The golden rule of prompting a consistent AI character: describe the scene, never the person. Her face, hair, and build come from her reference library — your prompt's only job is to set the stage. These fifty prompts follow that rule. Copy, paste, tweak the details.
Everyday lifestyle
- "at a sunlit coffee shop, holding an iced latte, candid iPhone shot"
- "browsing vinyl records in a thrift store, soft window light"
- "laughing at something off-camera at a farmers market, golden hour"
- "reading a paperback on a fire-escape balcony, morning coffee beside her"
- "carrying a paper grocery bag with baguette poking out, city sidewalk"
- "blowing on a spoonful of ramen at a tiny counter restaurant"
- "mirror selfie in a hallway with warm lamp light, casual outfit"
- "sitting cross-legged on the floor wrapping a gift, fairy lights behind"
- "walking a golden retriever through a leafy neighborhood, autumn"
- "leaning out a car window at sunset, hair moving in the wind"
Fitness & wellness
- "sweaty in sports bra on yoga mat, post-workout glow"
- "gym mirror selfie in athleisure, morning light"
- "mid-stretch on an outdoor track at sunrise"
- "tying running shoes on a park bench, fall layers"
- "drinking a green smoothie at a kitchen counter, gym bag on chair"
- "boxing wraps on, leaning against the ring ropes"
- "rolling up a yoga mat outside a studio, chatting with someone off-frame"
- "cold plunge tub, bracing expression, steam in the air"
Travel
- "walking down a cobblestone European street with a tote bag, sunny morning"
- "leaning on a ferry railing, harbor and gulls behind"
- "ordering at a street food stall at night, neon signage"
- "airport window seat selfie, passport and boarding pass in hand"
- "standing under temple gates in light rain with a clear umbrella"
- "reading a map on a scooter parked by a coastal road"
- "beach towel, linen shirt, wind in hair, book face-down beside her"
- "hiking switchbacks, looking back over shoulder at the valley below"
Fashion & editorial
- "goth outfit, headshot but candid at an angle"
- "posing in front of a brick building, side wall, no signs"
- "monochrome streetwear against a concrete parking structure"
- "trying on sunglasses at a boutique mirror"
- "long coat, crossing the street mid-stride, shallow depth of field"
- "sitting on studio apartment floor surrounded by outfit options"
- "evening dress on a rooftop at blue hour, city bokeh"
- "vintage denim jacket, leaning against a mural wall"
Food & home
- "cooking a colorful salad bowl in a bright kitchen, smiling mid-laugh"
- "pulling a tray of cookies from the oven, oven mitt on"
- "pour-over coffee ritual on a wooden counter, steam rising"
- "picnic blanket flat lay with her reaching into frame for a strawberry"
- "repotting a monstera on a balcony, soil on hands"
- "lighting a candle on a dinner table set for two"
UGC & product-style
- "holding a skincare bottle up beside her face, bathroom light, talking-to-camera framing"
- "unboxing a package on a couch, genuinely surprised expression"
- "phone-on-tripod framing, mid-sentence gesture, ring light catchlights"
- "comparing two products side by side at a desk"
- "applying lip balm in a car visor mirror"
- "gym shaker bottle in hand, mid-conversation with the camera"
Seasonal
- "first snow, knit beanie, breath visible, streetlight glow"
- "carving a pumpkin at a newspaper-covered table"
- "sparkler at arm's length on New Year's Eve, bokeh crowd"
- "cherry blossom street, petals falling, light trench coat"
Three rules that make every prompt above work
- Add a camera cue. "Candid iPhone shot," "shallow depth of field," "ring light" — the camera language controls the realism register more than any other phrase.
- One scene per prompt. Stacked scenes ("at the gym then the beach") confuse the model. Generate twice — at $0.25 per image, variations are the cheap part.
- Never restate her appearance. "Blonde 25-year-old" fights the reference photos and reintroduces face drift. The references already won that argument.
Don't have a character yet? Mint your first for $19 and run this list top to bottom.