The prompt (copy, paste, render):
SUBJECT: A woman in her late twenties laughing mid-conversation at a small kitchen table, holding a chipped coffee mug, hair slightly messy.
ACTION: caught off-guard mid-laugh, head tilted, one hand rising slightly.
ENVIRONMENT: a cluttered sunlit home kitchen in the early morning, dishes in the background, a half-eaten breakfast, soft curtains.
MOOD: warm, candid, intimate, unposed — a real moment, not a photoshoot.
STYLE: imperfect realism, casual smartphone snapshot aesthetic, anti-glossy, authentic everyday photography.
LIGHTING: soft natural window light from the left, mild overexposure near the window, gentle shadows, no studio polish.
CAMERA: handheld phone-camera feel, slightly off-center framing, minor motion blur on the rising hand, eye-level, 26mm-equivalent wide phone lens.
TEXTURE: visible skin pores and faint blemishes, fine flyaway hairs, slight sensor noise, natural fabric wrinkles, fingerprints on the mug.
QUALITY: realistic, naturalistic, believable imperfection over clinical sharpness.
NEGATIVES: no airbrushed skin, no studio lighting, no perfect symmetry, no text, no watermark, no plastic over-rendered look.
Why this works:
For years AI gave everyone the same thing: flawless skin, perfect light, unreal clarity. And our brains learned to spot it instantly. "That's AI." So the fix is counterintuitive — you have to add flaws back in.
Look at the TEXTURE line. Pores, flyaway hairs, a fingerprint on the mug, a little sensor noise. None of that is pretty. All of it is what makes your eye believe it. Real cameras and real skin are messy, and the mess is the realism.
The ACTION line is pulling weight too. "Caught off-guard mid-laugh" beats "smiling at camera" every time, because posed faces read as stock photos and stock photos read as fake. A half-second of motion — that rising hand, a little blur — tells your brain this happened, nobody set it up.
Then the camera plays dumb on purpose. "Handheld, slightly off-center, phone lens." You're rejecting the tripod-perfect composition the AI wants to give you. Slightly wrong framing is what a real person's photo looks like.
And here's the one to remember: the NEGATIVES are the secret weapon today. "No airbrushed skin, no plastic look" actively shoves the model away from its default glossy instinct. Most prompts use negatives to remove junk. Here, they're removing the AI's personality — and that's exactly the point.
Remix it: Swap the kitchen for a bus window, a backyard, a messy desk. Change who's in it. But keep TEXTURE and NEGATIVES word-for-word — those two blocks are the whole illusion. The rest is just set dressing.