Drag the clip sideways
A fifteen-second candid handheld home video in a modern kitchen on a dark rainy morning. @image_1 fights a tightly sealed glass jar in three escalating attempts - bare hands, then a tea towel wrapped around the lid, then the whole body braced against the worktop - and when the lid finally pops, dips a fingertip in to taste it and turns to the camera with a small proud smile. Natural location sound only, no music.
A fifteen-second premium 3D animated fantasy short. @image1 plays with a small white feathered dragon hatchling on a floating island above a sea of clouds, an enormous green-and-gold dragon comes up over the rim, and after a run and a hiding place in the rock the giant turns out to be no threat at all - the film ends with the hatchling settling on @image1's shoulder at the island rim, both dragons and the whole cloud sea behind.
SUBJECT ANCHOR: The pet in the input image is the protagonist. Keep its fur pattern, markings and eye colour identical in every shot. The input image defines the look of the whole film. TONE: The subject is worried, never amused. Mouth closed, eyes wide with concern. Wistful → alarmed → urgent → tender. Shot 1 (3s) — Profile shot at the window, slow push in. The subject gazes through rain-streaked glass, then spots something far below. Snap to close-up: pupils dilating, ears back. Rain on glass. Shot 2 (3s) — Wide kitchen, camera whip-pans with the movement. The subject bolts through the kitchen in one burst, knocking pots and pans flying behind it. Ceramic crashes, rolling metal. Shot 3 (2s) — Low-angle tracking down the hallway, the subject sprinting toward lens. It grabs a bright red knitted scarf from a coat rack in its jaws. The scarf streams behind it in the air, never touching the floor. Shot 4 (3s) — Handheld, low, through a rain-slicked city street at dusk, neon on wet asphalt. The subject weaves between honking cars. Car horns, heavy rain. Shot 5 (4s) — Slow push-in to a static wide. Under an awning, a tiny shivering puppy. The subject gently drapes the red scarf around it, then curls up beside it. Rain softens, quiet warm ambience. NEGATIVE: no smiling, no scarf dragging on the ground, no text, no captions, no logos, no watermarks.
SCENE 1 0 to 4s — arrival. Crisp morning, light snow falling. @avatar steps out of a timber chalet doorway, breath visible in the cold. Camera trails from behind down a snowy cobblestone lane lined with wooden cottages, then swings round into a close-up as @avatar turns, catches the lens, and laughs, shaking their head slightly. SCENE 2 4 to 8s — first ski run. @avatar glides down a gentle beginner slope, snow spraying at the skis, pines blurring past. A low chase shot from behind, then a front-facing action-camera angle catching @avatar laughing while wobbling and steadying. SCENE 3 8 to 12s — pine forest. A worm's-eye view up through frosted branches, pale sun filtering down with soft flares off ice crystals. Cut to a close-up of @avatar walking through untouched snow, one mittened hand trailing a low branch, stopping to exhale a long visible breath and look up. SCENE 4 12 to 16s — café break. @avatar sits at a frosted window inside a small timber café, both hands wrapped round a steaming mug of hot chocolate under whipped cream. Warm interior light against the blue-white world outside. Close-up of marshmallows melting, then @avatar's contented half-smile watching the snow drift past the glass. SCENE 5 16 to 20s — sledding. @avatar pushes off on a wooden sled at the top of a gentle hill and speeds down laughing, hair flying loose from under the beanie. Camera tracks alongside at sled height. @avatar tumbles softly into a snowbank at the bottom, still laughing, covers their face with one mitten, then drops it and grins straight at the lens. SCENE 6 20 to 24s — evening market. A glowing market strung with warm fairy lights, snow settling on wooden stalls, steam rising from vendor carts. @avatar weaves through the crowd sampling mulled cider and gingerbread, face lit by string lights and lantern glow. Close-up as @avatar's eyes widen at the taste, lights thrown into soft bokeh behind. SCENE 7 24 to 27s — sunset. A wide silhouette of @avatar on a snowy overlook as the sun drops behind the range, sky shifting pale gold into soft violet, scarf drifting in the wind. @avatar tilts their face toward the fading light, one hand holding the beanie in place. SCENE 8 27 to 30s — fireside. A cabin interior at night, fire glowing, snow at a frosted window behind. @avatar sits curled on a wool blanket, mug in hand, firelight flickering across the face. Final intimate close-up: @avatar looks straight into the lens, lifts the mug slightly in a small toast, and smiles. NOBODY SPEAKS. No lines are delivered to camera at any point. Every beat is carried by expression, gesture and movement. CAMERA. Authentic travel-vlog cinematography — handheld shake, slow push-ins, natural drifting movement, occasional POV, subtle motion blur. Smooth cuts between scenes, no flashy transitions.
