Most AI-generated photos of people look like AI-generated photos of people. Not because the models can't do better, but because the prompts are wrong.
We spent weeks generating over 28 photorealistic UGC influencer images using Nano Banana 2 in Masonry — iterating through three full prompt versions, failing in predictable ways, and eventually landing on a system that produces images people genuinely mistake for real content. Not "pretty good for AI" — actually indistinguishable.
The difference wasn't the model. It was the prompt architecture.
This guide covers everything we learned: the five layers of a photorealistic prompt, the specific mistakes that make images look fake, how to maintain facial consistency across an entire feed, and the complete battle-tested prompts for 10 different nationalities — each with a distinct city, fashion aesthetic, and lighting setup.
Why Most AI Portraits Look Fake (And How to Fix It)
The default output of any image model is a smooth, evenly-lit, perfectly symmetrical face staring into the camera. It looks like a LinkedIn headshot at best and a stock photo at worst.
Real UGC content is the opposite. It's a girl on a fire escape squinting into the sun. It's smudged kajal at dusk. It's a phone propped against a coffee cup on a café counter. The camera angle is slightly off. The hair is slightly messy. The light comes from two directions and creates shadows that don't quite match.
That messiness is what makes it feel real. And every piece of it needs to be in your prompt.
We found five layers that consistently separate photorealistic results from obviously-AI output:
- Layer 1: Camera and angle — "iPhone front camera selfie from slightly above" or "shot from coffee table level looking across." This sets the entire visual framework.
- Layer 2: Body in space — Not "sitting on a couch" but "lying sideways across a wide cream boucle sofa, head propped up on one hand, legs stretched out and slightly bent." The model needs spatial information.
- Layer 3: Skin imperfections — Pores, freckles, beauty marks, sun flush, slight dark circles, oil sheen from humidity. Include at least 2-3 per prompt. This is the single biggest lever for realism.
- Layer 4: Fashion, not costume — Don't dress your Indian character in a sari or your Korean character in a hanbok. Dress them like real fashion-forward people from those cities. A Mumbai fashion founder wears a black blazer suit with sculptural gold jewelry.
- Layer 5: Cinematic lighting — Light from two directions. "Warm amber floor lamp from camera-right, cool blue-pink neon from rainy window camera-left." This creates the depth and mood that flat lighting can't match.
Starting Simple: The iPhone Selfie Foundation
Before going global, we started with the basics — a single persona in 10 indoor settings. The goal was to make each image look like a real iPhone front camera selfie taken by a real person.
Morning Bed Selfie
Candid UGC lifestyle iPhone front camera selfie photo of a young American woman in her early 20s with light tan skin, natural visible pores, subtle faint freckles across nose and cheeks, hazel eyes with warm green-brown tones, long brown hair slightly messy spread softly across white pillow with loose natural waves. She has a calm slightly sleepy expression with a soft relaxed smile, lips gently closed. Wearing a fitted white ribbed tank top with clean scoop neckline sitting just below the collarbone, thin fitted straps showing natural shoulder line. Subtle cotton-ribbed texture visible across chest. She is lying on slightly wrinkled white bedsheets with soft natural folds in the blanket, in a minimal bedroom with green indoor plants near a sunlit window in soft background. Natural morning sidelight, soft and warm, gentle glow along jawline and cheekbones, delicate soft shadows across the pillow and neck. Slightly top-down perspective looking down at subject lying on bed, close-up portrait, handheld with arm extended out of frame. Sharp focus on face with soft natural edge falloff. Cozy and intimate mood. No phone visible anywhere in frame. No filters. No artificial skin smoothing.
Why this works:
- "Slightly top-down perspective looking down at subject lying on bed" gives the natural selfie angle you'd get holding a phone above yourself
- "Slightly wrinkled white bedsheets with soft natural folds" — imperfect environments signal authenticity
- "Sharp focus on face with soft natural edge falloff" mimics how phone cameras actually behave
- The prompt describes fabric texture ("cotton-ribbed"), not just the garment — this prevents the model from generating flat, textureless clothing
Maintaining Facial Consistency Across a Feed
A single prompt generates a single image. But influencer content is a feed — 9, 12, 18 images of the same person in different outfits and locations.
Without a reference image, the model creates a different face every time. Same prompt, different person. Every single generation.
