Transform into the Snow Sovereign — a shimmering ice-crystal royal outfit and crystal crown form in a palace of blue ice as the aurora glows. Upload a photo, you keep your own face. One continuous cinematic take, music only. 15s.
Watch a clip, read its prompt, make your own — free. Every link is remixable in one tap.
Transform into the Snow Sovereign — a shimmering ice-crystal royal outfit and crystal crown form in a palace of blue ice as the aurora glows. Upload a photo, you keep your own face. One continuous cinematic take, music only. 15s.
BECOMING THE DEMON OVERLORD. One continuous cinematic 20-second take, full realtime speed, an epic dark transformation that builds to a powerful reveal. (music only, mixed loud and full so the music dominates from the very first frame: a dark, menacing orchestral with a deep ominous choir, pounding war drums and a huge brass swell at the peak; original, no lyrics, no copyrighted melody) <a faint crackle of fire and a low rumble far under the music> @image1 keeps the exact same face and bare features as the reference the whole time — same eyes, same nose, same skin, only what is present in the reference — standing tall at the centre, powerful and commanding. Preserve their gender presentation exactly as shown. Use they / their / them only. SETTING: a vast dark gothic cathedral-hall of black stone, tall pillars, rivers of molten fire glowing in cracks across the floor, embers and ash drifting up, a deep red-black sky through broken arches. BEATS (20s), the whole figure transforming, each stage clear: 0–5s: @image1 stands in plain dark garb as embers gather around the body; heavy dark armoured plates form over @image1's own torso and shoulders, keeping their own build, a long tattered dark cloak settling behind. Camera: slow push-in. 5–10s: two huge curved horns grow up from the sides of @image1's head, and two massive dark leathery bat-like demon wings unfurl and spread wide from the back — the whole figure growing taller and more imposing as the wings open. Camera: ease back to reveal the full wingspan and horns. 10–15s: @image1 stands at full power, the dark armour laced with faint molten-glowing cracks; several winged stone gargoyle creatures swoop in from the shadows and circle around @image1, wings beating, orbiting their ruler. Camera: slow orbit around @image1 with the gargoyles wheeling past. 15–20s: @image1 spreads the demon wings wide, raises one hand, and the gargoyles wheel around in formation against the hellfire light; @image1 settles into a mighty commanding hero pose, ruler of the hall. Camera: rise to a wide low-angle hero shot, holding on the pose to the last frame. LOOK: ultra-photorealistic cinematic dark-fantasy, deep red-and-black infernal grade, real stone, metal, leathery wings and skin, embers and firelight, shallow depth of field, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
A 1950s DATE AT THE DINER. A charming 20-second montage shot to look like vintage 1950s Kodachrome home-movie film, full realtime speed, soft jump cuts, ending in a dance. (music only, mixed loud and full so the music dominates from the very first frame: an upbeat 1950s rockabilly / jive instrumental with a walking bass, bright guitar and a swinging, danceable rhythm; original, no lyrics, no copyrighted melody) <a faint diner chatter and hand-claps far under the music> TWO distinct people, both their identities kept EXACTLY from the references the whole time — @image1 is the star of the date with the exact face, hair and features of the first reference, playful and glamorous in a polka-dot 1950s outfit styled to suit their gender presentation as shown in @image1, their own hair exactly as in @image1; @image2 is their date with the exact face, hair and features of the second reference, in a cuffed white tee and jeans, their own hair exactly as in @image2. They are two clearly different people, never swapping faces or merging, and they are the only two people featured in frame. Preserve their gender presentation exactly as shown. Use they / their / them only. VINTAGE LOOK throughout: authentic 1950s Kodachrome film — warm saturated pastel colours, fine film grain, gentle flicker, slightly soft focus, nostalgic and cheerful. BEATS (20s): 0–5s: @image1 and @image2 sit across from each other in a red vinyl booth in a classic chrome-and-checkerboard diner on a date; @image1 waves over the server and a burger and fries are slid onto the table, both smiling at each other. Camera: slow push-in on the booth. 5–10s: @image1 and @image2 share one tall milkshake with two straws, leaning in close, and @image1 takes a sip and laughs. Camera: medium, easing in. 10–13s: @image1 hops up and drops a coin into the chrome jukebox; the music kicks in and @image1 snaps their fingers and spins round toward @image2. Camera: follow @image1 to the jukebox and back. 13–20s: @image1 and @image2 jive and swing-dance together in the middle of the diner — @image2 spins @image1 out and back, they bounce and step in sync, @image2 gives @image1 a quick lift, clothes flaring, other diners clapping from the booths — and they land together in a playful final dance pose. Camera: handheld, tracking the dance, holding on the final pose to the last frame. LOOK: nostalgic, glossy 1950s Americana grade, clean frame with no on-screen text, captions, logos or watermark.
BECOMING A VIKING WARRIOR — THE FULL ARMING, STEP BY STEP. One continuous cinematic take, full realtime speed, a slow fierce arming shown clearly and in order, each stage given its own time; every piece of gear is physically picked up by hand from a wooden bench and weapon rack and then put on over time. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a thunderous Nordic war instrumental with heavy drums, a low horn and a rising chant-like melody; original, instrumental only. 2. SFX — quiet, under the music: a cold wind, leather creaking, buckles clicking, the iron helm settling, and one big battle roar at the end. 3. VOICE — only the single wordless battle roar in the last beat; the warrior stays silent otherwise. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. TWO references of the SAME person: @image2 is the warrior BEFORE — their real face, how the clip begins, in a plain wool tunic. @image1 is the same person AFTER — the finished armoured Viking look the clip builds up to and ends on: copy the final gear, war paint and pose exactly from @image1 — the rounded iron knotwork nasal helm, the huge grey-brown wolf-fur mantle over a long dark cloak, dark leather armour, fur leg bindings and boots, the round planked wooden shield with iron rim and boss, the bearded axe, and three dark charcoal claw-stripes down each cheek. The face, eyes, nose, eyebrows, lips and identity are the same person in every frame from start to finish. HAIR: in every frame the warrior keeps their own natural hair colour and length from @image2; by the end, where it is long enough, it is worn in warrior braids with small metal beads under the helm, exactly as in @image1. SETTING (same as @image1): a snowy cliff edge above a dark grey fjord at misty dawn, wooden longships with glowing torches moored on the water below, drifting snow. Right beside the warrior stand a wooden weapon rack and a fur-draped bench that hold all the gear in view from the start: the folded wolf-fur mantle, the dark cloak, the leather armour, the iron helm, the round shield, the axe and a small clay bowl of charcoal paint. BEATS (25s), each piece visibly taken from the bench or rack by hand before it goes on, the look growing step by step from @image2 to @image1: 0–5s: BRAIDING, SEEN FROM THE FRONT: the warrior, with the face and own hair of @image2, faces the camera in a plain wool tunic, both arms raised and both hands reaching behind the head, working their own hair at the back of the head by feel; the hands stay behind the head the whole time while the face is the focus — a hard, fierce, dangerous stare straight into the lens, jaw set, eyes narrowed, the cold wind pushing loose strands across the face. Camera: frontal medium close-up slowly pushing in on the fierce face. 5–10s: WAR PAINT ON THE CHEEKS: the warrior dips two fingers into the clay bowl of charcoal paint, places the fingertips on the cheekbone just under the eye and drags them straight down the cheek toward the jaw, leaving three dark vertical claw-stripes on the left cheek, then dips again and draws the same three stripes down the right cheek; the stripes build up under the moving fingertips; forehead and eyebrows stay their own natural colour. Camera: tight on the face. 10–15s: the warrior picks the leather armour up off the bench and pulls it on, buckling it at the side; then lifts the dark cloak and swings it over the shoulders, then lifts the heavy wolf-fur mantle off the bench with both hands and settles it around the shoulders and neck, fastening it at the chest with an iron brooch. Camera: orbit around the shoulders. 15–19s: the warrior sits on the bench and wraps fur bindings around each shin with leather cords, one leg then the other, tying them off. Camera: push-in on the hands at the legs. 19–22s: the warrior lifts the iron knotwork nasal helm from the rack with both hands and settles it onto the head over their own hair, the nose guard sliding down between the eyes; takes the round shield from the rack onto the left arm and pulls the axe off the rack with the right hand. Camera: rise to a medium shot. 22–25s: HERO POSE exactly like @image1: the warrior plants both feet wide on the snowy cliff edge, crouches forward behind the round shield held up in front of the body, axe raised out to the side, and ROARS a huge battle cry straight into the lens, mouth wide open, cloak whipping in the wind, longships glowing behind. Camera: low frontal push-in to a close hero shot on the roaring face over the shield rim, holding to the last frame. LOOK: ultra-photorealistic cinematic, cold grey-and-steel grade, misty dawn light, shallow depth of field, real wool, fur, iron, wood, hair and skin, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
16:9 30 seconds. Seedance 2.5. Miniature tiny-person FX, photoreal live-action, cinematic macro, shallow DOF, giant indoor arcade. ONE person. No on-screen text. No logos. No narration. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build, gender presentation. Use they / their / them only. @image1 = FACE only. Photocopy this person. Do not copy studio wardrobe from the lock. They are PALM-SIZED, about 1:15 scale. A miniature person, exactly as in @image1, moving with quick athletic agility. Scene outfit: practical clothes that suit them for running and climbing, with a small backpack. Same size the whole clip. ARCADE: oversized claw machines, racing sims, basketball, air-hockey, giant plush, neon, prize shelves. Everything huge compared to them. 0–6s They sprint across enormous glowing arcade buttons, jump them, launch from a giant joystick, grab a hanging cable, swing, land on a racing cabinet. 6–12s Jump onto the steering wheel of a giant racing simulator as it accelerates. Hold on through the shake, jump off at the last moment. 12–18s Land on a huge air-hockey table. Massive puck races at them. They jump on it, ride across, leap the goal, land on oversized plush. 18–24s Inside a giant claw machine. Climb stuffed animals. Mechanical claw descends. They trampoline off a plush bear, grab the claw, swing out. 24–30s Land on a towering prize shelf, race across giant stuffed toys, leap onto a glowing neon sign above the entrance. Camera pulls far back to show their tiny scale. They raise both arms. Arcade lights flash. No end-card type. SFX: arcade beeps, buttons, footsteps, engine, air-hockey, plush, air, chimes. No spoken lines. Same person. Same tiny scale. Realistic physics. Never a studio still as frame 0.
