Dari X · Motion Transfer
Prompt Video AI
Prompt video AI siap salin dari kreator X, lengkap dengan klip hasilnya: motion transfer, dance, gerakan kamera, gambar ke video, dan lainnya. Dengan kredit kreator, siap dicoba.
Cyber Samurai Neon Rain Duel
Cyberpunk Neon Samurai Duel (High-energy action viral style)15-second ultra-cinematic cyberpunk VFX duel, anamorphic lens flares, 8K hyper-realistic, neon-noir rainy Shinjuku alley at night, torrential rain with realistic splashes and reflections on wet asphalt, pink/teal volumetric lighting, high-speed shutter with motion blur, realistic physics mixed with stylized slow-mo impacts, sparks and energy shockwaves, no text/logos, samurai maintains perfect face, glowing neural links, and katana consistency throughout, no deformation or extra limbs. 0–2s: Extreme low-angle tracking shot, cyber-samurai with glowing blue neural links strides powerfully through pouring rain, katana humming with electric energy. 2–4s: Wide whip-pan as holographic guards materialize, samurai launches into fluid sword flurry, blade collisions exploding bright sparks. 4–6s: Medium 120fps tracking shot, low sweeping kick sends one guard flying, camera follows the leg arc with realistic rain spray. 6–8s: Bullet-time 360° orbit, samurai deflects digital bullet rain with vibrating katana, creating expanding blue energy rings in slow motion. 8–10s: Vertical slow-mo jump spin, powerful downward strike cracks asphalt, concrete shards and rain droplets exploding upward in perfect spheres. 10–12s: Close-up whip-pan on samurai’s reflective visor, neon lights flickering as he spins to face the final guard. 12–14s: Extreme close-up on narrowed eyes, katana sheathes with sharp metallic click and electric discharge, background neon bokeh blur. 14–15s: Quick wide pull-back, samurai stands victorious amid settling rain and debris in a dramatic hero silhouette freeze.
Seedance 2.5 urban sunset drive with custom matte-black 911
Create a 30-second cinematic urban sunset driving sequence using the two reference images strictly. REFERENCE LOCK: [1] THE DRIVER — angular face, tan skin, curly black hair, goatee, black t-shirt, dark cargo pants, black boots. [2] THE CUSTOM 911 — matte black widebody Porsche 911, white "911" graffiti, BBS wheels with red brake calipers, red underglow, rear wing, dual center exhausts. VISUAL STYLE: Arcane and Spider-Verse 2D/3D hybrid artwork, hard-edge brushwork, stepped 12fps animation cadence, golden-hour sunset vs. neon city lighting. No music, only footsteps, ignition, flat-six roar, tire screeches, and gear shifts. 0–5 SECONDS — PUDDLE & IGNITION Low-angle shot of red underglow reflecting in a puddle as the Driver's [1] black boots step toward the Custom 911 [2]. Close-up: hand pulls handle → Driver [1] sits → turns ignition key → grips wheel. CUT. 5–10 SECONDS — LAUNCH & MIRROR GLANCE Dual center exhausts [2] blast white smoke as red underglow ignites. Hand shifts gear → Custom 911 [2] launches under elevated train tracks. Cockpit POV tilts up at the bridge, then snaps to the Driver's [1] eyes in the rearview mirror. 10–18 SECONDS — INTERSECTION DONUTS Top-down rotating shot as the Custom 911 [2] spins 360-degree donuts between yellow taxis. Close-up of the BBS wheel and red caliper [2] spinning violently, throwing sparks and red-lit smoke before rocketing away. 18–25 SECONDS — UNDER-TRUCK PASS & FENDER CAM Black boot [1] slams the gas pedal; hand upshifts by the glowing radio. Low camera shoots underneath a passing truck's chassis as the Custom 911 [2] streaks past. Cockpit camera pushes through the windshield onto the hood [2] with white anime speed-lines, cutting to a tight rear-fender camera locked onto the spinning wheel lip in thick pink-red smoke. 25–30 SECONDS — HIGHWAY LOOP & LIGHT TRAILS High aerial shot: the Custom 911 [2] drifts around a curved highway overpass, pouring twin trails of crimson tire smoke and glowing red taillight streaks toward the sunset skyline.
