Tips & Tricks8 min read
How to Make AI Images Look Real, Not AI-Generated
The tells that give AI images away — over-symmetry, unnamed light, plastic skin — and the specific prompt changes that remove them.
On this page
- Why AI images get spotted
- What our prompt library says people actually write
- Fix 1: Name one light source and one direction
- Fix 2: Give it a camera, not an adjective
- Fix 3: Skin needs texture, and texture means flaws
- Fix 4: Break the symmetry
- Fix 5: Photograph a moment, not a person
- A template you can paste and fill
- More prompts that already do this
- FAQ
- Why do my AI images still look fake?
- Does adding "photorealistic" to a prompt help?
- Which AI model makes the most realistic images?
- How do I get realistic skin without it looking dirty?
- Should I add film grain?
Why AI images get spotted
Nobody catches an AI image by counting pixels. They catch it on five physical mistakes: skin with no pores, light with no source, a face centred to the millimetre, a frame that is equally sharp from eyelashes to back wall, and a person who is posing instead of doing something.
Every one of those is a prompt problem, not a model problem. This guide is the five fixes, each with real prompts from our gallery that already apply them — copy the mechanics, swap the subject.
Run this checklist against your last render before you blame the model:
| The tell | What the model did | What a camera would have done |
|---|---|---|
| Poreless, waxy skin | Averaged a million retouched faces | Recorded pores, fine hair, uneven tone, shine on the forehead |
| No identifiable light source | Lit the subject from everywhere at once | One dominant source, one direction, one shadow edge |
| Perfect symmetry | Centred the face and balanced both halves | Off-centre framing, tilted horizon, an elbow cropped by the edge |
| Uniform sharpness front to back | Treated the whole frame as the subject | A focal plane — near eye sharp, ear already soft |
| Nothing is happening | Produced a pose | Caught a moment: mid-step, mid-laugh, mid-blink |
What our prompt library says people actually write
These counts come from our own gallery corpus: 9,667 published prompts, 9,599 of them longer than 20 characters, counted in August 2026.
26.4%
of prompts ask for photorealistic, ultra-realistic or hyper-realistic
2,533 of 9,599 prompts
53%
of those never mention skin at all
1,346 of the 2,533 realism requests
9.3%
of all prompts specify skin texture
pores 3.6%, freckles 1.6%
4.8%
of all prompts ask for a candid moment
motion blur 2.8%, an iPhone named 2.0%
That is the whole problem in one table. The most popular realism instruction in the library is the word "photorealistic" itself, and it is the instruction that carries the least information — it describes the result you want, not what to put in the frame. More than half of the prompts demanding photorealism say nothing whatsoever about skin, which is the first surface anyone looks at.
The models that get used for photographic work show the same split. Across our corpus, 45.5% of Nano Banana prompts carry photographic parameters, against 28.0% for GPT Image and 9.6% for Midjourney — a 4.7× gap between the top and bottom. If you want a photograph rather than an illustration, the corpus is telling you to write like a photographer. Our full breakdown of 9,667 prompts has the rest of the split.
Fix 1: Name one light source and one direction
"Cinematic lighting" is a mood; it does not tell the model where the shadow falls. In the corpus the most-used lighting terms are studio lighting (6.5%), cinematic lighting (4.8%), rim light (3.9%) and natural light (3.8%) — the first two are vibes, the second two are physics. Prefer physics.
✕ Bad
beautiful lighting, cinematic, dramatic, 8k, photorealistic✓ Good
single softbox at 45 degrees camera-left, one metre from her face, no fill on the right side, hard shadow edge running down the far cheek, dim tungsten practical on the back wallOnce there is one dominant source, the image gets a shadow logic that holds together — highlights on one side, falloff on the other, and a background that is lit differently from the face.