16:9 30 seconds. Seedance 2.5. Cinematic photorealistic London travel vlog. Handheld, smooth. ONE person. No on-screen text. No logos. No dialogue. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build, gender presentation. Use they / their / them only. @image1 = FACE only. Photocopy this person. Do not copy studio wardrobe from the lock. OUTFIT (same the whole 30s): a warm everyday coat over casual travel clothes that suit them. Small rolling suitcase at the airport only. 0–5s Heathrow arrival. They walk out with a small suitcase, excited. Handheld vlog. London atmosphere. 5–10s Big Ben & Westminster. They walk near Big Ben, look at it, smile at camera. Quick selfie then wide of Big Ben and the Thames. 10–15s Tower Bridge. They walk across filming themself, skyline behind, tracking shot. 15–20s London street. Red buses, black cabs, historic buildings. Brief café stop, take a coffee, keep walking. 20–25s London Eye, golden hour. They stand beside it, happy vlog look at camera. Slow circle, skyline behind. 25–30s Thames at sunset. They walk, turn, smile. Aerial pull-back of London skyline. Warm gold. End. Same person. Same coat. Real London. Never a studio still as frame 0.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGE IS MANDATORY AND FINAL: The attached image outranks every word written below it: wherever the attached image and this text disagree, follow the image. ASSET MAP: @image1 — the tiny living person. Photocopy face, hair, skin, and the clothes on the still, exactly as @image1 shows them. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a living subject — no other people or extra hands enter frame at any point. They are a tiny real person, not a doll. SCALE REALISM: @image1 is only a few centimetres tall in a real back-garden flower bed at dusk — among real moss, pebbles, fallen petals, a spanning strand of spider silk, and two large blooms. Match lighting and colour grade between the tiny person and the surroundings so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. Unbranded: no logos, no brand names, no readable text on pots, petals or stones. PROP CONTINUITY: the mossy stone step, the scattered petals, the spider-silk strand spanning the gap between the two blooms, and both blooms themselves are all visible from frame 1 or clearly established before first use — nothing new appears later unexplained. CAUSALITY LOCK: every event has a shown, physical cause — the petals scatter because a real gust moves through the garden first (grass and moss stir before the petals lift); the stumble happens because a loose pebble shifts underfoot, visibly; the silk strand sways because the same recurring gust catches it mid-crossing, not randomly; every recovery is a direct physical reaction to its cause, never unexplained. PACING: real, energetic, continuous motion for the full duration — no calm establishing pause longer than 1 second. Each beat below must read as a distinct, clearly readable event, not a blur — slow down fractionally at the peak of each moment so the action is visible, then keep moving. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. Never freeze on the reference still. SCENE: a magical-feeling corner of a real garden at dusk, warm golden-blue light, soft natural bokeh from real distant light sources (not added sparkle effects). The tiny living person, built exactly from @image1, is already climbing when the shot opens. Whimsical live-action macro realism, shallow depth of field. Tiny real person — not a doll, not a statue. 0–2s: Already mid-climb up a mossy stone step, they reach the top just as a gust moves through the garden — grass and moss stir first, a beat before the fallen petals around them lift into the air. 2–4s: Caught in the drifting petals, they duck and shield their face with one arm as petals swirl past, one grazing their shoulder. 4–6s: Stepping forward through the thinning petal-fall, a loose pebble shifts underfoot — they stumble, catch themselves on the stem of the first large bloom, steadying in one motion. 6–8s: Looking up, they spot a single strand of spider silk spanning the gap to a second, taller bloom. They test it with one foot, grip it with both hands, and start crossing hand-over-hand above the drop. 8–10s: Midway across, the same recurring gust catches the strand — it sways hard, their feet slip free for a moment, swinging, before they hook a leg back over the strand and keep moving, shaken but steady. 10–12s: Reaching the far side, they haul themselves up the last stretch onto the second bloom’s open petal edge and stand there, breathing hard but upright — the full dusk garden, both blooms, the crossed strand, and the soft distant lights all visible in this final frame. STYLE: photoreal live-action macro, real garden at dusk, warm golden-blue light, shallow depth of field. Vertical 9:16. Tiny real person among giant blooms — not a doll, not a statue. AUDIO (native, synced, no BGM, no voiceover, no narration): rustling grass and moss, a soft gust of wind, petals fluttering past, a small stumble sound, the silk strand creaking and swaying, an effort grunt mid-swing, a relieved exhale at the top. No on-screen text, no watermark, no subtitles.
Photo-to-video. First frame is this exact still — photocopy it. Same child, same clothes, same camcorder, same dog, same ball, same place, exactly as in the still. Preserve their gender presentation exactly as shown. Use they / their / them only. Not Pixar, not 3D cartoon. 15-second 16:9. Sound ON. 1997 VHS home video: tape grain, slight tracking, warm overcast daylight. THE CHILD MUST MOVE. They do not stay planted on one spot. 0–3s: the child takes a small step sideways, camcorder up, following the puppy. 3–8s: the child walks a few steps after the dog, laughing, and says clearly: "Hey, look!" Camera bounces like a kid walking. 8–12s: the child kneels, then stands again, panning with the ball. An off-camera adult: "Don't go in the flower bed!" 12–15s: the child shuffles forward as the dog reaches the ball. Tape wobble. AUDIO: the child's line, the off-camera yell, the dog's yaps, outdoor ambience, footsteps. Keep the orange timestamp if it is already in the picture. No new titles. Duration 15s.
OFF-ROAD PET ADVENTURE — 26 seconds, one continuous flow, no montage cuts. A person and two animals cross four wild terrains in one vehicle, and the cat never once hesitates. CHARACTER LOCK — HIGHEST PRIORITY: @image1 is the person in the attached reference image, exactly as shown. Preserve facial identity, facial structure, eye shape and spacing, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build, body proportions and gender presentation exactly as shown. Do not beautify, slim, de-age or restyle. Preserve their gender presentation exactly as shown. Use they / their / them only. No other people in frame. Every reference to the person in this prompt is @image1 — no pronouns, no description. REFERENCE LOCK — the animals and the vehicle come from their references, never from this text. @image2 is the dog. Reproduce exactly the animal in that reference — same breed, coat, markings, ears and size — in every frame. Do not use the reference's background. @image3 is the cat. Reproduce exactly the animal in that reference — same coat, markings, face and size — in every frame. Do not use the reference's background. @image4 is the vehicle. Reproduce exactly the vehicle in that reference — same body shape, paint colour, wheels, and any roof rack and load exactly as shown — in every frame. Do not use the reference's background; keep its own badges and logos exactly as shown, add no other lettering, and do not reproduce the number plate. @image1, @image2 and @image3 are the only living beings in the film. Keep each as the same single continuous individual from the first second to the last: do not duplicate, split or replace any of them. Blocking is locked: @image1 drives, @image2 rides the front passenger seat, @image3 rides the seat back and the window sill. PHYSICS AND TEXTURE. Emphasise extreme, realistic physical texture, and make water, dust and suspension strictly obey real-world physics: the bow wave pushed up by the tyres, water sheeting off the flanks in unbroken curtains, spray beading on cold glass, wet paint sheen, loose scree sliding underfoot, dry sand thrown backward in arcs by running paws, salt haze in low sun. The vehicle's weight is visible in how the springs load and release. Road dust accumulates on the paint progressively across the 26 seconds and never resets. LOOK. Naturalistic documentary photography, not commercial gloss. Warm low sun against cool shadow, deep greens and slate greys, shallow depth of field on the animals. Real optical flare, water spots on the lens, subtle film grain. Retain real fine pores and skin texture on @image1, and real individual hairs and wet fur on the animals. CAMERA. Full-frame cinema camera. 24mm for terrain wides, 35mm inside the cabin, 85mm for animal close-ups, 100mm macro for the paw inserts. Vehicle-mounted and handheld with fine breathing shake. The camera travels with the vehicle and holds a consistent left-to-right screen direction throughout. Transitions are motivated only: motion-relay past a passing tree trunk, a whip-pan that lands on the next terrain, one match cut on the cat's eyes. Prohibit rigid cutting. Prohibit objects appearing out of thin air. PERFORMANCE. Restrained and real. The animals behave like animals: brief head turns, ear flicks, blinking against wind, weight shifting with the chassis. @image1 barely performs at all — the competence shows in the hands on the wheel. BEATS (26s): 0-4s [Departure] Cold blue dawn fog lying flat on a forest track, beading on the windscreen. Low tracking shot gliding alongside the driver's door at mirror height. @image3 steps in through the open window onto the seat back, claws flexing once into the fabric, tail curling around the headrest. <a small metallic click from inside the cabin>. End state: three heads in one row inside the cabin, the vehicle rolling forward into the fog. 4-9s [Lake] First sun cuts the mist off a still mountain lake, gravel shore. The camera relays out from the tree line onto the water and travels level with the vehicle. @image2 pushes its head fully out of the window, ears inverting and flapping back, jowls trembling, eyes squeezed to slits against the wind; @image3 sits on the sill beside it, whiskers forward, perfectly still. <wet gravel crushing under tyres, then water>. End state: the tyres part the shallow water and the reflection on the lake surface breaks apart in the wake. 9-14s [The River] A fast brown river shoving hard against the vehicle's flank, cold spray hitting the glass. The camera drops to water level, close to the door, looking up. The water climbs above the sill. @image3 leans down and presses one front paw flat against the inside of the glass, pupils narrowing, and does not retreat; the hands of @image1 stay loose and steady on the wheel, one thumb tapping once. <water hammering the underbody; the engine note holding deep and even>. Emotion analysis: the cat is not being brave — the cat is reading the machine, and the machine is not worried. This is the emotional centre of the film. End state: the vehicle lifts out onto the far bank and water sheets off both flanks. 14-20s [The Climb] A broken rock shelf on an alpine flank, loose scree sliding out from under the tyres, dust hanging in a low sun beam. Low angle from below and ahead, the vehicle filling the frame against grey stone. The suspension articulates and one front wheel lifts clear of the rock and sets back down. @image2 plants its forelegs wide with claws spread on the seat; @image3 wedges calmly into the corner of the seat back, ears flat but eyes open. <suspension loading, a single stone skittering off the edge into silence>. End state: the vehicle crests the ridge line and open sky floods the frame. 20-26s [The Sea] The descent lands on hard wet sand at low sun, the beach mirroring the sky, spray hanging in the light, a sea stack far out in the haze. The camera tracks with the running animals, then slows, falls behind, and stops. @image2 runs flat out across the wet sand with @image3 riding on its back, claws set lightly in the coat, sand thrown backward in visible arcs; @image1 walks out of frame-left toward the vehicle and lays a palm flat on the warm front fender. The animals arrive and settle onto the bonnet, the cat curling in against the dog's chest. <surf, wet paws on packed sand, hot metal ticking as it cools>. End state and final image: the vehicle in profile against the sea, the cat curled against the dog on the warm bonnet, the hand of @image1 resting on the fender, all three still, the sun dropping. Hold this frame to the last second. GLOBAL. Maintain throughout: one person, one dog, one cat, one vehicle — same identities, same animals, same paint and any roof load, consistent left-to-right travel direction, progressively accumulating dust. The light travels from cold dawn fog through hard midday alpine light to warm low sun at the sea: a single continuous day. One unbroken flow with motivated transitions only. Audio is diegetic only: engine, water, wind, gravel, surf, paws. (a single low sustained string pad entering under the river crossing and resolving at the sea, no lyrics, no drums) Pure video. No subtitles, no on-screen text, no added logo (the vehicle keeps its own badges exactly as shown), no watermark.
A reaction reveal in two shots. Three people at a bus stop react one by one to something the camera never shows, then a hard cut reveals it: the person standing astride two identical copies of the uploaded car, arms folded, riding them down a sunlit street. Upload ANY face and ONE photo of ANY car — the second shot builds a matching twin of it. 10s, ambient audio, no dialogue.
