lighting across rider, tiger and forest. Editorial fashion photography, unretouched, no beauty filter, no skin smoothing. IDENTITY: Preserve the man from the reference — the Black man with short two-strand twists, close-cut sides, clean-shaven deep ebony skin, long chiseled face, small silver hoop in his left ear. Do NOT source his face, hair or skin from anywhere else. THE SUBJECT: he wears a shrunken, slightly too-small dark charcoal suit in a liquid-sheen fabric that catches soft highlights along the folds, the trousers cropped above bare ankles, black leather loafers, and over the suit a short, plush white faux-fur jacket. He sits bareback on a tiger, upright and composed, legs hanging naturally down its flanks, one hand resting lightly on the fur at its shoulders, expression calm and unbothered, gaze off toward frame-right. Bare skin, no foundation, no contour, no highlighter. Fine visible skin texture, faint pores on nose and cheeks, slightly uneven tone, tiny imperfections; mostly matte, only a faint shine on the nose tip and forehead. THE TIGER: a real adult Bengal tiger walking slowly toward frame-right in side profile, mid-stride, heavy shoulders rolling, deep orange coat with black stripes and a white belly, paws pressing into dry leaf litter. THE SETTING: an Indian sal forest — tall, straight, grey-barked sal trunks receding into depth, broad dry leaves and golden-brown leaf litter on a dusty forest track, a few low green shrubs at the edges. Low early-morning mist hangs between the trunks, thickening with distance so the far trees fade to pale silver-grey. THE LIGHT: early morning, soft and hazy. The sun itself is NOT visible; it sits low and out of frame behind the trees on frame-left, sending a few soft diffused light beams slanting through the mist from upper frame-left toward lower frame-right. The overall light is silvery and diffuse, gently brighter on the left side of rider and tiger, with soft shadows falling toward frame-right. Cool grey-green ambient fill in the shadows, low contrast, muted palette — NOT golden hour, NOT a sunset. Single scene light only, no studio light. The mist softens the background behind them and sits faintly around the tiger's legs. THE COMPOSITION: 9:16 vertical, Reels safe zone. Rider and tiger centred and pulled back, together filling about 45% of the frame height: his head about 27% down from the top of the frame, the tiger's paws about 72% down. Tall trunks and misty canopy above, forest floor below, nothing important at the far right edge. THE CAMERA: editorial on-location photograph, fine natural film grain across the whole frame including face and skin. Looks like a real model photographed on location, not a render.
Exact prompt
Create a new photorealistic image. A single photograph shot on location in a central Indian sal forest in early morning — rider, tiger and forest all captured in the same frame by the same light. Editorial fashion photography, unretouched, no beauty filter, no skin smoothing. THE ORIENTATION (LOCKED): the tiger walks toward FRAME-LEFT in side profile — its head and front paws on the left side of the frame, its tail and hindquarters on the right. The man rides facing FRAME-LEFT in the same direction, his face in three-quarter view looking off toward frame-left. Nothing faces or walks toward frame-right. THE SUBJECT: a striking male model in his mid-twenties, Black, with short two-strand twists and close-cut sides, clean-shaven deep ebony skin, a long chiseled face, a lean frame and a small silver hoop in his left ear. He wears a shrunken, slightly too-small dark charcoal suit in a liquid-sheen fabric that catches soft highlights along its folds, the trousers cropped above bare ankles, black leather loafers, and over the suit a short, plush white faux-fur jacket. He sits bareback on the tiger, upright and composed, legs hanging down its flanks, one hand resting lightly on the fur at its shoulders, expression calm and unbothered. Bare skin, no foundation, no contour, no highlighter. Fine visible skin texture, faint pores on nose and cheeks, slightly uneven tone, tiny imperfections; mostly matte, only a faint shine on the nose tip and forehead. THE TIGER: a real adult Bengal tiger walking slowly toward frame-left, mid-stride, heavy shoulders rolling, deep orange coat with black stripes and a white belly, paws pressing into dry leaf litter. THE SETTING: a dusty forest track through an Indian sal forest — tall, straight, grey-barked sal trunks receding into depth, golden-brown dry leaf litter on the ground, a few low green shrubs at the edges. Low early-morning mist hangs between the trunks, thickening with distance so the far trees fade to pale silver-grey. THE LIGHT: the low morning sun is OUT OF FRAME on frame-left, behind the trees, sending soft diffused beams slanting through the mist from upper frame-left toward lower frame-right. One of those beams falls across the rider and tiger, softly lighting his face, the front of the white fur and the tiger's head and chest with slightly warm light, while the back of his body and the tiger's hindquarters fall into cool, grey-green shade — a real directional light with visible falloff, NOT evenly lit. Soft shadows fall across the track toward frame-right. Muted silver-grey and green palette, low contrast, gentle haze — NOT golden hour, NOT a sunset. His exposure and contrast match the forest exactly — lit by the same misty light as the tiger and the trees, with the same haze, shadow density and grain; the mist softens the background and sits faintly around the tiger's legs. Single scene light only, no studio light, no fill on his face. THE COMPOSITION: 9:16 vertical, Reels safe zone. Rider and tiger centred and pulled back, small in the frame, together filling about 45% of the frame height: his head about 27% down from the top of the frame, the tiger's paws about 72% down. Tall trunks and misty canopy above, forest floor below, nothing important at the far right edge. THE CAMERA: editorial on-location photograph, fine natural film grain across the whole frame including face and skin. Looks like a real model photographed on location, not a render, not a composite.
The second one has no cast image attached at all. I wrote his twists, his long face and the silver hoop in his ear straight into the prompt, and he still reads as the same guy.
Why it works
Every model was built on the same backdrop, outfit and framing, so the only differences between them are the faces. Then each face got marks that are easy to spot and easy to write down, which is how the same faces kept showing up on an ice floe, a rock needle and a tiger.
Most of what broke happened when I reused the cast images, not when I made them. The studio light came along for the ride, the outfit drifted when I described it, and the framing changed from post to post until I gave it numbers.
Your turn: make your first cast member today. One plain outfit, a grey backdrop, and one or two marks you'd recognize from across the room.
Emero saves your cast and your rules, like the skin block and the framing numbers, so every new scene starts from the same people. Everything's at cost.
Get one of these every day
I test new workflows daily and send them with the exact prompts, the models and the costs
5,979 generations tested so far
More workflows
See all
Nano Banana Pro + Kling 3.0 - a surreal knitted wool world in three clips
Build a surreal world made of wool, dress a character to match and animate three short clips
Nano Banana Pro, Kling 3.0 · 10 min render
Nano Banana Pro vs GPT Image 2.5 Flare - one street ad keyframe in 12 tries
Fixing the light and the subject distance on one street ad frame across 12 tries
Nano Banana Pro, GPT Image 2.5 Flare, MiniMax H3 · 8 min render
Nano Banana Pro + Kling 3.0 - a cougar handoff and a pet in the same beat
A real-size cougar delivers a red crossbody bag while she takes it and pets it at the same time
Nano Banana Pro, Kling 3.0 · 8 min render