Controlled Gaze Under Dominant Flash
Nano Banana ProPrompt▾
Use reference image. Preserve identity and facial structure strictly. A slim Asian woman standing still, body deliberately composed rather than casual. Her posture feels arranged and intentional, not relaxed. She faces the camera directly, shoulders stable, center of gravity controlled. Her expression is calm and restrained — not playful, not overly sweet. A controlled, subtle smile, with emotional distance. Her gaze is steady and aware, fixed toward the camera, creating quiet dominance rather than friendliness. Hands & Props Right hand holding a hanging cluster of golden trumpet flowers, positioned deliberately as a visual accent Left hand holding a larger bouquet of golden trumpet flowers, lowered slightly Flowers function as props only, never overpowering the subject The woman remains the absolute visual focal point at all times Hair Long hair, slightly messy layered style Hair gently lifted by wind, with a few loose strands crossing the face Chic, slightly sexy, but controlled — never messy or cute Outfit White vintage-style puff-sleeve blouse Large puffed sleeves Layered ruffle cuffs Square neckline Drawstrings at the chest forming soft gathers Slightly long length covering the hips Front slit visible but restrained Light blue jeans Colorful beaded necklace as a subtle accent Makeup Fair skin tone Douyin + Korean beauty inspired Long curled eyelashes Soft pink blush on cheeks and nose tip Glossy pink-orange lips with juicy shine Makeup feels polished and deliberate, not cute Nails White French tip nails Subtle Christmas-themed nail art, low contrast Environment Standing directly in front of a fully blooming Golden Trumpet Tree (Thong Urai) The tree is dense with bright golden flowers, forming a saturated backdrop Some flower stems partially overlap the top edge of the frame, creating a soft foreground blur Background exists to frame the subject, not to compete with her Lighting (CRITICAL) Direct flash is the dominant key light Flash visibly overpowers ambient light Skin appears very bright with slight yellow flash tone Clear, hard-edged shadow cast behind the subject Natural sunlight exists only as weak ambient fill Lighting creates strong separation between subject and background Photography Style Compact camera aesthetic Fujifilm Pro 400H film look Vintage lens character Subtle film grain Ultra-sharp subject clarity Bright exposure with aggressive subject emphasis 8K resolution The image feels intentional, staged, and directed — not casual or spontaneous
CCD Flash Hotel Lazy Heel-Off Moment
Nano Banana ProPrompt▾
{"portrait_prompt":{"subject":{"description":"Based on <User Portrait>, a young woman lying on white sofa, seen from directly above, relaxed lazy pose, looking up at camera","features":{"hair":"Messy tousled hair spread on sofa cushion","expression":"Relaxed, slightly tired, intimate late-night vibe"},"pose":"Lying on back on sofa, one hand reaching down to touch her heel still on her foot, fingers hooked on the back of her high heel about to slip it off, other hand resting near her head or in her hair, legs slightly bent, the moment of taking off shoes"},"attire":{"type":"Based on user-uploaded clothing reference","details":["Black blazer dress","Sheer black stockings","Heels removed and held in hand"],"color":"As per clothing reference"},"composition":{"shot_type":"Top-down overhead shot, full body visible","focal_length":"28-35mm wide angle, compact camera","camera_angle":"Directly overhead, bird's eye view, looking straight down","camera_height":"Standing directly over subject","framing":"Subject fills frame diagonally, some sofa and floor visible around edges, casual snapshot framing","aspect_ratio":"9:16"},"lighting":{"type":"Direct on-camera flash, harsh and unflattering","direction":"From camera position straight down","mood":"Hard flash creating harsh highlights on skin and fabric, distinct shadow cast on sofa beneath subject, flash reflection in eyes, overexposed hot spots on forehead and cheekbones"},"color_palette":{"film_simulation":"CCD compact camera with flash","style":"Early 2000s point-and-shoot digital camera, direct flash, high ISO noise, warm incandescent ambient mixed with cool flash, slightly muddy colors","tones":["Cool flash on skin","Warm ambient in shadows","Blown highlights"],"grade_notes":["highlight: BLOWN OUT, clipped whites on skin and sofa","shadow: warm ambient light visible in corners","contrast: harsh from flash, deep blacks where flash doesn't reach","noise: VISIBLE high ISO digital noise, color noise in shadows","sharpness: slightly soft, compact camera lens quality","flash falloff: bright center, darker edges"]},"environment":{"setting":"Hotel room interior, nighttime","background":"White sofa, marble floor partially visible, window with city lights or curtains in background, indoor ambient warm lighting mixed with flash","atmosphere":"Late night, private moment, after-party vibe"},"mood":"Intimate snapshot, late night hotel room, caught on compact camera with flash, raw and unpolished","technical_tags":["top-down overhead shot","direct on-camera flash","CCD compact camera","high ISO digital noise","visible noise grain","blown highlights","harsh flash lighting","early 2000s digital photo","point-and-shoot aesthetic","NOT professional","candid snapshot"],"character_reference":"Based on user-uploaded character reference","clothing_reference":"Based on user-uploaded clothing reference","negative_prompt":"professional studio lighting, softbox, natural light only, side lighting, rim light, no flash, perfectly exposed, no noise, clean image, magazine editorial, too polished, AI look, plastic skin"}}Fix 2: Give it a camera, not an adjective
Only 12.0% of prompts in our library name a focal length, and only 6.2% name an aperture — while 11.1% ask for "shallow depth of field" and 5.4% ask for "bokeh". People request the effect and withhold the setting that produces it.