@image1 is the cat and the absolute identity lock. Reproduce this exact animal in every frame: their coat colour and pattern, fur length and texture, ear shape and set, eye colour and shape, muzzle, markings, whiskers and body build exactly as shown. Never substitute a different cat and never change their colouring. @image2 is the kitchen: same counters, same hanging racks, same light. ATTACHED IMAGES ARE MANDATORY AND FINAL. Wherever an attached image and this text disagree, follow the image. Handheld phone footage, one single unbroken take, filmed by a person standing in the kitchen@image2 exactly as it appears in @image2 — its own counters, fittings and utensils, with ladles hanging from a rack overhead, cooks in white jackets and tall toques working the line, a pan erupting in orange flame in the background, warm tungsten light. The camera opens already on the cat@image1, who is cooking. They sit upright on their haunches on the steel counter with both front paws wrapped around the handle of a long wooden spoon, stirring slow circles inside a big steel pot, steam rising into their face. They wear a tiny white chef's toque and a small white apron: the hat and apron fit to their own head and body naturally, whatever their size, build and fur, without altering the cat themselves. They are concentrating, completely serious. The camera operator's bare hand reaches in from the left and touches the cat@image1. First a flat palm stroking down their back, fur flattening under the fingers — they keep stirring, eyes narrowing. The hand comes back and pokes a fingertip into their cheek, pushing the fur in; their ears twitch flat. The hand comes a third time and tugs their ear between two fingers. That is the trigger. The wooden spoon drops from their paws and clatters into the pot. Their head snaps around, they scream with their mouth wide open, and they launch straight at the camera, a front paw slapping across the lens. Ladles clang and swing on the rack overhead, a pan tips and clatters, steam bursts, a cook flinches in the background. The camera whips sideways, the image smears, and it falls — hitting the tiled floor and skidding to rest at a tilt. Final beat: from floor level, lens pointed up, the image blurs then snaps into focus. The cat@image1 walks into frame and stops directly above the lens, chef's hat crooked, staring straight down into the camera in total silence. Whiskers twitch. Hold on that stare. Realistic domestic cat anatomy and real fur, paws gripping the spoon handle naturally, on all four paws when walking, never standing or walking on two legs. The phone doing the filming is never visible in frame. Raw amateur phone video: no stabilisation, autofocus hunting, motion blur, rolling shutter wobble, blown highlights from the kitchen lights, light grain. AUDIO: the roar of a working kitchen — sizzling, clattering steel, distant voices. Someone laughing as they poke, cut short by one loud furious cat scream and a crash of metal. Then sudden silence under the final stare. No music, no narration.
MONSTER ESCAPE #1 — PARKING GARAGE. One continuous cinematic take, 20 seconds, full realtime speed, a terrifying climax-of-a-horror-film chase. (music only, mixed loud and full so the score dominates from the very first frame: a pounding relentless chase rhythm, deep menacing drones, sharp stingers, and a huge terrifying hit and swell on the final creature reveal) <low mechanical groans, a distant metal clang, ragged breathing, wet heavy breathing, booming impacts on steel> @image1 is the person and the only human in frame, identity kept exactly from the reference the whole time. THE CREATURE — a huge, heavy, hunched predator, so tall that its head reaches the low concrete garage ceiling, forcing it to stoop and hunch down as it moves because it is too tall for the space; heavy slabs of muscle under slick, wet, grey-and-ashen mottled skin with veins and sinew showing through, hairless and glistening; a broad head with deep-sunken dead eyes and a heavy jaw that unhinges far too wide into uneven rows of jagged yellow-grey teeth; long thick arms ending in blunt claws; it moves with heavy, powerful, lurching, stooped strides and wet rattling breaths. Grounded, photoreal and realistically monstrous — matte wet flesh, dull natural colours, it looks like a real living nightmare animal caught on a security camera. Its design stays exactly consistent every time it appears. SPATIAL LOGIC — CRITICAL, keep it unambiguous: there is ONE red steel STAIRWELL DOOR. @image1 crosses to the STAIRWELL side and the creature is left on the GARAGE side. After the door slams, @image1 and the camera stay on the stairwell (near) side; the creature stays on the garage (far) side; ONLY the creature's hand and head come THROUGH the door toward @image1 and the camera. The creature always stays on the garage side until it breaks through, and there is only one creature. DOOR PHYSICS — the heavy steel door swings on its hinges realistically: @image1 pushes it open with both hands, runs through the opening, then swings it firmly shut behind them so it bangs closed into its frame and latches; the door stays a solid hinged slab the whole time, never bending, floating, sliding sideways or clipping through @image1. BEATS (20s): 0–4s: @image1 hurries through a dark half-lit underground parking garage — flickering fluorescents, a low concrete ceiling with pipes just overhead, concrete pillars, rows of normal-sized parked cars, oil-stained floor — then freezes at a deep wet growl. Camera: handheld, following @image1 from behind. 4–9s: the huge predator rises up hunched between the cars BEHIND @image1, its back scraping under the low ceiling; @image1 turns and sprints toward a red steel stairwell door at the far end, the creature loping after in a low stoop and gaining, looming in the background behind them. Camera: track backward ahead of @image1's run, the creature far behind them. 9–13s: @image1 grabs the door handle and pushes the heavy steel stairwell door open on its hinges, runs through the opening to the STAIRWELL side, then swings the door firmly shut behind them so it bangs closed into its frame — now @image1 and the camera are INSIDE the stairwell on the near side of the closed door, and the creature is shut out on the GARAGE side, on the far side of that same door. Camera: follow @image1 through the doorway, then whip around to face the closed door from the stairwell side, @image1 pressed back against the near wall in frame. 13–17s: heavy BOOMING impacts strike the door from the FAR garage side, denting the steel inward toward @image1; then a single huge clawed hand PUNCHES THROUGH the door from the far side, bursting into the stairwell and groping toward @image1 and the camera. @image1 is clearly on the near side, the hand reaching in toward them. Camera: hold on the door with @image1 on the near side. 17–20s: the creature tears the buckled door apart from the far side and forces its wide-splitting fanged jaws and sunken dead eyes through the broken gap into the stairwell, filling the frame inches from @image1's terrified face, to a hard black-out on the last frame. Camera: crane in to the close reveal of the maw through the gap. LOOK: ultra-photorealistic cinematic horror, 35mm handheld and 50mm close, shallow depth of field, cold sodium-and-fluorescent night grade, real wet textured grey monstrous flesh with dull natural colours, natural motion blur, film grain, photoreal throughout, clean frame with no on-screen text or logos.
ANCIENT EGYPT — BECOMING PHARAOH. One continuous cinematic take, 25 seconds, full realtime speed, a regal historical transformation. (music only, mixed loud and full so the score dominates from the first frame: an ancient-Egyptian-flavoured cinematic score — frame drums, plucked lyre and oud, a breathy reed flute and low horns, building through swelling strings to a regal orchestral peak on the final reveal) <soft crackle of torch flame, a faint chime of gold, a low resonant chord> IDENTITY — @image1 is the person, identity kept EXACTLY from the reference the whole time; keep the exact same face and natural features as the reference — same nose, same eyes, same skin, same face shape — only what is present in the reference. Only their adornment and styling change, never their identity. Preserve their gender presentation exactly as shown. Use they / their / them only. AUTHENTIC PTOLEMAIC-EGYPTIAN STYLING ONLY — real ancient tools and materials, using only the methods of ancient Egypt. JEWELLERY IS ALWAYS HAND-PLACED — every garment and every piece of jewellery is physically carried into frame in the attendants' hands and put onto @image1 with a clear, visible physical action: lifted up to them, wrapped around, clasped, or slid into place by their fingers. Each piece exists in an attendant's hands first and is then placed on their body, settling naturally with real weight. BEATS (25s): 0–4s: @image1 stands in a torch-lit sandstone chamber in a plain undyed linen tunic; their own hair is exactly as in @image1, ALREADY finished and dressed with a few small gold rings where its length allows, complete from the first frame and left untouched. Warm low torchlight. Camera: slow push-in. 4–9s: two attendants step in from either side holding the garments and jewels in their hands; they drape a finely pleated white linen robe over their shoulders, then one lifts a broad gold-and-carnelian wesekh collar up to their chest with both hands and fastens its clasp behind their neck, while the other slides carved gold armlets up each of their upper arms one at a time. Camera: slow orbit following the hands as they place each piece. 9–15s: an attendant lines @image1's eyes using a slender kohl rod drawn from a small alabaster kohl tube — a single bold matte-black line along the lash line, extended in one long tapered stroke toward the temple, authentic ancient kohl kept as a simple bold flat line; then a fingertip presses a little red ochre onto their lips and cheeks, kept to suit their gender presentation as shown in @image1. Camera: tighten to the face. 15–19s: an attendant lifts a golden uraeus circlet with a rearing cobra in both hands and lowers it onto @image1's head, settling it into place with deft fingers; @image1 opens their eyes fully — dark kohl-lined, commanding and serene — as faint gold dust drifts in the torchlight. Camera: ease back. 19–25s: @image1, now a full pharaoh-era ruler, turns to camera and settles onto a carved gilded throne with a slow, sovereign expression as the score peaks; torch flames flicker across the gold. Camera: slow push-in to the final portrait. LOOK: ultra-photorealistic historical cinematic, 50mm close and 35mm wide, shallow depth of field, warm torch glow against cool stone shadow, drifting gold dust, authentic finely pleated linen, broad gold wesekh collar, kohl-lined eyes with a touch of red ochre, their own hair with small gold rings, uraeus circlet, natural motion blur, film grain, photoreal throughout, clean frame with no on-screen text or logos.
VERTICAL 9:16 30 seconds. TikTok talking-head UGC. Handheld selfie, medium close-up, slight sway. Jump cuts. No full-screen app B-roll cuts — they stay on camera. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build, gender presentation. Use they / their / them only. @image1 = FACE only. Photocopy this person. Glasses only if they wear them in @image1. Their own top from @image1. Small black lavalier mic clipped to it. Phone in one hand. @image2 = THE APP on the phone. Photocopy this screen UI onto the phone when it is readable. SET: bright home office, wood-paneled walls, potted plant. 0.0–4.0s Lean in, eyebrows up, point at camera. “Okay, if you travel and you’re still buying SIM cards at the airport, stop. Watch this.” 4.0–11.0s Head-shake, eye-roll. “Last trip I landed in Vietnam, no data, standing in a line for a SIM card, trying to figure out which plan wasn’t a scam. Never again.” 11.0–21.0s Phone up next to the face. Screen shows @image2 as they tap. “Now I use this app. You type where you’re going, it shows me eSIM plans for that country, you pick one, buy it, and install it right in the app. Took me two minutes on my couch.” 21.0–25.0s Lower phone, count on fingers. “And it works alongside your normal SIM, so you still get texts from home.” 25.0–30.0s Smile, shrug. “Landed, connected, done. Just grab it in your app store. You’re welcome.” Same person. No extra people. Never a studio still as frame 0.