Create a first-person POV viral video showing the magic of architectural design coming to life. Real hands in black leather gloves hold up an architectural blueprint of a sleek modern villa, then pull it away to reveal the villa magically assembling in real-time. Concrete walls, steel frames, and glass panels rapidly materialize with cinematic effects, dust clouds billowing as construction happens at impossible speeds. The blueprint elements transform into real structural components. Set on a bright sunny construction site with clear blue skies.
ASSET MAP — every reference is bound here and cited by tag below: @avatar — the traveller. This is the person; it defines face and identity only. Its clothing, pose, framing, background and lighting are ignored completely. ATTACHED IMAGES ARE MANDATORY AND FINAL: the person in @avatar is the person in every one of the twenty shots — reproduce them exactly, and never invent a different face, hairline, hair colour, skin tone, body or age. The attached image outranks every word written below it: wherever the image and this text disagree, follow the image. CHARACTER LOCK — HIGHEST PRIORITY: The person in every shot is exactly the person in @avatar. Preserve their facial identity, facial structure, facial proportions, eye shape and spacing, eyebrows, nose, mouth, cheek structure, jawline, chin, exact skin tone and natural skin characteristics, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, from the first frame to the last. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not redesign, beautify, slim, age, de-age or restyle them. Only one person carries the camera, and no duplicate of that person ever appears. WARDROBE (comes from THIS TEXT, never from @avatar) Each shot has its own location-appropriate layer, named in that shot. @avatar shows face and build, not these clothes — put each outfit on them. Fit every garment to the body naturally without altering its design, cut, proportions or colour: the garment is fixed, only its drape adapts. An outfit belongs to its own shot only and never blends into the next one. ONE-SENTENCE SUMMARY @avatar travels the world in a beat-synced selfie hyper-lapse, holding the camera at arm's length in front of twenty famous landmarks, one hard cut every half second, shot on a wide-angle phone lens as a premium travel vlog. DETAILED PLOT — exactly 20 shots, one hard cut every 0.5 seconds, exactly 10 seconds in total Every shot: handheld selfie-stick perspective at natural arm-length distance, wide-angle phone lens with authentic perspective distortion, realistic handheld micro-movement and subtle motion blur, @avatar filling the near frame with the face fully visible and the landmark clearly readable behind. One landmark per shot, one gesture per shot. Hard cut on the beat between every shot, never a dissolve, a morph or a gradual change of place or outfit. The first shot opens already in motion, on a readable face — never on an empty frame or a fade from black. Shot 1 — the Eiffel Tower, Paris. Cream knit sweater and a wool beret. @avatar waves at the lens. Soft morning light. Shot 2 — Shibuya Crossing, Tokyo. Varsity bomber jacket. @avatar gives a peace sign as the crowd streams past. Neon dusk. Shot 3 — Times Square, New York City. Denim jacket. @avatar points up at the screens. Night, saturated billboard light. Shot 4 — the Colosseum, Rome. White linen shirt. @avatar gives a thumbs-up. Hard midday sun. Shot 5 — the Pyramids of Giza, Egypt. Sand-coloured shirt and a wide-brim straw hat. @avatar pulls a surprised open-mouthed reaction. Desert haze. Shot 6 — Christ the Redeemer, Rio de Janeiro. Green-and-yellow print shirt. @avatar makes a finger heart. Bright overcast. Shot 7 — Big Ben, London. Beige trench coat. @avatar points back at the clock tower. Flat grey daylight. Shot 8 — the Sydney Opera