The fix is simple. Nano Banana 2 supports reference images through the --ref flag. Pick your best generation, pass it as a reference, and the model preserves facial features while changing everything else.
masonry image "your prompt here" \ --ref reference_face.png \ -m gemini-3.1-flash-image-preview \ -a "4:5@2k" \ -o output.png
Using our morning bed selfie as the reference face, we generated 8 completely different scenarios — outdoor café, street style, rooftop sunset, park picnic — and the face stayed consistent across all of them.
Same Face, Different World
Same woman from reference photo taking an iPhone front camera selfie. She is sitting on wooden outdoor cafe chair wearing a white linen button-up shirt with sleeves rolled to elbows. Her brown wavy hair is loose and slightly wind-blown. She has a calm confident look with a slight smile looking directly into the camera. Warm golden hour sunlight, blurred outdoor cafe setting with greenery and string lights softly out of focus in background. Natural skin texture, warm sun glow on face. Effortless chic outdoor mood. No phone visible in frame. No filters. No skin smoothing.
Same woman from reference photo taking an iPhone front camera selfie. She is leaning against a sunlit brick wall on a city street wearing a camel-colored leather jacket over a simple white top. Her brown wavy hair is loose and flowing naturally. She has a cool relaxed expression with a slight smirk looking directly into the camera. Warm afternoon sunlight casting sharp shadows on the brick wall, blurred urban street with trees in background. Natural skin texture with warm sun glow. Urban street style mood. No phone visible in frame. No filters. No skin smoothing.
The key insight: reference images preserve structure (face shape, nose, eyes, bone structure) while giving the model freedom to adapt expression, hair styling, and context. It's not cloning — it's consistency.
Going Global: 10 Nationalities, 10 Cities, 10 Fashion Aesthetics
This is where things got hard — and where the prompt engineering actually matters.
We set out to create 10 UGC influencer images across 10 nationalities, each set in their home city with a distinct fashion aesthetic and lighting setup. We went through three complete prompt versions before getting results that felt real.
The Three Failures That Taught Us Everything
Version 1 was too editorial. We described full fashion-magazine makeup (razor-sharp contour, tightlined waterline, deep smoky eye) and runway outfits (structured bralette-style co-ord top, cold-shoulder blazer). The results looked like AI stock photos — technically impressive, emotionally dead.
Version 2 was too stereotypical. We fixed the realism problem by adding imperfections (dark circles, smudged liner, messy hair), but the outfits defaulted to cultural costume. Cotton kurta with glass bangles for India. Cable-knit cardigan with a baby tee for Korea. The images felt like a tourism poster, not an influencer's feed.
Version 3 was the breakthrough. We combined authentic skin and minimal makeup with fashion-forward personal style and — critically — specific body poses and camera angles. The Indian woman wears a double-breasted black blazer suit, not a kurta. The Korean woman wears a tan leather shacket, not a cable-knit. And instead of "sitting on a sofa," we wrote "sitting sideways on the arm of a modern sofa, one leg dangling down, the other tucked up, head tilted slightly."
The difference was enormous.
The Final 10: Complete Prompts and Results
Every prompt below was generated at 2K resolution (1856 × 2304) using Nano Banana 2 in Masonry. Each one is copy-paste ready.
Brazilian — Bedroom Golden Hour
iPhone selfie of a 22-year-old Brazilian woman lying back on a bed in a modern São Paulo apartment bedroom at golden hour, propped up on one elbow, body turned toward the camera, phone held up in her other hand capturing herself from slightly above. Deep sun-kissed olive skin with warm bronzed undertone, dewy luminous texture, natural pores, golden flush on cheekbones. Deep espresso brown long hair loose and fanning out naturally on the pillow behind her. Soft confident expression, lips slightly parted in a relaxed almost-smile, direct warm gaze into the camera lens. Dewy bronzed makeup, terracotta eyeshadow, peachy blush, nude-brown glossy lips. Wearing a matching two-piece satin co-ord set in deep burnt sienna — thin-strap crop camisole and high-waist wide-leg satin trousers, fabric pooling and catching golden light around her reclining body. Delicate layered gold chain necklaces at collarbone. Background: plush linen bed with terracotta throw pillows, rattan pendant light above, potted indoor palm. Warm directional golden hour sunlight from a side window. Intimate relaxed bedroom UGC influencer selfie energy.