BECOMING A KYOTO STAGE PERFORMER — THE FULL DRESSING, STEP BY STEP. One continuous cinematic 25-second take, full realtime speed, a slow and historically accurate Kyoto stage-performer preparation, each stage shown clearly and in order, nothing skipped. (music only, mixed loud and full so the music dominates from the very first frame: a serene traditional Japanese instrumental with a plucked koto, a soft shakuhachi flute and gentle taiko, building gently; original, no lyrics, no copyrighted melody) <a faint wind-chime and soft room tone far under the music> @image1 keeps the exact same face shape and features as the reference the whole time — same eyes, same nose, only what is present in the reference — seated at a low wooden mirror stand. Preserve their gender presentation exactly as shown. Use they / their / them only. SETTING: a traditional okiya dressing room — tatami-mat floor, paper shoji screens, a low lacquered mirror stand (kyodai), an open lacquered makeup box, hanging paper lanterns and a vase of blossoms. BEATS (25s), slow and clear, hands performing each real step in order: 0–4s: @image1 sits in a plain white cotton under-robe; a hand smooths a thin layer of bintsuke wax over the face, then a wide flat brush lays an even coat of white oshiroi over the face. Camera: close push-in on the face. 4–8s: the same white oshiroi is brushed down the neck and chest, and at the back of the neck the brush carefully leaves the traditional bare W-shape at the hairline (the unpainted nape). Camera: move around to the nape, then back. 8–12s: a small brush paints the lips with red beni, and the eyes and brows are softly lined in red and black. Camera: tight on the face details. 12–16s: THE HAIR — a glossy black lacquered nihongami wig styled in a shimada is lifted with both hands and settled carefully onto the head, then decorative kanzashi hairpins and a patterned kanoko hair ribbon are slid into the hair one by one by hand. Camera: rise around the head to show the hairstyle clearly. 16–21s: THE DRESSING — a white nagajuban under-kimono is wrapped across the body, then a richly patterned silk kimono is layered over it and the collars folded neatly low at the nape; a wide brocade obi is wound around the waist and tied into a large drum knot (taiko musubi) at the back, secured with an obijime cord at the front. Camera: orbit from the back to the front to show the obi and layers. 21–25s: @image1 rises in the full stage-performer garb, takes small graceful steps, lifts a folding paper fan, opens it and turns into a final poised pose. Camera: ease back to a wide hero shot to the last frame. LOOK: ultra-photorealistic cinematic, soft warm lantern grade, shallow depth of field, real silk, lacquer, powder and skin, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
BECOMING THE STAR OF A BOLLYWOOD WEDDING — THE FULL PREPARATION, STEP BY STEP. One continuous cinematic 25-second take, full realtime speed, a slow and real wedding preparation where every item is physically held in hands, taken from a tray or box, and visibly worked onto the body over time — nothing ever appears on its own or jumps into place — ending in a joyful dance. (music only, mixed loud and full so the music dominates from the very first frame: a festive Bollywood instrumental with a driving dhol rhythm, bright strings and a joyful melody, building to a big celebratory dance peak in the final section; original, no lyrics, no copyrighted melody) <a faint warm courtyard ambience far under the music> @image1 keeps the exact same face and bare features as the reference the whole time — same nose, same eyes, same skin, only what is present in the reference — seated then standing at the centre. Preserve their gender presentation exactly as shown. Use they / their / them only. SETTING: an ornate sunlit courtyard draped with marigold-flower garlands, carved pillars, hanging lanterns, with a low table holding an open wooden jewellery box and a brass tray. BEATS (25s), each step a real visible process performed by hands, taking its full time: 0–5s: an attendant's hands lift the folded red-and-gold embroidered silk wedding outfit off the brass tray — a sari or a sherwani with a draped stole, whichever suits their gender presentation as shown in @image1 — and dress @image1 in it layer by layer, tucking the pleats and drawing the long drape up over one shoulder, smoothing the fabric — the outfit visibly going on layer by layer. Camera: slow orbit following the dressing. 5–9s: a hand holds a henna cone and draws looping mehndi patterns onto @image1's open palm and the back of the hand stroke by stroke, the reddish-brown lines appearing under the moving tip and building up into a full design. Camera: tight on the hand as the pattern is drawn. 9–13s: hands lift a gold head ornament out of the open jewellery box — a maang-tikka or a jewelled turban brooch, whichever suits their gender presentation as shown in @image1 — and set it in place at the hair or headwear; then gold earrings or ear studs are picked up and fitted one at a time, only if they suit @image1. Camera: push-in to the head and ears. 13–17s: hands lift a layered gold necklace from the box, bring it around @image1's neck and fasten the clasp at the back; then red-and-gold bangles or gold kadas are slid over @image1's hand onto each wrist one by one, several stacking up on each arm. Camera: follow the hands to the neck then the wrists. 17–20s: @image1 raises the finished hennaed hands and looks at them, then turns slowly in the full wedding outfit as rose petals drift down around them. Camera: ease back to a wide. 20–25s: @image1 breaks into a joyful Bollywood wedding dance — expressive hand mudras, wrists turning, a spin that flares the outfit, light bouncy footwork, bangles glinting — and lands on a radiant final dance pose with both arms raised. Camera: following the dance, holding on the final pose to the last frame. LOOK: ultra-photorealistic cinematic, warm rich festive grade, shallow depth of field, real silk, gold, henna and skin, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
DAY IN THE LIFE OF A SAFARI ADVENTURER — RIDING THE ELEPHANT. A warm montage with a golden savanna film look, full realtime speed, soft jump cuts between short scenes, full of movement. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: an adventurous African-percussion-and-strings instrumental with a warm, sweeping, uplifting melody; original, instrumental only. 2. SFX — quiet, under the music: a soft savanna breeze, distant birds, a splash of river water and one soft elephant trumpet. 3. VOICE — none; @image1 stays silent the whole clip. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. @image1 is the adventurer and the star of every scene, identity kept EXACTLY from the reference the whole time — same face, hair and features throughout — in a khaki explorer's outfit and a wide-brim hat. BEATS (20s): 0–5s: @image1 rides high on the back of a large African elephant that walks steadily across golden savanna grass, @image1 holding the harness and grinning with the wind in their face. Camera: sweeping alongside, rising to a wide. 5–9s: the elephant wades into a wide river and lifts its trunk to spray a glittering arc of water into the air, @image1 laughing and shielding their face on its back. Camera: tracking across the water. 9–13s: at a sunny wildlife sanctuary @image1 crouches beside a relaxed leopard lying in the shade and gently strokes its back and head, the big cat calm and content, @image1 smiling in wonder. Camera: slow push-in on the gentle moment. 13–16s: back riding the elephant, @image1 passes a group of giraffes and zebras and points toward them. Camera: over the shoulder past @image1 to the animals. 16–20s: the elephant stops on a ridge at sunset overlooking a herd on the plain below, and @image1 rises on its back and throws both arms wide against the blazing orange sky. Camera: crane up and hold on the pose to the last frame. LOOK: warm golden safari grade, soft film grain, sweeping and cinematic, clean frame with no on-screen text, captions, logos or watermark.
RULER OF THE WHITE WOLVES. One continuous cinematic epic-fantasy take, full realtime speed, like a scene from a big fantasy film. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a majestic Nordic-fantasy orchestral with deep drums, low horns, warm strings and a soaring choir-like synth pad building to a grand theme; original, instrumental only. 2. SFX — quiet, under the music: soft paws crunching through snow, a cold forest wind, and one long wolf howl at the end. 3. VOICE — none; @image1 stays silent the whole clip. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. @image1 is the ruler of the wolf pack, identity kept EXACTLY from the reference the whole time — same face, hair and features throughout — in a long dark wool cloak with a thick white fur collar over dark leather clothing. Preserve their gender presentation exactly as shown. Use they / their / them only. THE WOLVES — a pack of eight enormous white wolves, photoreal and lifelike like real animals: thick white-and-silver fur moving in the wind, calm pale-gold eyes, powerful shoulders. The pack leader is the biggest, the size of a horse, with a broad silver-white chest. All of them loyal and calm with @image1, their family. SETTING: a vast snowy pine forest at blue dusk, tall dark trees heavy with snow, soft snowfall, a full moon rising over a rocky cliff at the forest edge. BEATS (15s): 0–4s: wide cinematic shot — @image1 walks alone through the deep snow between the giant pines as the white wolves step out of the trees one by one on both sides and fall in beside @image1, walking with them. Camera: wide, slow dolly alongside showing the whole pack. 4–7s: the huge pack leader walks up and lowers its head to @image1, who smiles and presses a hand into the thick fur of its neck; the wolf closes its eyes calmly. Camera: slow push-in on the gentle moment. 7–11s: @image1 swings up and sits on the broad back of the pack leader, holding its thick neck fur, and the giant wolf lopes forward through the snowy forest at an easy steady pace, the whole pack running in formation behind, snow spraying up from their paws, cloak streaming. Camera: low tracking shot racing parallel alongside at the same speed. 11–15s: they burst out of the trees onto the rocky cliff edge under the huge full moon; @image1, still sitting on the pack leader, raises one arm high as all eight wolves lift their heads and howl at the moon together. Camera: crane up and pull back to a wide silhouette of @image1 and the howling pack against the moon, holding to the last frame. LOOK: ultra-photorealistic cinematic fantasy, cold blue moonlight grade with warm highlights on the faces, real fur, snow, wool and skin, shallow depth of field, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
16:9 30 seconds. Seedance 2.5. Early-2000s Los Angeles friends’ night, consumer MiniDV, handheld. THREE friends only. Fun, friendly, clothed. No sultry posing. No on-screen text. No logos. No dialogue. CHARACTER LOCK: @image1, @image2 and @image3 are exactly those people. Photocopy FACE and the outfit on each lock. Preserve their gender presentation exactly as shown. Use they / their / them only. Never recast. Never swap clothes. @image1 = PERSON 1. Photocopy this face AND this exact outfit. @image2 = PERSON 2. Photocopy this face AND this exact outfit. @image3 = PERSON 3. Photocopy this face AND this exact outfit. ONLY THESE THREE. No fourth friend. Background extras ONLY inside the club. LOS ANGELES early 2000s. Yellow cabs, neon, Y2K club. No smartphones. 0–4s GETTING READY — @image1 only, small apartment. Finishes getting ready in that outfit, a last mirror check, bag, smiles at camera. 4–8s PRE-GAME — living room. @image1 walks in to @image2 and @image3 in their exact outfits. Three only. Giggles, clink drinks, show outfits. 8–11s STREET — the THREE walk a neon LA street. Glance back and smile. 11–14s CAB — the THREE in a yellow-cab back seat. Wave, laugh. 14–23s CLUB — early-2000s LA lounge. The THREE dance together, sip drinks, laugh. Crowd extras here only, out of focus. 23–26s BATHROOM — disposable film camera. The THREE group snap: smiles, peace signs. Flash. Laugh. 26–30s WALK HOME — the THREE on a quiet LA street. One carries their shoes. Last over-the-shoulder smile. Same three. Same outfits. MiniDV shake. Never a studio still as frame 0.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGE IS MANDATORY AND FINAL: The attached image outranks every word written below it: wherever the attached image and this text disagree, follow the image. ASSET MAP: @image1 — the tiny living person. Photocopy face, hair, skin, and the clothes on the still (their own hair, outfit, jewellery and any bag exactly as shown in @image1 — nothing invented, nothing swapped). Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. SUBJECT LOCK — HIGHEST PRIORITY: The tiny subject in every shot is exactly what is in @image1, only shrunk in scale. Do not beautify, restyle, slim or redesign it. No other version of it appears anywhere in frame. NATURAL FLIGHT — HIGHEST PRIORITY: they are a living person riding the dandelion, not a posed still pasted onto it. Do not copy the seated cafe-chair pose from @image1 into the air. They nest in the hollow of the seed head like a soft nest: hips sunk into the seeds, knees loosely tucked, torso relaxed and slightly forward, face clear above the fluff rim from frame 1 — never buried in the seeds. Hair, sleeves, hems and any loose straps stream and lift in the real airflow. On every tilt they lean with gravity; on every dip their body sinks a beat then recovers with the puff. Soft living micro-motion the whole time — blinks, looking where they are going, breath in the shoulders. Not a doll, not a statue, not a frozen smile, not hovering above the puff, not glued on top. HANDS: if the subject has hands, they are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. Hands rest loosely in their lap or on the seed rim — they do not grip, hold or manipulate anything. SINGLE SUBJECT: only the @image1 subject rides the dandelion as the story’s subject — a single real moth may appear as natural wildlife in the scene, not as a character, and no other people, figures or hands enter frame. They are a tiny real person, not a doll. PHOTOREAL LIVE-ACTION: real camera, real garden at dusk, real light — not CGI, not a painting, not an animated illustration. Match grain, colour and lighting between the subject and the dandelion so the whole shot reads as one photograph. SCALE REALISM: the subject, nested in the hollow centre of a single dandelion seed head, is a few centimetres tall relative to the real garden below — blades of grass, a low garden fence, and a ring of mushrooms at a tree’s base all read at true towering scale beneath it. Match lighting and colour grade between the subject and the dandelion throughout so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. Unbranded: no logos, no brand names, no readable text on seeds, grass or mushrooms. PROP CONTINUITY: the dandelion seed head carrying the subject, the grass blades and low fence below, and the ring of glowing mushrooms at the destination tree are all visible from frame 1 or clearly established before first use — nothing new appears later unexplained. CAUSALITY LOCK: every event has a shown, physical cause — the dandelion tilts because a real gust of wind visibly bends the grass below it first; it dips suddenly because a real moth’s wingbeat disturbs the air close by, visible a beat before the dip; it rights itself and descends only because the wind settles and gravity brings it down, a visible physical process, never an unexplained recovery. PACING: real, continuous motion for the full duration — no calm establishing pause longer than 1 second except the brief settled beat at landing. Each beat below must read as a distinct, clearly readable event, not a blur. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. Never freeze on the reference still. SCENE: a real garden at dusk, warm fading light low on the horizon, cool blue shadow rising from below. The tiny living person, built exactly from @image1, is already airborne, nested in the dandelion hollow, rising fast past the grass blades when the shot opens. Whimsical live-action macro realism, shallow depth of field. Tiny real person — not a doll, not a statue. 0–2s: Already rising fast, the dandelion carries them up past the grass blades and over the low garden fence. Their hair and clothing hems stream in the airflow; face readable above the fluff; hips nested in the seeds. 2–4s: A real gust bends the grass below first, then catches the dandelion — it tilts hard to one side. They lean with it, hair flattening with the fluff, then the puff levels and they settle back into the hollow. 4–6s: A real moth flutters close, its wingbeat disturbing the air — the dandelion dips suddenly. Their body sinks with the drop, then the puff catches air and they recover with it. 6–8s: Steadied, it drifts higher through a loose scatter of real fireflies, their light catching the fluff strands and the moving hair as it passes through them. 8–10s: Below, the garden opens into a ring of glowing mushrooms circling the base of a large tree — they look down toward the destination coming into view. 10–12s: The dandelion settles gently down at the edge of the mushroom ring, fluff coming to rest, they still nested and living at its centre, soft mushroom-glow lighting the whole scene — the full miniature garden world visible in this final frame. STYLE: photoreal live-action macro, real garden at dusk, warm fading light, shallow depth of field. Vertical 9:16. Tiny real person nested in a giant dandelion seed head — not a doll, not a statue, not a posed still. AUDIO (native, synced, no BGM, no voiceover, no narration): a steady breeze, fluff rippling and rustling, a soft gust gusting past, a faint moth wingbeat, gentle settling as it lands, soft ambient dusk insects at the end. No on-screen text, no watermark, no subtitles.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGE IS MANDATORY AND FINAL: The attached image outranks every word written below it: wherever the attached image and this text disagree, follow the image. ASSET MAP: @image1 — the tiny living person. Photocopy face, hair, skin, and the clothes on the still, exactly as @image1 shows them. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a living subject — no other people or extra hands enter frame at any point. They are a tiny real person, not a doll. SCALE REALISM: @image1 is only a few centimetres tall on a real windowsill among full-size objects (a small potted plant, a window-blind cord, the sill’s edge dropping to the floor below). Match lighting and colour grade between the tiny person and the surroundings so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. Unbranded: no logos, no brand names, no readable text on the plant pot, tag, blinds or sill. PROP CONTINUITY: the plant pot, the hanging blind cord and the drop to the floor are all visible from frame 1 — nothing new appears later unexplained. CAUSALITY LOCK: every event has a shown, physical cause — a gust from the cracked-open window knocks a small paper tag off the sill first, visibly, before they react to it; their slide down the cord happens because they grab it mid-fall, not instantly; the catch at the bottom is a direct physical response to the tag’s falling speed, not a lucky unexplained grab. PACING: real, energetic, continuous motion for nearly the whole duration — no calm establishing pause longer than 1 second. Each beat below must read as a distinct, clearly readable event, not a blur — slow down fractionally at the peak of each moment so the action is visible, then keep moving. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. Never freeze on the reference still. SCENE: a windowsill at dusk, warm indoor light mixing with cool blue evening light through the glass. @image1, only a few centimetres tall, is already moving the instant the gust hits. Whimsical live-action macro realism, shallow depth of field. Tiny real person — not a doll, not a statue. 0–2s: A gust through the cracked window knocks a small blank paper tag off the sill edge — they, standing nearby, immediately turn and sprint after it as it tumbles off the edge. 2–4s: At the sill’s edge, with no time to climb down safely, they leap and grab the hanging blind cord, swinging out over the drop. 4–6s: Sliding fast down the cord, both hands gripping tight, passing the plant pot’s rim at speed — a near-miss as a leaf brushes past their shoulder. 6–8s: Lands hard on the floor below, rolls once to absorb the impact, and scrambles forward just as the tag flutters down — catching it in both hands a breath before it hits the ground. 8–10s: Stands up straight on the floor, tag held up triumphantly, looking back up at the long drop they just came down, the windowsill towering above them. STYLE: photoreal live-action macro, dusk windowsill, warm indoor light mixed with cool evening window light, shallow depth of field. Vertical 9:16. Tiny real person on a giant windowsill — not a doll, not a statue. AUDIO (native, synced, no BGM, no voiceover, no narration): a soft window-gust whoosh, quick tiny footsteps, the cord creaking under grip, a brief rush of air during the slide, a soft thud on landing, a satisfied exhale at the catch. No on-screen text, no watermark, no subtitles.
VERTICAL 9:16 22 seconds. Handheld selfie UGC in an SUV, driver’s seat. Hard cuts. ONE person. No on-screen captions. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build, gender presentation. Use they / their / them only. @image1 = FACE only. Photocopy this person. Do not copy studio wardrobe from the lock. Casual car clothes: plain t-shirt, no hat. @image2 = THIS drink, FRONT. Photocopy the can exactly as in @image2, its colours and label as printed. NEVER a different can. @image3 = THIS drink, BACK / nutrition label. Photocopy the label panel. NEVER invent a different label. They hold the real @image2 can whenever the can is in frame. 0.00–6.00 Driver’s seat selfie. Holds @image2 up to camera. ON CAMERA: “All right, hot take here, but this is way more important than coffee in my cup holder in the morning. Like, I don’t leave without it.” 6.00–9.00 Hard cut. Close-up of @image3 nutrition label, readable. ON CAMERA: “And look at what’s in it.” 9.00–13.00 Speaks to camera, then cut to close-up of @image2 front of the can. ON CAMERA: “It’s so good and it just hits every time. I literally bought a whole case again last night.” 13.00–22.00 Points at the label on the can, then talks to camera. ON CAMERA: “See that? That’s my drink, and I reach for it every day. Don’t sleep on this. If you’re looking for your new favourite, this is it right here.” Same person. Same can as the photos, both sides. SUV interior. No extra people. Never a studio still as frame 0.
ATTACHED IMAGES ARE MANDATORY AND FINAL. They outrank every word written below them: wherever an attached image and this text disagree, follow the image. @image1 is the absolute identity lock for the cat — its exact coat colour, pattern and markings, face, eyes, ears, build and its white apron, with no drift in any frame. Never invent a different animal. @image2 is the absolute lock for the drink — the same vessel, the same contents, the same colour and the same fill level in every serving on the tray. Handheld vertical phone footage, documentary look, inside a traditional Balkan kafana tavern. The cat@image1 walks upright on its hind legs like a waiter. One front paw is raised high above its head, balancing a round metal tray that carries exactly five identical servings of the drink@image2, every one matching the reference exactly. The camera tracks alongside at cat height, low to the wooden floor, bouncing slightly with each step, table legs and chair legs rushing past, the cat clear of furniture and never hidden behind it. Around it a warm dim kafana: dark wood panelling, long tables under red-and-white checkered tablecloths, bentwood chairs, framed old photographs on the walls, amber pendant lamps. The guests at the tables turn to look, laughing and clapping, a few lifting phones to film. In the final beats the cat reaches an occupied table, slows, brings both front paws to the rim of the tray and slides it flat down onto the checkered tablecloth, then lets go and lowers its paws; the five servings rock once and settle, the glasses clink, and the guests burst out laughing. Realistic fur detail, warm tungsten interior light, shallow depth of field, natural handheld shake, normal speed. AUDIO: kafana ambience — overlapping chatter and bursts of laughter, chairs scraping, glasses clinking as the tray lands, faint live tamburitza music underneath. No voice-over, no music sting.
A two-step reaction film. Step 1 places @image1 in a dim museum gallery beside a colossal Renaissance portrait, pointing and grinning — one still that fixes the face, the set and the wardrobe. Step 2 takes that frame as its opening shot and runs in four beats: the gallery lights cut, the painted eyes ignite red, the canvas bursts and a chrome cyborg strides out, then a tight recoil on @image1 with the cyborg's arm still in frame.
VERTICAL 9:16 30 seconds. Handheld selfie GRWM, close on the face. Hard cuts. ONE person. Light grey walls. No on-screen captions. No brand names spoken. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build. Preserve their gender presentation exactly as shown. Use they / their / them only. @image1 = FACE only. Photocopy this person. Do not copy studio wardrobe from the lock. WARDROBE: a simple plain top. Any jewellery exactly as in @image1. Hair DONE — styled, not wet, not in a towel. Skin exactly as in @image1, fresh and healthy-looking, lashes naturally defined. SHOW EACH PRODUCT TO CAMERA BEFORE THEY USE IT. @image2 = THIS tinted sunscreen. Photocopy it exactly as in @image2, label as printed. NEVER a different product. @image3 = THESE brightening drops. Photocopy the bottle. @image4 = THIS concealer. Photocopy it exactly as in @image4. @image5 = THIS lip liner pencil. Photocopy it. @image6 = THIS lip balm. Photocopy it exactly as in @image6. TALKING vs VOICEOVER: 0.00–4.00 INTRO — they TALK to camera, lips match. 4.00–21.00 VOICEOVER — lips still, they do the steps. 21.00–30.00 they TALK again, lips match. 0.00–4.00 Smiles at camera, hair already done. ON CAMERA: “Hey everyone. My favorite kind of makeup look is skincare makeup, and I’m gonna show you how I do it.” 4.00–15.00 FIRST hold @image2 and @image3 up to camera, both readable. THEN mix BOTH on the FACE: apply the @image2 tinted sunscreen onto cheeks the way its own packaging dispenses it AND add the drops onto the same spots so both products are visibly on the skin at once, then rub in. VO: “I mix this tinted sunscreen with these brightening drops. It gives such a luminous, glowy effect. This is what I do when I want that skincare makeup look—it just evens out my skin tone.” 15.00–21.00 FIRST hold @image4 to camera. THEN a tiny dot on the chin, blend with a finger. VO: “If I am breaking out, I just add a little dot of concealer and blend it with my finger.” 21.00–26.00 They TALK. Hold @image5 to camera, then draw: ON CAMERA: “Then just a little lip shaping” Then hold @image6 to camera and apply: ON CAMERA: “and some lip balm.” 26.00–30.00 Adjusts hair, glowy skin, smiles. ON CAMERA: “And this is it. Very light, still glowy, but more on the skincare side than makeup.” Same person. Same products as the photos. No extra people. Never a studio still as frame 0.