✕ Bad
shallow depth of field, blurry background, professional photo, bokeh✓ Good
85mm at f/1.8, focus locked on the near eye, her ear already soft, background wall six metres behind her dissolved into round highlightsA phone is a legitimate answer here, and often the better one. Naming a phone brings its whole signature: wide 24mm perspective, deep focus, aggressive processing, a little chromatic noise in shadows. That reads as a real snapshot in a way no "professional photography" tag ever will.
Stairwell Glance
Nano Banana ProPrompt▾
{
"meta": {
"quality": "ultra photorealistic",
"resolution": "8k",
"camera": "iPhone 15 Pro Max",
"lens": "24mm",
"aspect_ratio": "9:16",
"style": "dim stairwell lighting, raw iphone candid"
},
"scene": {
"location": "apartment stairwell",
"time": "night"
},
"subject": {
"action": "walking upstairs",
"pose": {
"one foot higher step": true,
"looking back": true,
"expression": "playful"
},
"outfit": {
"top": { "type": "tight crop sweater", "color": "grey" },
"bottom": { "type": "micro skirt", "color": "black" },
"legs": { "type": "black thigh-high stockings" }
}
},
"camera_perspective": {
"pov": "few steps below",
"angle": "slightly low"
}
}
Platinum Hair Duck Face Escalator Selfie
Nano Banana ProPrompt▾
{
"prompt": "Candid selfie photo from above, young woman with platinum silver white short wavy hair, looking up at camera making duck face kiss pout expression, light blue grey eyes, natural minimal makeup with pink nude lips, wearing cream beige tank top with thin straps, black leather shoulder bag strap visible, black hair scrunchie on wrist, silver ring on finger, white manicured nails, standing on escalator looking up, metal escalator steps visible below, another person partially visible behind in polka dot shirt, shopping mall or metro station setting, overhead high angle selfie perspective, natural indoor lighting, casual street style, social media influencer aesthetic, iPhone selfie style",
"negative_prompt": "studio lighting, outdoor nature, formal clothing, serious expression, side angle, full body shot",
"style": "candid street selfie, social media content, casual everyday moment",
"aspect_ratio": "3:4",
"camera": {
"type": "smartphone front camera selfie",
"angle": "high overhead angle looking down at subject",
"framing": "close up face and upper body from above"
},
"lighting": {
"type": "natural indoor ambient light",
"quality": "soft even lighting, no harsh shadows",
"atmosphere": "casual everyday indoor setting"
},
"mood": "playful, fun, casual, flirty, spontaneous",
"color_palette": "platinum white hair, cream beige, black accents, silver metal, neutral tones",
"subject_features": {
"hair": "platinum silver white, short shoulder length, soft waves, middle parted",
"skin": "fair, clear, natural, porcelain",
"eyes": "light blue grey, looking directly up at camera",
"lips": "natural pink, pursed in kiss pout duck face",
"expression": "playful duck face kiss pout, looking at camera",
"eyebrows": "natural dark brown, well groomed"
},
"wardrobe": {
"top": "cream beige ribbed tank top with thin spaghetti straps",
"accessories": "black leather shoulder bag with strap across shoulder, black fabric hair scrunchie on wrist, silver chunky ring on finger",
"nails": "white or light pink manicure"
},
"setting": {
"location": "escalator in shopping mall or metro station",
"background": "metal escalator steps, black handrail, another person behind",
"context": "casual everyday urban moment"
}
}Fix 3: Skin needs texture, and texture means flaws
This is where most portraits die. Untouched instructions produce retouched skin, because retouched skin is what the training data mostly contains. You have to ask for the imperfections by name.