House, Sydney. Striped cotton tee. @avatar laughs at the lens with the harbour behind. Clean blue afternoon. Shot 9 — the Grand Palace, Bangkok. Light printed overshirt. @avatar holds a namaste. Gold hour glare off the spires. Shot 10 — the Taj Mahal, India. Patterned scarf over a pale shirt. @avatar tilts the head and smiles. Warm early sun. Shot 11 — the Great Wall, China. Grey fleece shell. @avatar waves with the wall running away behind. Cool mountain light. Shot 12 — the Burj Khalifa, Dubai. White linen set. @avatar points straight up along the tower. Bleached noon sky. Shot 13 — Hagia Sophia, Istanbul. Woven pashmina. @avatar raises one hand in a small greeting. Low afternoon sun. Shot 14 — the Grand Canal, Venice. Navy striped top, seated in a gondola. @avatar lifts both hands in delight. Soft reflected water light. Shot 15 — Machu Picchu, Peru. Knitted andean-pattern layer. @avatar gives a thumbs-up on the terraces. Thin high-altitude light. Shot 16 — the Acropolis, Athens. White cotton shirt. @avatar shades the eyes with one hand and squints into the sun. Harsh white daylight. Shot 17 — the Sagrada Família, Barcelona. Soft ivory shirt. @avatar leans back with an amazed look up at the spires. Clear blue sky. Shot 18 — the windmills of Zaanse Schans, Amsterdam. Brown corduroy jacket. @avatar claps once, delighted. Windy bright overcast. Shot 19 — Gyeongbokgung Palace, Seoul. A pale traditional Korean jacket rented at the palace gate. @avatar gives a small bow toward the lens. Even autumn light. Shot 20 — Marina Bay Sands, Singapore. Dark evening shirt. @avatar grins and swings the camera slightly as the skyline lights burn behind. Blue-hour night. ADDITIONAL NOTES PACE AND CUTS Twenty shots of equal length, one hard cut every 0.5 seconds, 10.0 seconds in total. Every cut lands on the beat. No shot is held longer than another, no location is blended with the next, and no outfit carries across a cut. LOOK Ultra-realistic cinematic travel photography with a premium travel-vlog finish: natural colour science, realistic depth of field, authentic wide-angle perspective, subtle cinematic contrast, real motion blur. Physically believable location-specific light in every shot — natural sunlight, accurate shadows and reflections, atmospheric depth, consistent exposure across the cuts. Natural skin texture with visible pores and fine facial detail, physically accurate hair strands, real fabric behaviour, believable local crowds and environmental movement. Nothing looks rendered or retouched. HANDS Hands stay anatomically correct with five fingers on each hand in every gesture, and the hand holding the camera stays out of the way of the face. TEXT IN FRAME Only the real signage that belongs to each location appears. Do not generate a logo and do not generate a watermark. No subtitles. AUDIO An upbeat electronic travel track at about 120 BPM with a clean beat landing on every cut, plus a thin layer of location ambience under each shot — crowd, traffic, wind, water. No dialogue.
12-second cinematic rally film, 16:9. Six shots, hard cuts, fast pace. @avatar — the driver. @image1 — the car. Keep the driver's face and the car's livery identical in every shot. Wardrobe: white racing suit, helmet, goggles, gloves. Location: red desert plateau, low sun, dust, open dirt track. Shot 1: the car drifts past camera, low and wide, dust everywhere. Shot 2: inside the cabin, close on the driver's face. Shot 3: ground level at the rear wheel, gravel at the lens. Shot 4: wide tracking alongside, tail out through a curve. Shot 5: back in the cabin, the driver catching the slide. Shot 6: the car stops side-on, dust clears, whole car in frame. Hold. Style: photoreal motorsport, heavy dust, motion blur. Audio: engine, gravel, wind. No music, no dialogue. Avoid: no crashes, no other cars, no text or logos.