GHOST APPARITION #2 — THE POSSESSION. One continuous eerie cinematic take, full realtime speed, a connected build that resolves at the very end. (music only, mixed loud and full so it dominates the entire soundtrack from the very first frame: cold ethereal strings and a ghostly choir shimmer rising through swelling dread and warping demonic drones to a peak, then softening and clearing as it passes at the end) <a low spectral wind, a rising whisper, a bone crack, then a soft release and a settling breath> @image1 is the person and the only person in frame, identity kept exactly from the reference the whole time. SETTING — a dim candlelit room at night, old furniture, drifting mist, cold moonlight through a window, real depth. BEATS (20s), one connected build that resolves: 0–4s: @image1 stands alone in the dim room, then their whole body suddenly stiffens and jerks violently on its own, head snapping sharply to one side, as the candles flare and objects on the shelves tremble. Camera: slow push-in. 4–8s: @image1's body rises up off the floor on its own, arms falling limp, head tipping back, hair and clothing streaming upward as a cold wind swirls around them. Camera: crane up with the rising body. 8–12s: @image1's body arches into a deep unnatural backbend in mid-air, fingers splayed and rigid; the furniture slides across the floor and the candles gutter. Camera: slow orbit around the levitating figure. 12–16s: @image1's head snaps forward and their eyes roll back to solid, flat, blank white — no pupils and no irises at all, dull and matte with no glow or light coming from them; the possession peaks at its most intense. Camera: tighten on the face. 16–20s: it passes all at once — @image1's body sinks gently back down to the floor, the white clears from their eyes as their own normal pupils and living face return; they straighten up, themselves again, breathing hard but calm, as the mist clears and the candles steady. Camera: ease in on the restored, normal face, held to the last frame. LOOK — eerie ultra-photorealistic cinematic ghost-horror, 35mm and 50mm, shallow depth of field, cold blue moonlit grade with warm candle glow, drifting mist, real spectral refraction and cold rim glow, natural motion blur, film grain, photoreal throughout, clean frame with no on-screen text or logos.
HAUNTED HOUSE TOUR — THE BANQUET OF THE DEAD. One continuous eerie cinematic take, full realtime speed, a connected build to one loud terrifying climax. (music only, mixed loud and full so it dominates the entire soundtrack from the very first frame: creeping strings and a warped waltz motif rising through swelling dread into a violent orchestral stinger at the final scare) <a faint clink of silverware, a slow chair scrape, a cold wind, a sudden shriek> TWO SEPARATE PEOPLE — never merged, never the same face: • @image1 is the LIVING GUEST — the person in @image1, identity kept EXACTLY from the reference from the first frame to the last. Their face stays a NORMAL, LIVING, HUMAN face the entire time and NEVER changes, greys, decays, or turns into a skull or skeleton. They are an adult, fully clothed and decent, holding a lit candelabra. • The DEAD HOST is a DIFFERENT person from @image1 — a tall gaunt spectral figure in decayed black Victorian robes with a long tattered veil; the HOST's face is the skeletal one — a bare skull-like face with hollow black eye sockets and shrivelled grey skin. The skull face belongs ONLY to the host, never to @image1. SETTING — a grand candlelit dining hall in an abandoned mansion at night: a very long banquet table set with tarnished silver, cobwebbed candelabra and wax-dripped candles; rows of high-backed chairs holding silent seated corpse-guests in decayed formal Victorian dress with hollow eyes; faded tapestries, a great stone fireplace with dying embers, cold blue moonlight through tall arched windows, drifting dust motes and cobwebs; real, richly detailed depth. BEATS (20s), one connected build: 0–4s: @image1 pushes open the tall double doors and steps into the long hall, candelabra raised, their breath faintly visible in the cold; the seated dead are motionless, facing the table. Camera: slow push-in from behind @image1 down the length of the table, their living face lit warm by the candles. 4–8s: as they walk past, the gaunt seated corpses slowly turn their heads in sequence to watch them, jaws slack; they stop, frightened, their normal living face clear in the candlelight. Camera: track past the turning heads. 8–12s: the candles snuff out one by one down the table toward the far end; at the head of the table the tall DEAD HOST rises and LIFTS OFF THE FLOOR — floating upward into the air, hovering HIGH ABOVE the table with the host's feet dangling a full metre above the tabletop and the long robes trailing beneath, hollow eye sockets fixed on @image1. Camera: tilt up to reveal the airborne host well above the table. 12–16s: the DEAD HOST GLIDES THROUGH THE AIR high above the length of the table toward @image1 — drifting, weightless, floating, NEVER once touching, standing on, walking or running along the tabletop, the host's feet always off the surface — while the seated corpses rise in unison; @image1 turns and flees back toward the doors. Camera: whip and track their retreat. 16–20s: @image1 reaches the tall doors but they SLAM SHUT in their face; they spin around — the DEAD HOST is now directly in front of them, the host's skeletal SKULL-face and hollow eye sockets inches from @image1's terrified LIVING human face, the host's jaw wrenching open into a deafening shriek that fills the frame as the last candle dies to black. Throughout this, @image1's own face stays fully human, living and unchanged — only the host is skeletal. Camera: whip to the two faces side by side, snap to black. IDENTITY LOCK: @image1 (the living guest) and the DEAD HOST are two distinct separate people who appear together in the same frame at the climax. The skull / skeletal face is the HOST's alone. @image1's face stays a living human face the entire clip. Preserve their gender presentation exactly as shown. Use they / their / them only. LOOK — eerie ultra-photorealistic cinematic horror, 35mm and 50mm, shallow depth of field, cold blue moonlight and warm candle glow, drifting dust, real decayed Victorian and banquet detail, natural motion blur, film grain, photoreal throughout, clean frame with no on-screen text or logos.
Transform into a forest elf — pointed ears, leaf-and-silver braids, elven armour and a longbow in an ancient sunlit forest. Upload a photo, you keep your own face. One continuous cinematic take, music only. 15s.
Live a day as a billionaire sheikh — palace villa, pet tiger, falcon, supercars and a private jet, over-the-top luxury. Upload a photo, you keep your own face. Music only. 15s.
FASHION WEEK — THE GRAND RUNWAY FINALE. One sleek continuous take, full realtime speed, a huge, grand, star-making couture show. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a chic, pulsing electronic fashion-show beat with a cool confident groove building to a big triumphant hit; original, instrumental only. 2. SFX — quiet, under the music: rapid camera shutters, a rising crowd cheer and applause. 3. VOICE — none; @image1 stays silent the whole clip. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. TWO references with strictly separate jobs: @image1 provides the FACE, head, hair and identity — the headline star — kept EXACTLY the same the whole time (same face, bone structure, eyes, nose, mouth, skin tone, hairline), always centre-frame and in focus. Preserve their gender presentation exactly as shown. Use they / their / them only. @image2 provides ONLY the OUTFIT: take only the dress from @image2 — its exact shape, silhouette, length, colour, pattern, fabric and details — and fit it naturally onto @image1 for the whole clip. If @image2 shows a model's face or body, that is not used; the person is always and only @image1. @image1 is clearly the most important model in the room, wearing the @image2 dress. SETTING: an enormous grand fashion-show hall — a very long, wide, brightly illuminated glossy catwalk stretching far into the distance, towering lights and giant screens, a huge packed audience in glamorous outfits seated and standing along both sides, and banks of photographers crouched at the end. BEATS (20s), showing the full scale and that @image1 is the star: 0–4s: a wide high establishing shot reveals the vast hall — the enormous runway, the huge crowd on both sides, dramatic lights — as @image1 emerges through a grand backlit entrance at the top of the catwalk in the @image2 dress and begins to walk. Camera: high and wide, then easing down toward @image1. 4–9s: @image1 strides down the vast catwalk as the clear star, the fabric of the @image2 dress moving with each step, while several other elegant runway models walk in spaced formation behind and ahead of @image1 — @image1 stays centre and in front — as the crowd rises to their feet and hundreds of camera flashes fire from both sides. Camera: tracking backward ahead of @image1. 9–14s: the glamorous audience is on their feet applauding and filming on phones, photographers firing cameras up at @image1, who commands all the attention walking straight toward the lens. Camera: low hero angle looking up at @image1 with the crowd and flashes behind. 14–20s: @image1 reaches the end of the runway and strikes a powerful pose in the @image2 dress as the flashes peak and the other models fan out in a line behind; @image1 turns and leads them back up the catwalk. Camera: pull up and wide to reveal the whole grand scene, holding to the last frame. LOOK: ultra-photorealistic cinematic runway grade, crisp bright show lighting, shallow depth of field, real fabric, skin and crowds, natural motion blur, film grain, clean frame with no on-screen text, captions, logos or watermark.