✕ Bad
perfect flawless skin, beautiful face, smooth complexion, high detail✓ Good
visible pores across the nose and cheeks, faint freckles under the eyes, slight redness around the nostrils, natural shine on the forehead and chin, one flyaway hair crossing the temple, no retouchingThree specifics beat ten adjectives: pores, an uneven patch, and one stray hair. Add the environment's effect on skin — sweat, cold flush, wind, damp — and the face stops looking rendered and starts looking exposed to a room.
8K Finnish Sauna Goddess Portrait
Nano Banana ProPrompt▾
High-fidelity 8K cinematic portrait. Subject in ultra-luxury Finnish sauna, skin drenched in heavy glistening layer of sweat and condensed steam. Deep heat-induced flush across décolletage and cheeks, chest rising and falling from humidity. Thick white terry-cloth towel wrapped precariously low, damp and clinging to silhouette. Low-key amber lighting from beneath cedar benches creating sharp rim-lights on wet shoulders and collarbones. High-gloss lips slightly parted. 35mm f/1.4 lens, intense direct eye contact, volumetric steam swirling around body.
Orange Sunglasses Selective Color Portrait
Nano Banana ProPrompt▾
A studio-style close-up editorial portrait of a person with strong, well-defined facial features and slightly imperfect, natural skin texture. The subject wears a black tailored turtleneck with sharp, clean lines, layered under a high-collared black jacket in a minimalist contemporary fashion style. The subject wears semi-transparent orange acetate sunglasses — rectangular frames with softly rounded edges, glossy finish, and amber gradient lenses — serving as the only colored element in the image. Color concept: selective color photography — monochrome black-and-white image with only the sunglasses in vivid orange. Mood is calm and confident, serious expression, direct gaze into the camera. Lighting is soft frontal studio light with gentle shadows, even skin tones, cinematic contrast, and visible natural skin texture. Shot on a professional portrait camera, f/2.0, ISO 100, 1/125s. High resolution, ultra-sharp focus on the face. Style: editorial luxury fashion portrait, photorealistic, professional studio photography, no illustration, no painterly effects.
Fix 4: Break the symmetry
Symmetry appears in 5.1% of our corpus and centred composition in another 1.3%, and it is the single easiest tell to remove. Real photographs are asymmetric almost by accident: the subject sits in a third of the frame, the horizon leans a degree, a shoulder or a cup gets cut by the edge, and something in the foreground is closer to the lens than the subject is.
✕ Bad
a woman standing in a street, symmetrical composition, centered, looking straight at camera✓ Good
subject in the left third, turned three-quarters away, one arm cropped by the frame edge, a blurred shoulder of a passer-by in the near foreground, horizon tilted slightly rightNegative space helps too — 2.7% of the corpus asks for it — but the point is placement, not emptiness. Put the subject somewhere specific, and give the model something else to fill the rest of the frame with.
Urban Still Shadow Crowd Flow
Nano Banana ProPrompt▾
Ultra-realistic cinematic street portrait of a young woman standing still in a crowded city street, sharp focus on her face with calm, intense expression. Long-exposure effect with people moving around her creating strong directional motion blur, streaked crowd movement while the subject remains perfectly still. Natural soft daylight, shallow depth of field, creamy bokeh, realistic skin texture, subtle freckles, neutral makeup, dark winter coat and scarf. Emotional, introspective mood, urban storytelling photography, DSLR quality, 85mm lens look, f/1.8, high dynamic range, 8K, professional color grading.
Fix 5: Photograph a moment, not a person
The last tell is the hardest to unlearn. Most prompts describe a person standing there being attractive. Photographs, even bad ones, are usually of something happening: someone reaching, laughing, blocking the lens, mid-step, about to leave.