Why this works: The reclining pose with phone held up creates authentic selfie geometry. The satin fabric "pooling and catching golden light" gives the model specific texture instructions. Terracotta pillows, rattan pendant, indoor palm — three props that immediately say "São Paulo apartment."
Russian — Marble Bathroom Morning
Bathroom mirror selfie of a 24-year-old Russian woman standing in a luxury Saint Petersburg marble bathroom at early morning, holding her smartphone up with one hand capturing her full reflection, face clearly visible and front-facing in the mirror. Naturally pale porcelain skin with cool undertone, slight tiredness under eyes, real fine skin texture with visible pores. Striking ice blue almond-shaped eyes with thin precise brown liner. Platinum blonde shoulder-length hair in a fresh slightly-slept-on blowout, one side tucked behind ear. Direct and unbothered expression, slight knowing smirk, one eyebrow barely raised, effortless cool-girl morning energy. Soft mauve lip stain, brushed-up natural brows. Wearing an oversized structured camel wool blazer worn as a dress, belted loosely at waist with a thin gold chain belt, hem falling mid-thigh, one sleeve pushed up revealing a vintage gold watch. White fluffy slippers. Small gold huggie hoop earrings. Smartphone visible in her raised hand in the mirror reflection. Background in mirror: white and grey veined marble walls, floating double vanity with matte black fixtures, backlit mirror, folded white towels on matte black rail. Cool diffused flat morning daylight from above.
Why this works: "Slight tiredness under eyes" and "slightly-slept-on blowout" are the kind of lived-in details that AI images never have. The camel blazer worn as a dress with a chain belt and fluffy slippers is a fashion statement, not a costume.
American — Brooklyn Sunday Afternoon
Candid lifestyle shot of a 27-year-old American woman with mixed heritage lying sideways across a wide cream boucle sofa in a Williamsburg Brooklyn apartment on a late Sunday afternoon, body fully reclined along the sofa length, head propped up on one hand with elbow on the sofa arm, other hand resting loosely on her hip, legs stretched out and slightly bent, looking directly at the camera with a genuine relaxed smile. Medium warm beige skin with golden undertone, prominent natural freckles densely scattered across nose and cheeks, slight sun flush, real visible pores. Green eyes with warm amber flecks, slightly squinting in the afternoon light. Dark brown hair with sun-bleached auburn ends in a messy high bun with many loose strands falling around her face. Just tinted SPF letting all freckles show, warm bronzer, clear lip gloss. Wearing a vintage oversized dark brown leather bomber jacket worn open over a cropped sage green ribbed tank, low-rise wide-leg cream linen trousers, woven brown belt. Layered thin gold chains, small signet ring on pinky, tortoiseshell sunglasses pushed up on head. Bare feet visible at the end of the sofa. Shot from coffee table level looking across at her full reclining pose. Background: cream boucle cushions, chunky knit throw, abstract art print on white wall, rattan lamp, trailing pothos.
Why this works: "Prominent natural freckles densely scattered" and "just tinted SPF letting all freckles show" — two layers of the same instruction, reinforcing that the model should not cover the freckles. "Shot from coffee table level looking across" creates a perspective that no AI default would ever choose.
French — Parisian Bedroom
Candid lifestyle shot of a 23-year-old French woman sitting cross-legged on a low platform bed in a minimalist Paris apartment bedroom on a warm afternoon, leaning slightly forward with both hands resting in her lap, chin tilted down slightly, looking up at the camera with a quiet knowing expression and closed-lip half-smile, classically Parisian and completely at ease. Fair skin with cool neutral undertone, slightly peachy at cheeks, smooth refined fine pore texture, faint natural flush. Deep warm brown slightly heavy-lidded eyes defined by clean thin black liner. Dark chocolate brown jaw-length choppy textured French bob falling naturally. Classic red matte lip as the hero, strong defined brow, thin black liner flick, minimal blush. Wearing a fitted black ribbed long-sleeve mock-neck bodysuit tucked into a perfectly tailored high-waist camel midi skirt with a centre-front split. Small gold stud earrings, single thin gold bangle. Black pointed-toe kitten heel mules placed beside the bed visible in frame. Background: ivory linen duvet and black accent pillow, white oak herringbone floor, tall arched window with warm Paris afternoon light, single sculptural vase on bedside table.
Why this works: "Classic red matte lip as the hero" — one bold choice, everything else minimal. That's real Parisian style. "Kitten heel mules placed beside the bed visible in frame" tells a story: she took her shoes off to sit on the bed.