DAY IN THE LIFE OF AN '80s MTV POP STAR. A glamorous high-energy montage shot to look like a real 1980s MTV clip on VHS tape, full realtime speed, packed with action, soft home-video jump cuts between six short scenes of the pop-star lifestyle. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a bright, upbeat 1980s synth-pop instrumental with a punchy drum machine, shimmering synths and a catchy hook, building to a big triumphant chorus; original, instrumental only. 2. SFX — quiet, under the music: a cheering crowd, popping camera flashbulbs, a clink of glasses, and a big stadium roar at the end. 3. VOICE — none; @image1's singing is shown visually only, the lips moving in time with the music. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. @image1 is the famous pop star and the star of every scene, identity kept EXACTLY from the reference the whole time — same face, hair and features throughout. Confident, radiant and glamorous, in bold sequined 1980s stage-and-streetwear with bright stage makeup, their own hair exactly as in @image1. VINTAGE LOOK throughout: authentic 1980s camcorder/VHS footage — soft slightly-washed neon colours, fine analog video grain and gentle tape noise, mild chromatic bleed, light handheld sway, slightly blown highlights. BEATS (25s), six glamorous scenes: 0–4s: @image1 performs on a neon soundstage, holding a chunky microphone right up to the mouth and clearly singing into it — the lips and jaw move expressively in time with the music, belting the song, eyebrows and face animated — dancing with backup dancers behind, coloured lights sweeping. Camera: push-in on the singing performance. 4–9s: @image1 is the celebrity in front of a barrier on a red carpet, leaning over it to sign posters and photos that a crowd of cheering fans hold out from the far side — @image1 stays on the star's side, grinning and slapping fans' hands across the barrier as flashbulbs pop. Camera: over @image1's shoulder to the reaching fans. 9–12s: @image1 clinks a champagne glass with glamorous friends at a glittering rooftop party at night, laughing and tossing their hair, city lights behind. Camera: handheld, easing in. 12–15s: @image1 strikes fierce poses at a glossy magazine photo-shoot, a wind machine lifting the hair as studio strobes fire. Camera: circling slightly. 15–19s: @image1 rides in the PASSENGER seat of an open convertible being driven by a chauffeur at the wheel; @image1 stands up through the open roof on the passenger side and waves to people on the street as the car cruises along, the chauffeur clearly steering. Camera: tracking alongside. 19–25s: @image1 performs on a huge concert stage before a massive roaring crowd and walks downstage straight toward the camera, then smoothly and naturally pivots on the spot to face the audience behind and throws both arms up as confetti rains — one continuous gradual turn, believable and smooth. Camera: stays with @image1 through the smooth turn, holding to the last frame. LOOK: nostalgic, glossy, aspirational 1980s MTV grade, clean frame with no on-screen text, captions, logos or watermark.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGES ARE MANDATORY AND FINAL: The attached images outrank every word written below them. Wherever an attached image and this text disagree, follow the image — including the product’s exact shape, features and proportions. ASSET MAP: @image1 — the tiny person. Photocopy face, hair, skin, and the clothes and accessories exactly as on the still. Ignore the grey studio pad around @image1 — that pad is not a set. Ignore any restaurant tables, chairs, windows or city skyline leftover in the still — those are not the location. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. The pose on the still is not the action. @image2 — the product, exactly as shown — same shape, colour, materials, label and logo; nothing invented or redesigned. Unchanged and not regenerated — same shape, colour, materials and hardware in every shot it appears. Ignore the grey studio pad and any thin white JPEG fringe around @image2 — those are not part of the product. No invented branding, text or extra hardware — keep the product’s own label and logo exactly as shown. No giant wearer or user. No other people, hands, or faces enter frame at any point. HANDS: if the subject has hands, they are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. The story does not require the subject to grip, hold or manipulate anything — only to be carried along by the water and ride out the motion. SINGLE SUBJECT: only the @image1 subject appears as a character in this scene — no other people, figures or hands enter frame. PHOTOREAL LIVE-ACTION: real camera, real drizzle, real water — not CGI, not a painting. Match grain, colour and lighting between the subject, the droplet and the product’s surface so the whole shot reads as one photograph. SCALE REALISM: the subject is a few centimetres tall relative to @image2, which stands at its true full size — like a towering landscape. A natural slope or curve on the product’s own surface forms the run the water follows. Match lighting and colour grade between the subject and the product throughout so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. PROP CONTINUITY: the slope on the product’s surface, a sharp curve partway down it, and a small ridge near the bottom — all taken from the product’s own real shape as shown in @image2 — are all visible from frame 1 or clearly established before first use — nothing new appears later unexplained. CAUSALITY LOCK: every event has a shown, physical cause — a real raindrop lands and gathers into a single large droplet at the top of the slope first, visibly, before it breaks free and starts rolling; the subject is swept along only because the moving water physically reaches and carries them; the brief airborne moment happens because the ridge is a visible bump in the product’s surface, not an unexplained jump; the splash at the end happens because the droplet and subject reach a real puddle that has visibly formed at the product’s base. PACING: real, continuous motion for the full duration — no calm establishing pause longer than 1 second except the brief settled beat at the very end. Each beat below must read as a distinct, clearly readable event, not a blur. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. SCENE: light drizzle falling onto @image2's real surface, overcast daylight, shallow depth of field. The subject, exactly as in @image1, is already perched at the top of the slope as a droplet forms nearby when the shot opens. Whimsical live-action macro realism. 0–2s: A single large droplet gathers from the drizzle right beside the subject, then breaks free and starts rolling down the product’s natural slope — the subject, caught at its edge, is swept along with it. 2–4s: Gathering speed together down the slope, water spraying off to the sides. 4–6s: The slope curves sharply — droplet and subject swing wide around the curve, water fanning outward at the turn. 6–8s: They hit the small ridge near the bottom — a brief airborne moment, both lifting off the surface together — before landing back on the slope, still moving fast. 8–10s: Reaching the bottom, droplet and subject splash together into a small puddle that has formed at the product’s base. 10–12s: Settled at the puddle’s edge, drizzle still falling softly around them, the subject sits or rests there, @image2 towering behind them — the full scale of the product landscape visible in this final frame. AUDIO: natural foley only — soft drizzle, water gathering and breaking free, a rushing trickle as it rolls and speeds up, a brief whoosh at the airborne moment, a soft splash at the bottom, gentle rain ambience settling at the end. No music, no BGM, no voiceover, no narration. No on-screen text, no watermark, no subtitles.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGE IS MANDATORY AND FINAL: The attached image outranks every word written below it: wherever the attached image and this text disagree, follow the image. ASSET MAP: @image1 — the tiny living person. Photocopy face, hair, skin, and the clothes exactly as in @image1. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. SUBJECT LOCK — HIGHEST PRIORITY: The tiny subject in every shot is exactly what is in @image1, only shrunk in scale. Do not beautify, restyle, slim or redesign it. No other version of it appears anywhere in frame. HANDS: if the subject has hands, they are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. The story does not require them to grip, hold or manipulate anything — only to ride, sway and settle with the boat’s motion. SINGLE SUBJECT: only the @image1 subject rides the leaf-boat as the story’s subject — a real frog and real fireflies may appear as natural wildlife in the scene, not as characters, and no other people, figures or hands enter frame. They are a tiny real person, not a doll. PHOTOREAL LIVE-ACTION: real camera, real miniature stream, real water, real light — not CGI, not a painting, not an animated illustration. Match grain, colour and lighting between the subject and the water so the whole shot reads as one photograph. SCALE REALISM: the subject, riding on a single large fallen leaf, is a few centimetres tall relative to a real miniature garden stream — the blades of grass on either bank tower like canyon walls, a lily pad ahead is the size of a small island. Match lighting and colour grade between the subject and the water throughout so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. Unbranded: no logos, no brand names, no readable text on leaves, petals or water. PROP CONTINUITY: the leaf-boat the subject rides, the grass-blade banks, and the overhanging curtain of petals partway downstream are all visible from frame 1 or clearly established before first use — nothing new appears later unexplained. CAUSALITY LOCK: every event has a shown, physical cause — a real frog leaps from the bank into the water first, visibly, before the ripple-wave it creates reaches the boat; the boat tips because that same wave physically strikes it, not randomly; it rights itself because the wave passes and the water settles, a visible physical process, never an unexplained recovery. PACING: real, continuous motion for the full duration — no calm establishing pause longer than 1 second except the brief settled beat at the very end. Each beat below must read as a distinct, clearly readable event, not a blur. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. Never freeze on the reference still. SCENE: a miniature garden stream at dusk, warm fading light on the water, grass blades rising like canyon walls on either bank. The tiny living person, built exactly from @image1, is already riding the leaf-boat downstream when the shot opens. Whimsical live-action macro realism, shallow depth of field. Tiny real person — not a doll, not a statue. 0–2s: Already racing downstream, the leaf-boat weaves between the grass-blade banks, a fine spray kicking up at its bow, the subject riding steady at its centre. 2–4s: Ahead on the bank, a real frog leaps into the water — the splash sends a ripple-wave racing toward the boat. 4–6s: The wave strikes — the leaf-boat rears up and tips hard to one side, the subject sliding toward the rim, before the wave passes and the leaf levels out flat on the water again. 6–8s: The current carries the boat under a low curtain of overhanging petals — they brush just overhead as the boat ducks beneath, emerging into open water on the far side. 8–10s: The stream widens into a still, glowing pool, real fireflies hovering low over the surface, their light scattered across the water. 10–12s: The leaf-boat drifts to rest against a large lily pad at the pool’s edge, the subject settled and steady at its centre, fireflies circling softly above — the full miniature river world visible in this final frame. STYLE: photoreal live-action macro, real garden stream at dusk, warm fading light, shallow depth of field. Vertical 9:16. Tiny real person on a giant leaf-boat — not a doll, not a statue. AUDIO (native, synced, no BGM, no voiceover, no narration): rushing water, a frog’s splash and the ripple that follows, water lapping against the leaf, a soft rustle as petals brush overhead, gentle water settling at the pool, faint insect hum at the end. No on-screen text, no watermark, no subtitles.
CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. Refer to them only as they/them in this prompt. ATTACHED IMAGE IS MANDATORY AND FINAL: The attached image outranks every word written below it: wherever the attached image and this text disagree, follow the image. ASSET MAP: @image1 — the tiny living person. Photocopy face, hair, skin, and the clothes exactly as they appear in @image1 ( their own hair exactly as in @image1 — never add a braid, ponytail or length it does not have; any jewellery or bag only exactly as in @image1). Ignore the grey studio pad and any gym, brick wall, banners, weights, or extra people behind @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a living subject — no other people or extra hands enter frame at any point. They are a tiny real person, not a doll. SCALE REALISM: @image1 is only a few centimetres tall on a real, full-size home-office desk among everyday objects (a stapler, a phone charger cable, a stack of books, loose coins). Match lighting and colour grade between the tiny person and the surroundings so the whole shot reads as one real photograph, not a composite. The scale contrast is clearly visible in the very first frame. Unbranded: no logos, no brand names, no readable text on the stapler, books, cable or desk. PROP CONTINUITY: the charger cable, the stapler and the stack of books are all visible from frame 1, placed across the desk — nothing new appears later unexplained. CAUSALITY LOCK: every obstacle has a shown, physical cause — the cable swings because it was already gently moving from a draft, not randomly; papers stir from a desk-fan gust a half-second before that same gust reaches them; their catch and recovery are a direct physical reaction to that gust, never an unexplained save. PACING: real, energetic, continuous motion for nearly the whole duration — no calm establishing pause longer than 1 second. Each beat below must read as a distinct, clearly readable event, not a blur — slow down fractionally at the peak of each obstacle so the action is visible, then keep moving. Do NOT default to a calm, mostly-still shot. Shot 1 must open already in motion. Never freeze on the reference still. SCENE: a cluttered home-office desk at warm late-afternoon light. @image1, only a few centimetres tall, is already running across the desk surface toward a stapler at the far end, weaving between obstacles. Whimsical live-action macro realism, shallow depth of field. Tiny real person — not a doll, not a statue. 0–2s: Already sprinting — they dash across open desk space, dodging a stray coin rolling slowly in their path, vaulting over it without breaking stride. 2–4s: A phone charger cable hanging off the desk edge swings into their path like a pendulum, already mid-swing from an earlier draft. They time it, duck low, and slide underneath just as it passes overhead. 4–6s: A sudden gust from an unseen desk fan hits — loose papers nearby flutter and lift first, a half-second before the gust reaches them. They stagger, grab the edge of a book to keep from being blown back, and hold on through the gust. 6–8s: Gust passes. They let go, climb the stacked books like steps, using the raised spine of the top book as a final foothold to reach the base of the stapler. 8–10s: They reach the top of the stapler — the highest point on the desk in view — and stand there, chest heaving, looking out across the whole desk landscape they just crossed. The full size contrast is clearly visible in this final frame. STYLE: photoreal live-action macro, warm late-afternoon desk lighting, shallow depth of field. Vertical 9:16. Tiny real person on a giant desk — not a doll, not a statue. AUDIO (native, synced, no BGM, no voiceover, no narration): quick tiny footsteps, a whoosh as the cable swings past, a gust of wind and papers fluttering, a grunt of effort at the book-grab, a relieved exhale at the top. No on-screen text, no watermark, no subtitles.
Stylized 3D animated film, exaggerated cartoon physics, squash and stretch. The attached first frame is the set, the cast and the flat unassembled pieces of THE PRODUCT. Every piece keeps the exact colour, material, texture, trim, hardware and markings it has in the first frame. THE PRODUCT is never redesigned, recoloured or relettered, and it never changes scale: it stays the same small piece of pet furniture the pieces belong to, in true proportion to the kittens. It must never grow into a shed, a cabin or a human-sized building. Beats are written for 10 seconds; at another length, scale every window in proportion. 0-2s - Two kittens plant their paws on the blueprint; one taps the drawing twice, another bats the chunky calculator. All of them look up from the paper toward the pieces on the grass. 2-4s - They haul the first upright panel into standing position, three kittens shoving it from behind while a fourth steadies the edge with both paws. 4-7s - The second panel is walked into place and joined into a standing corner, then the next piece is slotted in between them. THE PRODUCT now stands half-built. 7-9s - Two kittens carry the last large piece over their heads and lower it into place; another fits the last small piece into place, turning a screw with one paw. In the final moments, the kittens step back onto the grass and look up at what they have built. Camera: low and slow, a push-in over the blueprint, then a steady arc around THE PRODUCT as it rises. Warm golden afternoon light, green lawn, wooden fence and flowering shrubs behind. High-end CG animation, lively and slightly unreal. No added text anywhere - THE PRODUCT keeps its own markings and lettering exactly as shown; no lettering on the hard hats. No cardboard boxes or cartons anywhere in the frame. AUDIO: upbeat comedic orchestral score running the whole take, soft knocks and clacks as pieces meet, a couple of screwdriver ratchets, busy excited kitten meows, light garden ambience with birds. No dialogue and no human voice of any kind.