This is the kind of Seedance 2.5 experiment I LOVE.
Ultra-realistic live-action bar scene in bright daytime, premium cinematic realism, natural window light, shallow depth of field, subtle handheld camera, slow push-ins, realistic skin texture, glass reflections and room ambience. Everything is photorealistic EXCEPT the woman, her cocktail and her smartphone, which are rendered as polished hand-drawn 2D manga animation with clean linework, expressive eyes, cel shading and fluid anime motion. The visual contrast must remain stable throughout. Man: adult, late 20s, realistic live-action human, attractive but extremely shy, slightly awkward posture, nervous smile, casual dark shirt, trying hard to seem confident. Woman: adult, late 20s, hyper-stylized manga woman, glamorous and very sexy but tasteful, long dark hair, elegant fitted outfit, expressive eyes, confident relaxed body language. She is fully 2D animated at all times. Only these two characters are important. Upscale modern cocktail bar during the day, warm sunlight through large windows, wooden counter, glass shelves, plants, quiet atmosphere. She sits alone at the bar looking at her manga-style phone with a manga-style colorful cocktail beside her. 0-6s: Medium tracking The man notices her from several meters away. He freezes for one second, quietly rehearses under his breath: MAN: “Hi… no. Hey… worse.” He exhales and walks toward her. 6-12s: Two-shot He stops beside her stool. MAN: “Excuse me… is this seat taken?” She looks at the completely empty bar around them, then back at him. WOMAN: “You mean the other twenty seats?” 12-18s: Close-up on man He realizes how bad that sounded. MAN: “Yeah… I panicked halfway through the sentence.” She lowers her phone, amused. WOMAN: “At least you’re honest.” 18-24s: Slow push-in He glances at her animated cocktail. MAN: “Can I ask you something weird?” WOMAN: “You already did.” Beat. MAN: “Why does everything around you look real… except you?” 24-30s: Viral payoff She slowly smiles, leans slightly closer. WOMAN: “Maybe you’re the one who’s drawn badly.” He stares at her, completely speechless. She takes a sip of her animated cocktail. WOMAN: “Relax. That was your chance to flirt back.” He quietly says: MAN: “I’m still loading.” She laughs. Hard cut. Live-action man and environment remain fully photorealistic, woman remains fully 2D manga throughout, her cocktail and smartphone remain manga-styled objects, no visual blending between styles, precise eye lines, natural English dialogue, excellent comedic timing, expressive acting, believable awkward chemistry, stable character continuity, no readable text, no logos.
Gave Hailuo H3 and Kling 4.0 the exact same prompt and character sheet.