Generate a twenty-second vertical ultra-realistic cinematic aerial stunt film from the provided first frame. ASSET MAP - every reference is bound here and cited below: @image_1 - the first frame of this video. It is the exact opening image, and the person in it is the athlete for the whole film. CHARACTER LOCK - HIGHEST PRIORITY: the athlete in every segment is exactly the person in @image_1. Preserve the facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions - identically, segment to segment. Preserve the gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. No face changes, no identity drift. One single athlete appears in this film. ATTACHED IMAGE IS MANDATORY AND FINAL FOR THE OPENING AND THE IDENTITY: start from @image_1 as the exact first frame, and wherever @image_1 and this text disagree about a face, hair, skin tone, body, age, the jumpsuit, the harness or the helicopter interior, follow @image_1. The camera path, the timing and everything that happens after the first frame come from this text. WARDROBE LOCK: the athlete wears the matte black aerodynamic jumpsuit with dark red panels, the compact safety harness and parachute pack, black gloves, lightweight boots, clear goggles and the small helmet-mounted camera exactly as established in @image_1, unchanged from the first frame to the last. No visible logos, brand marks, printed words or numbers on any garment or on the equipment. Nothing is ever strapped to the feet of the athlete - no board, no ski and no equipment of any kind, at any point in the film. Both hands keep five fingers in every segment. Summary: a professionally supervised fictional stunt - the athlete exits a helicopter above Paris at golden hour, free-falls the full height of a vast wrought-iron lattice tower, passes it within metres, stabilises, deploys a guided parachute with a hard snap, and lands on marked open ground at the far end of the esplanade, well clear of the tower. Visual style: ultra-realistic vertical sports footage with the grit of real helmet-cam and chase-cam material - hard low golden light on grey zinc rooftops and cream stone, heavy motion blur, rolling horizon, lens vibration, fine grain, deep readable shadows. Cinematic but never glossy or staged. SPEED RULE - this is what sells the fall: the camera never travels at the same speed as the athlete during free fall. In every falling segment something must rush past inside the frame - the tower's ironwork, a platform edge, a cloud layer, the rooftops - so the speed is visible against a moving foreground rather than a still one. The horizon rolls, the frame vibrates, and the body is thrown by the airflow. 0-2 seconds: the film starts exactly on @image_1 and immediately moves. The camera pushes in fast from the doorway framing to a tight shot of the face of @image_1, wind hammering the jumpsuit and tearing at the hair, the rotor beat shaking the lens. The goggles come down over the eyes, the jaw sets, and the crew member deeper in the cabin slaps the shoulder of the athlete once. <rotor beat, hard wind at the open door, a single hand slap on fabric> （a low tense drone, one slow pulse） 2-4 seconds: one continuous shot from inside the door, camera locked to the doorframe. The athlete pushes off hard with both hands and falls away from the camera, shrinking fast against the rooftops in under a second while the doorframe stays in frame in the foreground for scale. Then the camera whips down after the athlete with full motion blur. The exit is fully visible - the body leaves the frame downward, never by a cut. <a hard push off metal, the wind opening into a roar, the rotor snapping away above> （the pulse cuts to silence） 4-7 seconds: helmet-camera point of view from the athlete looking down along the body - the boots against the rooftops far below, the iron tower rushing up toward the lens from the bottom of the frame, the city rotating slowly around the edges. The frame vibrates hard. <a solid wall of rushing air, the suit drumming, breathing inside the helmet> （a rising low string swell） 7-10 seconds: the camera drops to a chase position beside the athlete as they pass the tower within metres, belly-down and arms spread. The lattice ironwork rips past the frame at speed, one platform edge ticking by close to the lens, the golden light strobing through the girders. <ironwork wind roar, a hard flutter of fabric> （the swell keeps climbing） 10-12 seconds: a tight close-up on the face of @image_1 in free fall, filling the frame - the cheeks pulled back by the airflow, the goggles pressed against the skin, loose hair hammering across the temple, the eyes locked forward and calm. Brief slow motion, then back to real time. The face, hair and skin tone stay exactly as in @image_1. <the roar going slightly muffled, then full> （the swell thins to a single held note） 12-14 seconds: the hand of @image_1 finds the deployment handle and pulls it hard. The pack cracks open, the suspension lines rip out and snap taut, and a compact red-and-white canopy slams open above the athlete. The harness takes the weight with a