Drag the clip sideways
An emotional cinematic café short: @image1 walks into a cozy café on a rainy evening and sees @image2 kissing @image3 at a window table across the room. @image1's face shifts from confusion to heartbreak as quiet tears form, then @image1 turns and walks out. Restrained, photoreal, handheld cinematic, warm rainy light, diegetic sound only, no dialogue.
Create a 15-second ultra-realistic cinematic 9:16 video of @product being built as a luxury miniature countryside villa, then revealed at golden hour. Keep @product locked to the reference: same silhouette, proportions, materials and colours once the form appears. One object, never duplicated. Two maker hands only — no faces, no other people. 0–3s — empty plot. A miniature countryside diorama: lush grass, olive trees, natural rocks, rustic wooden fencing, rolling farmland and distant hills under an orange-pink sunset. The footprint of @product is marked in the earth. Soft cinematic aerial, shallow depth of field. 3–7s — villa rises. Two hands, macro, miniature tools. Beige natural-stone walls of @product are laid, cement spread, a rustic terracotta tiled roof set, wooden doors and windows fitted, a covered outdoor dining patio added. Warm interior lights begin to glow. No cuts that break the same miniature. 7–11s — pool and water. Hands tile a crystal-clear turquoise swimming pool in front of @product: stone paving, wide steps, a dramatic infinity-style waterfall edge. Water fills; the cascade starts to flow with realistic liquid physics and photorealistic reflections. 11–15s — aerial pull-back. Smooth cinematic aerial pull-back revealing completed @product: glowing interiors, outdoor furniture, lanterns, olive trees, flowers, waterfall edge, the whole miniature property in golden-hour farmland. Hold the hero. No on-screen text. Sound: tiny masonry taps, water fill, waterfall, evening insects, a soft wind. No dialogue. No titles, no subtitles, no watermark, no logos that are not already on @product. Duration: exactly 15 seconds. Aspect ratio: 9:16.
A private water taxi glides down the Grand Canal toward a historic palazzo aglow with camera flashes — your avatar's phone catches it all in raw, handheld selfie-vlog style. Stepping onto the torch-lit landing, they're swept into a wall of photographers, a red-carpet climb, and an unexpected embrace with a fellow gala guest. A brief reporter interview, a candid laugh as the crowd jostles the frame, then a step inside: a frescoed ballroom, glowing chandeliers, and elegant guests at long dining tables. The night closes on one final selfie and a knowing smile before the phone drops. A 30-second documentary-style fashion-event vlog, unscripted in feel from first frame to last.
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, 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 reinterpret it. Do not beautify, slim, de-age or restyle them. The attached image outranks every word below it: wherever the image and the text disagree, follow the image. Frame 1 is already in the scene — never a studio portrait, grey backdrop, or the attached still as an opening plate. Use they/them throughout. No other people. Create a 15-second ultra-realistic early-2000s consumer camcorder home video in London. 9:16. The person in the attached reference is the only person. OPENING — MANDATORY: Frame 1 is already in the London street, handheld camcorder running. They are already mid-run toward a bus. Do not open on a studio portrait, grey backdrop, still headshot, or the attached image as a plate. NEVER insert the attached still, a grey seamless, or a beauty-headshot as a cutaway. Every single frame is handheld location footage. The spoken line is delivered on the street, in the hoodie, after missing the bus — not as a studio talking-head. IDENTITY: preserve the attached face exactly, including green eyes, freckles, black wavy hair, and the pointed elf ears — both ears stay pointed in every frame. Do not humanize the ears. Do not change the face. Wardrobe, locked: navy hoodie over a white top, beige cargo shorts, gray sneakers, burgundy crossbody bag, silver watch. Fit these clothes to the attached person's body. Do not restyle the hair except as wind and running move it. STYLE: handheld footage from an old consumer DV camcorder on a hot summer afternoon in a quiet older London neighborhood. Brick terraces, a red double-decker bus, a corner café with a takeaway window, a shop front with a glass display. Strong natural handheld shake, imperfect framing, autofocus hunting, faded DV colours, mild tape noise. No cinematic gimbal. Never show the camera operator. No other faces (barista hands only if needed, no face). 0–3s — chase: Already running after a red double-decker pulling away from the stop. Chaotic handheld chase, breathing, footsteps. They miss it by a second. 3–6s — the line: They stop, out of breath, turn fully toward the camera and clearly say in English, exact lip sync: "Oh, I guess I'll walk today." Those exact words. Then a small resigned laugh. Do not say "I'm going to drive it." 6–10s — coffee: They duck to a neighbourhood café takeaway counter, grab a paper takeaway coffee cup with a lid (plain cup, no brand logo), pay quickly, keep walking. Cup stays in one hand for the rest of the clip. 10–13s — shop window: They slow down and peer through a shop window at something in the display, coffee in hand, a beat of curiosity, then keep walking. 13–15s — heat: It is baking hot. They pull a small paper fan from the bag and fan their face while holding the coffee, slightly overheated, a little laugh, camera still following. Abrupt DV cut. AUDIO: one spoken English line only — "Oh, I guess I'll walk today." Plus running, breathing, bus engine, café clink, street, fan flutter. No music, no subtitles, no logos, no watermark. No on-screen text. Duration: exactly 15 seconds. Aspect ratio: 9:16.