STYLE FIRST: a real Japanese TV anime, like a scene from a seasonal anime opening. Hand-drawn character animation on twos and threes, hand-inked lines that change weight, flat cel color with one hard-edged shadow tone, painted anime background art with soft brush texture. Anime compositing only: gentle backlight bloom, light leaks at sunset, depth-of-field blur on painted layers, rain drawn as thin streaks. No 3D, no CG, no glossy skin or hair, no AI polish. This applies to every shot. CHARACTER: use @[image1] as the character design sheet and match its hand-drawn look. Keep her face, violet eyes, very long straight pink hair with blunt bangs, white cropped high-neck top with black collar, oversized off-shoulder white techwear jacket with orange panels, black harness belt with white hip pouch, orange baggy cargo pants, one orange fingerless glove, white and orange sneakers. Her face stays on-model in every shot. MAIN GOAL: a 30-second calm slice-of-life showcase, no fighting. One evening in a Japanese city from golden hour to the first star. 7 shots, each about 4 seconds, each a different classic anime shot type with one slow camera move. Hard cuts. Let every shot breathe, no rushing. Her hair and jacket always move in the wind. (0–4.5s) SHOT 1, extreme wide: a city rooftop at golden hour, she sits on the railing edge, skyline and long orange sky behind her, wind lifting her hair. Slow push-in from far away. (4.5–8.5s) SHOT 2, extreme close-up of her violet eyes, the sunset reflected in them, a strand of pink hair crossing her face. Rack focus from the hair strand to her eyes, then she blinks once. (8.5–13s) SHOT 3, inside a commuter train: she stands by the window holding a strap, orange sunlight strobing across her face through passing poles. The camera tracks sideways with the train. (13–17s) SHOT 4, insert close-up: her gloved hand on the train window as the first raindrops hit the glass, city lights blurring behind. Slow push-in on the hand. (17–21.5s) SHOT 5, medium wide at a rainy bus stop at dusk: she is large in frame under the shelter beside a glowing vending machine, rain falling past the streetlight. She shakes rain off her sleeve and looks down the street. Slow pan toward her. (21.5–25.5s) SHOT 6, low side angle from the ground: her reflection in a puddle, her sneaker steps in and sends ripples. The camera tilts up from the side and stops on her face. (25.5–30s) SHOT 7, the rain stops. Side profile: she slowly tilts her head back and looks up, her chin lifting, as the clouds part and the first star appears in a deep blue sky. Slow crane up from her face to the sky and hold on the single star until the end. Sound: city ambience only, wind, train, rain, the rain fading to quiet in the last shot. No music, no voice, no text, no subtitles, no watermark, no other characters.
Seedance 2.5 on Dreamina_AI Prompt: Create a 15-second vertical 9:16 li
Create a 15-second vertical 9:16 live-action dance video using the person in @ Image1 Setting: a high school classroom — desks, chairs, chalkboard. One continuous take: no zooms, no cuts. - Prioritize @ Image1 for the face, hairstyle, outfit, colors, and overall silhouette - Only one person in the frame - Not intense breakdancing — a short live-action dance MV where a high school student casually dances along to the music, just for fun - Don't repeat the same choreography; mix in steps, turns, hand movements, footwork, profile angles, and swaying hair - Amateur camerawork, as if a friend is filming on an iPhone — handheld shake, slightly off-center framing that loosely follows the dancer
Seedance 2.0 mini honey badger animated continuation with character voices
Create a 15-second animated continuation starting exactly from the uploaded final frame of the corrected Part 1. Use this single image as the visual reference for both characters, their poses, environment, lighting, and art style. Characters: 1. Honey badger: preserve the stout upright body, dark bluish-purple fur, cream-white fur cap, small ears, thick limbs, long tail, and shoulder strap visible in the image. No axe. Initially seated astride the dragon immediately behind his mane. Serious, frightened during flight, then relieved and exasperated. 