hard jolt that throws the legs forward and the descent breaks sharply. Slow motion across the snap, then real time. <the pack cracking, lines snapping tight, a deep canopy boom, the harness creaking under load> （the held note breaks into a wide warm chord） 14-17 seconds: the camera settles behind and slightly above the athlete, hanging in the harness with both arms spread wide, the long formal lawns, gravel paths and clipped tree rows of the esplanade sliding beneath and a gilded dome standing in the distance ahead. The canopy breathes in the air above. <calm air, fabric luffing, distant city> （the chord opens up） 17-20 seconds: a low camera at grass height as the ground rushes up - the lawn filling the frame, then the boots slam down and take two hard running steps, the canopy collapsing forward onto the grass behind. Ground crew in high-visibility vests stand well back beside coloured markers. The camera rises and settles locked off on the face of @image_1, who straightens up facing the lens, still breathing hard, the dome and the grey Paris rooftops glowing behind. The composition holds still to the final frame. <ground rush, boots hitting grass, fabric collapsing, hard breathing, wind dropping away> （the chord holds and fades） Continuity and physics: one single athlete throughout, always the person in @image_1, in the same jumpsuit, harness and goggles from the first frame to the last. Every falling segment is the same uninterrupted fall, and the athlete is lower and closer to the ground in each one. Real air resistance and real mass: the body accelerates for the first seconds, the fabric flutters and snaps under load, the canopy inflates in stages and takes the weight before the descent slows, the lines stay attached and visible, and the landing carries forward momentum into two steps. The iron tower, the esplanade and the city stay background geography and are never touched, never landed on and never damaged; the landing ground is open, marked and well away from the tower. The jump is a supervised fictional stunt: the harness, the crew and the marked landing ground are visible, the canopy deploys cleanly, and the athlete lands safely on both feet. Constraints: no subtitles and no on-screen text of any kind anywhere in the film. Do not generate a watermark. Do not generate a logo.
@avatar is the person walking - it defines face and identity only. Preserve their facial identity, skin tone, hairline, hair, apparent age, build and gender presentation exactly as shown, in every frame; never invent a different face. The attached image outranks this text wherever they disagree. Wardrobe comes from this text, never from @avatar: a sharp all-black tailored suit - single-breasted jacket with narrow lapels worn closed, straight-leg trousers, a plain black shirt buttoned to the collar with no tie, polished black shoes, one thin gold chain. Monochromatic, unbranded, no lettering. Fit each garment to their body naturally without changing its design or cut - the garment is fixed, only its drape adapts. Same outfit in every second. 10-second hyper-realistic red-carpet arrival, vertical 9:16, one unbroken shot, late at night. @avatar walks straight toward the camera down a deep red carpet holding direct eye contact with the lens, framed head to toe. A press line runs along both sides behind a low barrier: photographers shoulder to shoulder, long lenses up, firing out of sync. THE LOOK IS HARD DIRECT FLASH. Harsh on-camera flash dominates every frame - bright speculars on skin and black fabric, deep contrast, a crisp hard shadow on the carpet behind @avatar, and fast falloff into black past them. Side flashes rake in at different intensities with overlapping shadows and occasional lens flare; between bursts the scene drops to dim tungsten and the carpet reads deep and saturated. Each burst freezes @avatar razor-sharp mid-stride while the crowd behind smears. Eye level, 35mm, flash at the camera - aggressive tabloid arrival, not soft editorial. High-ISO grain, real skin texture, five fingers, correct legs and feet. The clip opens on the peak of a flash burst with @avatar already mid-stride - no fade from black. @avatar walks at an even confident rhythm, arms swinging subtly, shoulders relaxed, never an exaggerated runway pose. Midway a volley hits from the left, then the right, lighting their face from both sides at once as the framing tightens toward three-quarter. In the last beat @avatar arrives close to the lens, still walking and still holding eye contact, one final burst fills the frame, and the clip ends on that motion. Diegetic sound only: dense overlapping shutter clicks and flash recycle whine, steps on carpet, photographers calling over each other, a low crowd murmur behind the barrier. No BGM; generate only environmental sounds and action sounds. No subtitles.
A relaxed boardwalk stroll turns into a slick, deadpan action sequence when a bathrobe-clad parent takes down three cocky skateboarders for ruining their kid's ice cream — capped with an unimpressed freeze-frame.
Scarlet '87Recreate workflow