Indian — Mumbai Dusk
Candid lifestyle shot of a 25-year-old Indian woman sitting sideways on the arm of a modern sofa in a curated South Mumbai apartment living room at dusk, one leg dangling down and the other tucked up, leaning back slightly with one hand resting on her raised knee and the other hand lightly touching her collarbone, head tilted slightly, looking directly at the camera with a warm magnetic half-smile and lifted chin. Rich warm medium-brown skin with deep golden undertone, natural healthy sheen, real skin texture with visible pores, small beauty mark near left jawline. Deep dark brown large expressive eyes with smudged kajal on waterline, soft bronze shimmer on lid, natural full lashes. Deep black thick hair blown out voluminously in a side-part past shoulders. Wearing a sharply tailored double-breasted cropped matte black blazer with gold buttons, deep V framing collarbone and décolletage, matched with high-waist wide-leg black tailored trousers. Chunky gold chain bracelet, gold sculptural ear cuff on left ear, thin gold hoops on right, delicate chain at collarbone. Gold strap heeled mule on her dangling foot. Shot from slightly below eye level looking up at her seated on the sofa arm. Background: cream plaster wall with large modern abstract canvas, designer amber floor lamp glowing, Mumbai city lights beginning through floor-to-ceiling glass at blue hour. Warm amber interior light from floor lamp camera-right, cool blue dusk from glass window camera-left, cinematic warm-cool split.
Why this works: "Shot from slightly below eye level looking up" — a power angle that makes the subject look commanding. The warm-cool light split creates cinematic depth. "Gold sculptural ear cuff on left ear, thin gold hoops on right" — mismatched earrings are a real fashion choice, not a prompt error.
Korean — Rainy Seoul Evening
Candid lifestyle shot of a 22-year-old Korean woman sitting on the floor of her Hannam-dong Seoul apartment bedroom on a rainy weeknight evening, back leaning against the side of the low platform bed, knees pulled up to her chest, one hand loosely wrapped around her knees and the other hand raised with fingers lightly touching her cheek, head tilted sideways with a soft knowing smile looking directly into the camera, completely at ease. Porcelain skin with natural cool-pink undertone, genuinely good skin with subtle texture on cheeks, real and unairbrushed. Dark brown-black slightly monolid eyes, thin smudged brown pencil liner, natural lashes with one coat brown mascara. Glossy black hair long past shoulders, center-parted, one side tucked behind ear revealing a small sculptural silver ear cuff. Dewy cushion foundation, soft peach cream blush on apples of cheeks and nose tip, gradient rose lip tint fading from center outward. Wearing a matching oversized oat cream chunky ribbed knit two-piece lounge co-ord — boxy cropped crewneck pullover and matching high-waist wide-leg knit trousers. One silver ring on index finger. Fluffy ivory knit socks. Shot from a low angle at floor level looking across at her sitting against the bed. Background: dark wood platform bed behind her, tall arched black floor lamp with warm amber glow, large window showing rainy Seoul street with neon Hangul signs reflecting on wet pavement in soft bokeh. Warm amber floor lamp glow from camera-right, cool blue-pink neon from rainy window camera-left, cinematic colour split.
Why this works: "Gradient rose lip tint fading from center outward" — this is the classic Korean gradient lip technique. "Neon Hangul signs reflecting on wet pavement" creates the rainy-Seoul atmosphere without being heavy-handed. The oat cream knit co-ord with fluffy socks is cozy influencer content, not cultural costume.
Italian — Amalfi Villa Bathroom
Candid lifestyle shot of a 26-year-old Italian woman sitting on the edge of a marble bathroom vanity counter in a contemporary Amalfi villa bathroom on a mid afternoon, legs crossed at the ankle dangling off the counter, one hand flat on the counter beside her supporting her weight, other hand resting lightly in her lap, leaning slightly back and looking directly into the camera with a warm open smile, lips parted, completely natural and effortlessly beautiful. Warm golden-olive deeply sun-kissed Mediterranean complexion, smooth with warm natural glow, subtle pore texture, light sun-touch across nose and cheeks. Warm hazel eyes with strong green and amber tones. Deep dark brown long hair in loose romantic dark curls past chest, natural volume and sea-salt texture. Sheer foundation, terracotta and bronze eyeshadow, coral blush, nude coral gloss lips. Wearing a plunging wrap mini dress in fluid deep burgundy satin with thin self-tie straps, hem riding up slightly as she sits on the counter, bare golden-olive legs dangling. Thin delicate gold chain necklace, small gold hoop earrings. Strappy gold flat sandals on her dangling feet. Background: white Calacatta marble walls with grey veining, brushed gold fixtures, fluffy white towels on gold rail, small bud vase with eucalyptus stems, frosted window with warm diffused Mediterranean light.