DUO DANCE — Rooftop Hip-Hop Audio: music only, mixed loud and full so the music dominates the entire soundtrack from the first frame: one continuous hard-hitting modern hip-hop / trap beat with a deep 808 bassline, crisp hi-hats and a catchy hook, high-energy and bouncy throughout. NO background or ambient sound of any kind, no footsteps, no foot taps or shoe-scuff, no foley, no recognisable copyrighted song or melody, no spoken dialogue, no invented language. IDENTITY LOCK — TWO distinct people: @image1 and @image2. Keep each one's exact face, bone structure, eyes, nose, mouth, skin tone, hairline, hairstyle, apparent age, build and proportions identical throughout — @image1 stays @image1 and @image2 stays @image2, two clearly different recognisable people, never swapping faces, never merging, never turning into each other or anyone else. The attached references outrank every word: where text and image disagree, follow the images. Both are adults and they are the only two people in the frame. Preserve their gender presentation exactly as shown. Use they / their / them only. WARDROBE — @image1 and @image2 each wear fitted, modern dance streetwear styled to suit their own gender presentation as shown in their photo; both stylish and modern, fully clothed and decent at all times, torso and legs covered, nothing revealing. THE DANCE — @image1 and @image2 perform an EXPLOSIVE, complex, non-stop DUO HIP-HOP routine packed with real moves, mostly in tight side-by-side sync with sharp moments of trading: a hard bounce and groove, the running man, quick two-step and skate footwork, body rolls and chest pops, snappy arm waves and hits, a shoulder shimmy, a spin, and a couple of trade-offs where one hits a freeze while the other rips a fast footwork solo, then they lock back into unison — dense, bouncy and precise, MANY moves flowing one into the next, every hit explosive and locked to the beat, never pausing, never simple. STAGE TIMING — always forward, one continuous routine packed with combos building in intensity: synced bounce and running man, footwork and body rolls, a trade-off solo each, a synchronized combo, a final duo freeze. Never reversing time. 0-3s — @image1 and @image2 explode in side by side with a hard bounce into the running man in tight sync, arms sharp; the camera holds a wide. 3-6s — the pair fire quick two-step and skate footwork with chest pops, perfectly synced; the camera tracks in. 6-10s — @image1 hits a sharp freeze pose while @image2 rips a fast footwork-and-wave solo, then they swap — @image2 freezes while @image1 rips a body-roll-and-arm-wave solo; the camera circles slightly. 10-13s — @image1 and @image2 lock back into a synchronized combo of body rolls, arm hits and a shoulder shimmy; the camera pushes in. 13-16s — the pair do a synchronized spin into snappy cross-arm hits and a bounce combo; the camera stays with them. 16-20s — @image1 and @image2 blast a final rapid synced combo and snap into a sharp duo freeze pose facing camera, on the last beat, held to the last frame. DIRECTION — one continuous energetic handheld take following the pair: wide and medium framings keeping both full bodies and feet visible, smooth tracking and a slight circle, easing in for the final freeze. Continuous screen direction; no cuts, no teleporting, no extreme shake. SCENE — a sunny rooftop at golden hour with an urban backdrop: a graffiti wall, a city skyline in the distance, warm low sun and long light, a lively street vibe with depth. A real outdoor setting with depth — never a plain grey or black empty studio void. Let each beat play out fully; ONE continuous coherent take, no cuts and no scene changes. CHOREOGRAPHY — believable, bouncy high-energy hip-hop physics: real bounce and weight, fast precise footwork, sharp arm hits, tight timing between the two both in sync and when trading; powerful, dense and clean, never floaty or glitching. ANATOMY — @image1 and @image2 each have exactly two arms and two hands with five fingers per hand, two legs and two feet; natural proportions and joints throughout, no distortion. LOOK — ultra-photorealistic cinematic with a warm golden-hour grade, 35mm and 50mm, shallow depth of field, realistic grain, natural motion blur, cinematic shutter. Never plastic, never cartoon, never CGI-looking.
Stylized feature-animation 3D commercial. Peak morning rush, total chaos in a packed specialty coffee bar, and one animal barista is barely keeping up. Fast, loud, energetic. [Reference] @Image 1 is a REFERENCE IMAGE, not a first frame. It defines the barista's appearance and nothing else: its species, build, coat colour and markings, head and ear shape, tail, eye colour and body proportions. Do not use the image's background, the pose it is in, its lighting, or its photographic realism. ANIMAL LOCK — HIGHEST PRIORITY: the barista in every frame is exactly the animal in @Image 1. Preserve its species, its coat colour, pattern and layering, its head and ear shape, its tail, its eye colour, and its body proportions, identically from first frame to last. Do not change it into another species or breed. Do not beautify or restyle it. No second animal, no other animals, no human staff behind the bar. Render it as a stylized 3D animated character with groomed fur and individual guard hairs — preserve what @Image 1 shows, not its photographic realism. This is stylized animation: the animal has the dexterity to grip and operate the tools with its own forepaws exactly as they appear in @Image 1. Never replace its forepaws with human hands, and never put gloves on it. Stage the bar, the machine and the grinder at the animal's own scale, whatever its size, so it can reach and work comfortably. [The emotion, translated into the body] It is losing the race and refusing to stop. Ears pinned back and flicking, eyes darting between the order rail and its paws, quick shallow breathing, mouth slightly open, fur ruffled along the shoulders, always doing two things at once. Never calm, never posed, never smiling at the camera. [Global setting] 07:20, a small specialty coffee bar jammed wall to wall with people. The walnut bar runs left to right: espresso machine left, grinder centre. THE CROWD IS IN THE FOREGROUND AND IN FOCUS — customers cross between the camera and the bar, lean over the counter, reach in from the frame edges, talk over each other. Bodies wipe the lens. Low winter sun through the street window rakes the bar as hard backlight; warm gold, cool shadow, steam and coffee dust hanging in the beam. THE ANIMAL IS THE ONLY ONE WHO TOUCHES THE COFFEE. Customers never hand it a cup, never place a cup on the bar, never pass drinks across the counter and never pay for anything on camera. Coffee only ever travels outward, from its paws. Camera: handheld, fast and restless — whip pans between the rail, its paws and the crowd; quick push-ins; one speed ramp into the pour. Deliberate hard cuts on the beats below. Cut logic: cut on action, always on a movement already in progress. The camera stays behind the bar for the whole film and never turns to the shopfront door. Prohibit rigid cutting, prohibit objects appearing out of thin air. Physics regime: steam, milk and water strictly obey real-world physics. Milk pours and folds as liquid, never as jelly or cloth. Coffee grounds fall as loose granular particles. Fur stays fur under fast motion — no plastic look, no smearing, no clipping through the counter. It wears only a worn canvas apron. [Beat 1 — the opening beat] Opening frame is already chaos: <the door bell SLAMS> and a wave of people pushes in behind the bar line. Whip-pan to the animal, ears flat, eyes wide, the order printer already spitting paper past its shoulder. End state: the shop full, the animal caught mid-turn. [Beat 2] The order rail. Blank thermal paper unspools in a long strip that has already reached the floor, kept soft and out of focus. It tears a handful off with one paw. Hands and elbows of customers push into frame from both edges. <Overlapping voices, half a dozen orders at once>. End state: a fistful of torn paper, the strip still growing. [Beat 3] Two things at once: one paw tamps a portafilter, the other reaches across to knock the machine lever. Its tail sweeps a spoon off the counter behind it. A customer leans right over the bar and talks at it; it does not look up. End state: portafilter tamped, spoon hitting the floor. [Beat 4] The grinder screams. Beans rattle and bounce; one ricochets off its nose and it blinks it away without stopping. A customer crosses fast through the foreground and blurs the whole frame. End state: basket filled, grounds scattered across the bar. [Beat 5] Milk. It snatches a steel pitcher, jams it under the steam wand and cracks the valve open. <A rising screech of steam>. The milk climbs and folds into a tight spinning vortex; its ears flick back at the noise but its eyes stay locked on the surface. The crowd still presses at the bar behind it. End state: the milk spinning glossy in the pitcher, steam rolling off the top. [Beat 6] It cuts the steam, knocks the pitcher twice against the walnut to burst the bubbles, and swirls it until the surface goes to wet paint. With its other paw it drags the waiting cup of espresso in front of itself. Crowd noise still at full. End state: glossy microfoam in one paw, dark crema in the cup in front of it, ready to pour. [Beat 7 — the closing beat] THE PAUSE, AND THE ENDING. Speed ramps down into slow motion and the noise drops away to almost nothing. Macro on the cup: the pitcher comes down close to the crema and the milk falls in — a rosetta opens across the surface, leaves spreading outward, the last stroke drawn back through the centre. The animal then sets the finished cup down on its saucer and pushes it forward to the front of the walnut, and holds there, looking into the lens with the crowd still packed behind it. The film ends on that frame. End state: the finished cup centred on the walnut with clean empty counter in front of it, the animal behind it, the shop still full. Beats 1 to 6 divide the first two thirds of the running time into six equal windows; beat 7 holds the whole final third by itself. [Audio] Dense and overlapping throughout: crowd chatter with individual voices cutting through, the door bell, the order printer, the grinder, the steam wand, cups on saucers, coffee machine hiss. (A driving, percussive music bed — tight, fast, kick-and-shaker, pushing the pace). Audio drops almost to silence for beat 7 and stays low to the end. No dialogue, no lyrics. [Maintain consistency] Same animal, same apron, same bar geometry, same light direction and colour temperature throughout. The espresso machine stays at the left end of the bar. The crowd never thins out — the shop is full in every single beat including the last. [Prohibitions] No customer ever hands the animal a cup, places a cup on the bar, or passes a drink across the counter — coffee moves outward from its paws only. No payment of any kind: no money, no banknotes, no coins, no card readers, no phones held out over the counter. No menu board, no price board, no chalkboard, no signage on walls or doors, no printed receipt in focus, no shopfront door in frame at any point. No rows of finished cups lined up on the counter. The cafe is never empty and never calm. The crowd is never a soft blur in the background — people are in the foreground, in focus, crossing the lens. No human staff, no second animal. No legible text anywhere. No latte art before beat 7. No milk behaving like jelly or fabric. No human hands replacing the animal's paws. No fur smearing or clipping. No subtitles, no on-screen text, no watermark. Prohibit rigid cutting, prohibit objects appearing out of thin air.
You're a tomb raider who grabs a cursed golden idol — and petrify from the hand up, setting into a temple statue among the others. Upload a photo, you keep your own face. One continuous cinematic take with a loud score. 15s.
You sprint through a dark snowy forest at night while THE ANTLERED — a towering elk-skulled creature — chases you down and lunges. Upload a photo, you keep your own face. One continuous cinematic take with a loud score. 15s.
Your pale spectral double — wearing your exact face — peels out of a fogged mirror behind you, drifts closer, then rushes straight at the camera. Upload a photo, you keep your own face. One continuous cinematic take with a loud score. 15s.