Only 4.8% of prompts in our corpus ask for a candid, and 2.8% for motion blur. Both are cheap to add and change the read immediately, because they imply a photographer who was slightly too slow.
✕ Bad
beautiful young woman standing in a convenience store, smiling at camera, high quality portrait✓ Good
caught mid-step as she turns back over her shoulder, one hand rising toward the lens, slight motion blur on the hand, shy half-formed smile, shot from chest height while walking
Tokyo Night Candid Capture
Nano Banana ProPrompt▾
Ultra-realistic candid smartphone photograph, 9:16 vertical. Nighttime on a quiet Tokyo city sidewalk beside a building with white ceramic tile walls. The subject is walking past the camera, partially side-on, captured mid-step in a rushed, accidental moment. The subject reacts instinctively to being photographed: turning her head back mid-walk as if surprised, raising one hand close to the lens to partially block the camera, shy, restrained smile forming as she suppresses laughter, body angled in half-profile while continuing to walk forward, motion carries her slightly out of frame, imperfect framing. Shot on a modern smartphone, handheld, rushed capture. Strong uncontrolled camera shake. Heavy directional motion blur across the raised hand, hair, and face. Flash-only lighting fired mid-motion. Harsh flash highlights on skin and hand. Deep surrounding darkness. Extremely noisy. Heavily blurred. Raw smartphone look.
Playful Convenience Store Selfie
Nano Banana ProPrompt▾
{
"portrait_prompt": {
"subject": "Woman with long straight hair, wearing a simple white sleeveless top and a delicate pearl necklace. She is leaning playfully forward over the handle of a black wire shopping cart in a dimly lit convenience store aisle. She has a cute, spontaneous expression, winking one eye and puckering her lips in a playful kissy face. She holds a smartphone in her hand. The aisle is lined with shelves stocked with colorful beverage cans and bottles.",
"composition": "Medium shot, vertical framing, 35mm focal length, capturing the depth of the supermarket aisle behind her. The composition feels like a candid snapshot taken by a friend.",
"camera_angle": "Slightly high angle looking down at the subject, creating a cute and intimate perspective.",
"lighting": "Direct on-camera flash style lighting (snapshot aesthetic) that illuminates the subject brightly while letting the background fall into shadow. Mixed with ambient overhead store lights visible as bokeh points.",
"color_palette": "Fujifilm Classic Negative simulation, strong film texture and grain, high contrast. Cool greenish-blue tones in the shadows typical of fluorescent store lighting, contrasted with natural skin tones. Vintage lo-fi aesthetic.",
"mood": "Playful, spontaneous, late-night urban vibe, authentic, fun, Gen Z aesthetic, 'grocery store photoshoot' trend.",
"negative_prompt": "Unrealistic features, anatomical errors, stiff pose, over-processed, harsh lighting, blurry, low resolution, bad composition, distorted faces, unnatural colors, extra limbs, watermark, text, bright daylight"
}
}
Kneeling Mirror Selfie in Soft Indoor Light
Nano Banana ProPrompt▾
{
"prompt_data": {
"subject": {
"appearance": "Young woman with light skin tone, blonde hair pulled back into a high ponytail with loose strands framing the face, rosy cheeks, visible small tattoo on the side of the neck/behind ear.",
"body_type": "Fit, curvy physique.",
"accessories": "Gold ring on ring finger, gold anklet."
},
"clothing": {
"top": "Black long-sleeved tight-fitting compression top or bodysuit.",
"bottom": "Black tight-fitting mini shorts (booty shorts).",
"footwear": "Barefoot (pink Crocs visible on the floor nearby)."
},
"pose_and_action": {
"posture": "Kneeling on a white fluffy rug, body turned in profile to the mirror, back arched fully.",
"action": "Taking a mirror selfie with a smartphone, right hand holding the phone, left hand resting gently on chest/collarbone.",
"gaze": "Looking at the phone screen/mirror reflection."
},
"environment": {
"setting": "Modern bedroom or dressing room.",
"furniture": "White modern shelving unit/drawers on the left, white vanity chair with gold legs in background.",
"decor": "White fluffy area rug (shag carpet), pink artwork/poster on the wall.",
"objects": "Pink Crocs on the rug, clutter on dresser (cups, toiletries), smartphone with a black case covered in planet/space stickers."