Why this works: "Hem riding up slightly as she sits on the counter" — real-world clothing behavior that AI images never include. "Sea-salt texture" in the hair tells a story. Calacatta marble, brushed gold fixtures, eucalyptus stems — three details that say "Amalfi villa" without stating it.
British — Notting Hill Autumn
Candid lifestyle shot of a 24-year-old British woman sitting on the floor of a Notting Hill London apartment living room on an overcast autumn afternoon, back resting against the front of a low greige boucle sofa, legs stretched out and crossed at the ankle, one hand holding a ceramic mug resting on her knee, the other hand pushing her hair back from her face mid-gesture, looking at the camera with a quiet amused smile caught mid-moment, effortlessly composed. Light neutral skin with cool peachy undertone, very faint freckles on nose and upper cheeks, fine pore texture, slight rosy flush. Blue-grey clear eyes. Strawberry blonde hair past shoulders in loose natural waves. Wearing a fitted square-neck camel ribbed knit midi dress with structured square neckline, full collarbone exposed, form-fitting ribbed fabric falling to mid-calf, tan leather belt at natural waist. Pearl drop earrings, thin gold watch. Brown leather pointed-toe loafer flats beside her on the floor. Background: greige boucle sofa behind her, cream plaster wall, stack of coffee table books beside her, tall window with overcast grey London light, warm amber side lamp glow. Diffused cool overcast light from window balanced with warm amber lamp.
Why this works: "The other hand pushing her hair back from her face mid-gesture" transforms the image from a posed portrait to a caught moment. The model generated a book stack with "Notting Hill" on the spine — we didn't even specify the titles.
Mexican — Mexico City Mirror Selfie
Bedroom mirror selfie of a 23-year-old Mexican woman standing in front of a full-length mirror in a Colonia Roma Mexico City apartment bedroom on a warm afternoon, holding her smartphone up in one hand at a slight downward angle, other hand resting on her hip with a confident stance, looking directly into the camera with a wide radiant smile and bright warm eyes. Warm medium-tan skin with rich golden-copper undertones, vibrant and full of life, smooth healthy natural warmth. Deep warm dark brown large expressive eyes dramatically framed by full lashes and bold liner. Rich dark black-brown long hair past chest, thick and voluminous, half-up half-down with top section loosely pulled back. Bold full lashes, warm terracotta-brick eyeshadow, vibrant coral-orange bold lip. Wearing a matching white embroidered off-shoulder co-ord set — fitted off-shoulder crop top with traditional Mexican floral embroidery in bright coral, yellow, and green, wide elasticated neckline pulled below both shoulder joints, paired with matching high-waist wide-leg white linen trousers with the same embroidery along the hem. Multiple layered gold choker necklaces, large gold hoop earrings. Bare feet on warm terracotta tile floor. Smartphone visible in her raised hand in the mirror. Background in mirror: terracotta-painted bedroom wall, low rattan bed frame with bright woven blanket, hanging macramé wall art, Mexican ceramic vases, dappled warm afternoon light through sheer curtain casting leaf shadow patterns.
Why this works: This is the one prompt where we intentionally leaned into cultural fashion — the embroidered off-shoulder blouse is a genuine style choice in Colonia Roma, not a costume. "Dappled warm afternoon light through sheer curtain casting leaf shadow patterns" gives the model specific light behavior to render.