A cinematic night car showcase featuring @image1: your car on wet neon-lit streets - macro on the badge with neon crawling over the glossy paint, headlights igniting, the car rolling through puddles with real spray, ending on a low hero shot. Rain, reflections and a driving techno beat throughout. Photoreal, premium, physics-accurate, no dialogue.
Deep in a misty rainforest cacao plantation, a gold foil wrapper peels back in one fluid motion to reveal a glossy dark chocolate bar, its surface catching dappled sunlight breaking through the canopy above. A close-up on ripe cacao pods sets the scene before the bar snaps cleanly in two, releasing a fine drift of cacao powder through a shaft of light. The camera flows through the plantation mist and out onto an open-air terrace, where warm golden light gathers into a final hero shot: the bar resting on rustic dark wood, framed by cacao pods and lens flare. A refined, 10-second product reveal built entirely around continuous, connected motion.
A bustling city street turns chaotic when a thief rips an elderly woman's purse from her arm and bolts through the crowd. Your avatar springs into action — vaulting a market stall, weaving through shoppers, cutting through a narrow side street. A quick, controlled scuffle gives way to the money shot: a slow-motion flying tackle that sends the thief sprawling and the purse spinning through the air. Your avatar catches it clean, returns it with a respectful nod, and the street erupts in applause. A tight, 10-second cinematic hero moment, warm and family-friendly.
Create a 15-second 9:16 ultra-realistic miniature model-kit assembly video of a candy-apple red 1960s fastback die-cast. a candy-apple red 1960s fastback die-cast is the exact die-cast in the reference image: candy-apple red 1960s fastback coupe, white GT side stripes, chrome bumper and grille, five-spoke silver wheels, glossy toy-model proportions. Same shape, paint, stripes, chrome and wheels in every frame. One object, never duplicated. Never turn it into a full-size car. OPENING: Frame 1 is already the workshop table with the sealed kit box — no still packshot as a plate, no grey seamless, no portrait. Hands only. Two realistic adult hands, five fingers each. No faces, no other people. 0–3s — unbox: Overhead / POV macro. A premium sealed model-kit box sits on a clean workshop mat. Hands set it down, open it, peel wrapping, lift fitted trays. Parts inside match a candy-apple red 1960s fastback die-cast: red body shells, chrome bumper, wheels, tyres, interior, screws. 3–8s — assemble: Fast but readable time-lapse. Chassis on the mat, engine and suspension, axles, wheels tightened, interior, doors, red fastback body panels, chrome, lights, GT stripes already on the sides as in the reference. Miniature screwdriver and ratchet. Same two hands throughout. 8–12s — finish: Last bolt, wipe with a detailing cloth, place a candy-apple red 1960s fastback die-cast on a small rotating display stand. Workshop lighting becomes cinematic. Light sweep. Headlights click on. 12–15s — hero: Slow orbit of the completed candy-red fastback, white GT stripes catching the light, chrome gleam. Subtle engine sound. End on a strong 9:16 hero. No on-screen text. Sound: cardboard, tape, tray clicks, screws, ratchet, panel snaps, cloth, headlight click, a small engine blip. No narration, no music bed, no subtitles, no watermark in the prompt (the platform may stamp one). No extra fingers, no floating tools, no full-size car, no morphing paint, no logo distortion. Duration: exactly 15 seconds. Aspect ratio: 9:16.
Under a single dramatic spotlight on a darkened stage, your avatar stands perfectly still as a pair of giant, unseen hands work with the precision of professional stylists. One by one, the pieces arrive — combat boots laced onto each foot, a leather jacket settled onto the shoulders, a chain and cuff fastened, dark shades placed just so, and finally an electric guitar slung into position. A quick restyle of the hair completes the look. As the hands withdraw, your avatar stands transformed: fully stage-ready, lit like the cover of a rock album. A polished, 10-second continuous-shot transformation reveal.
A crisp autumn afternoon on a tree-lined city path takes an unexpected turn when a passing cyclist clips your avatar's shoulder mid-stride. In that instant, time suspends — earbuds, a latte cup and a small parcel are caught in extreme slow-motion bullet-time, rotating through warm autumn light as coffee arcs and leaves drift mid-air. Time resumes, everything falls, and your avatar turns to camera with a perfectly deadpan "did that really just happen" expression. A polished, 15-second cinematic bullet-time moment blending high-fashion editorial style with a touch of everyday comedy.
Garage monster horrorRecreate workflow