},
"lighting": {
"type": "Soft indoor diffused lighting.",
"source": "Natural light mix with room ambience.",
"quality": "Even illumination, soft shadows, flattering skin tones."
},
"styling_and_mood": {
"aesthetic": "Influencer lifestyle, casual at-home aesthetic, Gen Z style.",
"mood": "Relaxed, confident, candid, selfie mode."
},
"technical_specs": {
"camera_simulation": "iPhone rear camera or high-end mirrorless mimicking a mirror selfie.",
"focal_length": "24mm (wide angle)",
"aperture": "f/2.0",
"quality_tags": [
"Ultra Photorealistic",
"8k resolution",
"Raw photo",
"Hyper-detailed skin texture",
"Masterpiece",
"Sharp focus",
"Nano Banana Pro optimized",
"High fidelity",
"Depth of field"
]
}
}
}Do not stack the fixes with contradictions
Asking for a 24mm phone snapshot and f/1.4 background melt in the same prompt gives you neither — phones do not render that separation. Pick one camera story and let every other instruction agree with it.
A template you can paste and fill
Every fix above, in the order the model reads best. Replace the bracketed parts and delete nothing else.
Candid photograph of [subject: age, build, hair, clothing], [what they are
doing at this exact second — reaching, turning, laughing, leaving].
Location: [ordinary specific place], [time of day].
Light: [one source] from [direction], [hard or diffused], no fill on the
[other] side, [what the background is lit by].
Camera: [phone model or 35mm body], [focal length], [aperture], shot from
[height], subject in the [left/right] third.
Skin: visible pores, uneven tone, faint redness at [nose/cheeks], natural
shine on forehead and chin, one flyaway hair, no retouching.
Imperfections: slight motion blur on [moving part], [something] cropped by
the frame edge, focus locked on the near eye and falling off behind it.
Run it once, then change exactly one line at a time. If you change three, you will not know which one worked.
More prompts that already do this
Every card on this page opens to the full prompt in the prompt library — copy one, paste it into the generator, and change one line. If you are keeping the same face across several of these, the method is different and it is covered in the character consistency guide; if your realistic shot is going on a phone screen or a print, pick the ratio first.
FAQ
Why do my AI images still look fake?
Almost always one of five things: skin rendered without pores or unevenness, light with no single source, a dead-centre symmetrical composition, uniform sharpness across the whole frame, or a subject posing rather than doing something. Fix them in that order — skin first, it is the surface people read first.
Does adding "photorealistic" to a prompt help?
Barely. A quarter of the prompts in our library use it and more than half of those say nothing about skin, light or camera. It is a wish, not an instruction. Replace it with the physical details that would have made the photo real: a lens, an aperture, one light source, and pores.
Which AI model makes the most realistic images?
Our corpus does not run controlled tests, so treat this as what practitioners do rather than a benchmark: 45.5% of Nano Banana prompts carry camera parameters against 28.0% for GPT Image and 9.6% for Midjourney, and conditional instructions — the "keep this, change only that" kind — are where Nano Banana separates in our library. Midjourney prompts skew toward style words, which is a different job. See the model comparison for the wider differences.
How do I get realistic skin without it looking dirty?
Name the flaws small and specific: pores on the nose and cheeks, faint freckles, a little redness at the nostrils, shine on forehead and chin. Avoid global words like "rough", "textured" or "acne" — those get applied to the whole face at full strength. Realism is one or two visible imperfections, not a skin condition.
Should I add film grain?
Grain appears in about one in ten of our prompts and it does help, mostly because it breaks the plastic uniformity of the render. Use it as the last item in the prompt, ask for it fine, and do not combine it with "8K ultra sharp" in the same breath — the model will try to give you both and give you neither. If you have one good render of a face already, a reference image beats every text trick on this page: roughly 12% of the Nano Banana prompts in our library work from an uploaded image for exactly that reason.
LocalBanana Team
We run Nano Banana, GPT Image and other models side by side, and publish every prompt in our gallery. These guides are written from that corpus.
Updated @LocalBanana_io