Scandinavian — Copenhagen Kitchen Morning
Candid lifestyle shot of a 25-year-old Scandinavian Danish woman standing barefoot in a minimal Copenhagen apartment kitchen on a soft cool morning, leaning with her back against the kitchen counter, both hands loosely wrapped around a large ceramic coffee mug held at chest height, head slightly tilted and looking directly at the camera with a barely-there soft smile and serene morning calm, completely present and unhurried. Fair cool skin with faintest pink undertone, clean and Nordic, very smooth fine skin texture, slight natural pink flush along cheeks. Grey-blue clear sharp eyes set wide apart. Ash blonde long hair past shoulders in a loose effortless low bun at nape, several face-framing pieces left loose. Barely-there sheer dewy coverage, clean thin brown mascara only, minimal natural brow, very sheer natural rose blush, clear lip balm, no-makeup makeup Scandi ideal. Wearing a matching oversized oat-white premium modal knit two-piece lounge co-ord — relaxed wide-neck long-sleeve top with dropped shoulders and matching high-waist wide-leg knit trousers. No jewellery. Barefoot on warm white oak herringbone wood floor. Background: white oak cabinetry with brushed brass handles, poured concrete countertop, open cookbook on the counter, large arched window flooding cool Scandinavian morning daylight, single sculptural dried flower arrangement in a ceramic vase.
Why this works: "No jewellery" is as important as any positive instruction — it tells the model to leave something out, which is the essence of Scandinavian minimalism. "Completely present and unhurried" is a mood instruction that affects the entire composition.
The 8 Rules We Learned the Hard Way
After 28+ images and three complete prompt rewrites, these are the rules that consistently separate real-looking results from obviously-AI output.
- Lead with camera and perspective
"iPhone front camera selfie from slightly above" or "shot from coffee table level looking across at her full reclining pose." The perspective sets the entire visual framework. Default eye-level produces default results.
- Describe bodies in space, not just positions
"Sitting on a couch" gives the model nothing. "Lying sideways across a wide cream boucle sofa, head propped up on one hand with elbow on the sofa arm, other hand resting loosely on her hip, legs stretched out and slightly bent" gives it a complete spatial map.
- Skin imperfections are the single biggest lever
Real people have pores, freckles, beauty marks, sun flush, slight dark circles, oil sheen from humidity, and uneven texture. Include at least 2-3 per prompt. This is non-negotiable for realism.
- Fashion is personal, not national
Don't dress characters in cultural costume unless it's a genuine fashion choice in that city. A Mumbai fashion founder wears a black blazer suit with sculptural gold jewelry. A Seoul cool-girl wears a tan leather shacket over a black square-neck top. A Brooklyn creative wears a vintage 90s leather bomber with layered gold chains.
- Environmental micro-details sell the whole story
"Coffee table books" is generic. "Stack of coffee table books with a candle and brown leather loafers beside her on the floor" tells a story. The model picks up on narrative context and generates more coherent scenes.
- Two-direction lighting creates cinema
"Warm amber floor lamp glow from camera-right, cool blue-pink neon from rainy window camera-left" creates a colour split that makes the image feel like a frame from a movie, not a flat photograph.
- Keep negative prompts short and specific
"No phone visible, no filters, no skin smoothing" works. Long lists of negatives confuse the model and produce worse results than three precise ones.
- Iterate in versions, not tweaks
If an image feels fake, don't change one adjective. Rethink the pose, outfit, environment, and lighting together. Our biggest quality jumps came from rewriting prompts from scratch, not from adjusting them incrementally.
Technical Setup
All images were generated using Masonry with the following configuration:
- Model: Nano Banana 2 (gemini-3.1-flash-image-preview)
- Resolution: 2K (1856 × 2304 at 4:5 aspect ratio)
- Reference images: --ref flag for facial consistency
- Negative prompts: --negative-prompt flag for constraints
Generate an image
masonry image "your prompt here" \ -m gemini-3.1-flash-image-preview \ -a "4:5@2k" \ -o "output.png" \ --negative-prompt "phone visible, filters, skin smoothing"
Generate with face reference
masonry image "your prompt here" \ -m gemini-3.1-flash-image-preview \ -a "4:5@2k" \ --ref reference_face.png \ -o "output.png"
Start Building Your Own Feed
The prompts in this guide aren't templates to copy verbatim — they're patterns to adapt. Take the five-layer structure, apply it to your brand's aesthetic, and iterate. Change the city. Change the outfit. Change the lighting. Keep the specificity.
The model can generate anything you can describe. The question isn't whether the AI is good enough. It's whether your prompt is.
Start with one image. Get it right. Then build the feed.
All images in this post were generated using Nano Banana 2 via Masonry. No real people were photographed.
For prompting techniques across other models, see our guide to prompting AI image models; for product-focused shots, the best AI image model for product photography roundup.