2. Goofy Chinese dragon: preserve the long serpentine blue body, yellow-gold mane, cream belly, long whiskers, round eyes, oversized buck teeth, and small limbs visible in the image. No wings. Inexperienced, innocent, and absurdly confident. Style: Cute traditional Chinese ink/watercolor cartoon, soft paper texture, expressive ink outlines, watercolor shading, clean silhouettes, smooth readable hand-drawn animation, and expressive cartoon comedy. Match the opening image’s colors, proportions, warm daylight, and layered mountain scenery. No 3D or realistic CGI. VFX: Thin watercolor cloud wisps curl around the dragon during the turn and reveal a sunlit grassy terrace beside a mountain lake. Wind gently trails the mane and whiskers backward. Subtle grass movement and a small dust puff during landing. No magical aura, lightning, transformation, or explosive effects. Setting and continuity: Start in the exact positions and flight direction shown in the uploaded frame. The rock face is ahead; an open route passes around its outer edge toward SCREEN RIGHT. A broad grassy terrace lies beyond the rock at a slightly lower elevation. Show this continuous route: approach, turn around the rock, descend to the terrace, land. Keep camera angles on the same side of the action axis and preserve predominantly LEFT-TO-RIGHT travel. No reverse flight, teleportation, or sudden changes of location. Critical anatomy and contact: The badger has exactly TWO arms and TWO hands attached to his own shoulders. During every airborne shot, both hands grip the dragon’s upper back immediately behind the mane. No pointing. Keep the near-side hand at its visible contact point; the far-side hand may remain naturally occluded. No badger hands or arms appear beside the dragon’s face or beneath his jaw. The badger’s hips stay seated on the same section of the dragon’s back, with legs securely astride it. The dragon’s head, neck, torso, and tail follow one continuous curved flight path without stretching or tangling. The badger releases his grip only after the dragon is fully stationary on the ground. Dialogue and voice direction: All dialogue is spoken in ENGLISH with clear, funny animated-character voices. Honey badger voice: A compact, slightly gravelly mid-low cartoon voice, dry and skeptical, rising sharply into alarm. Clear articulation, never a monstrous growl. Dragon voice: A bright, rounded, slightly nasal cartoon voice with a smiling, innocent delivery. Sound like an enthusiastic young adult, not a baby. His confidence is completely sincere. Use consistent voice identities throughout. Natural conversational pacing, no overlapping speech. Only the speaking character moves his mouth for speech; the other reacts silently. Accurate English lip sync while preserving animal mouth anatomy, especially the dragon’s two oversized buck teeth. Speak only the exact lines below. No narration, extra dialogue, or additional vocalizations. SHOT 1 — 0–3s: Camera: Side-tracking medium two-shot matching the opening frame. Show both faces and the approaching rock edge. Dragon action: Keeps looking ahead, eyes wide with uncertainty. Continues forward flight and asks: “Which way?” As the badger finishes answering, begins a gentle bank toward the open route on SCREEN RIGHT.
Seedance 2.5 phone-footage viral wrestling clip: grandma wins
Ultra-realistic amateur smartphone footage, 20-second viral wrestling clip. Filmed by a random spectator from the audience, raw handheld phone camera, strong shake, bad digital zoom, autofocus hunting, exposure pumping, compression artifacts, occasional heads blocking frame, crowd screaming, arena echo, no cinematic stabilization, no polished camera movement. Main subject: one elderly female professional wrestler, clearly a grandmother, silver hair tied back, colorful classic wrestling outfit, confident and fearless, surprisingly athletic. Opponent: one adult male professional wrestler, large athletic build, standing inside the ring, fully consenting, prepared for the staged move. Referee and crowd remain secondary. Small indoor wrestling arena, ring surrounded by excited spectators, cheap lights, folding chairs, phones raised everywhere. 0-4s: [Messy discovery] Camera is already filming the ring badly from several rows back. The filmer suddenly notices the elderly wrestler climbing onto the corner turnbuckle. FILMER in English: “Wait… WAIT… grandma’s going up?!” 4-8s: [Build-up] Shaky zoom on her balancing at the top of the corner. The crowd starts screaming louder. The male wrestler looks up from center ring, shocked but ready. FILMER: “No way she’s doing this!” 8-13s: [THE JUMP] She launches into a huge dramatic flying wrestling jump from the top corner, fully committed, body extended in midair. The phone camera jerks upward trying to follow, briefly loses focus, then catches her descending toward the opponent. 13-17s: [Impact + chaos] She lands the staged move cleanly onto the prepared wrestler and both drop safely onto the ring mat. The ring bounces hard. Crowd explodes. Camera shakes violently from the filmer screaming. FILMER: “OH MY GOD! SHE ACTUALLY DID IT!” 17-20s: [Viral payoff] The grandmother immediately sits up, raises one fist like a champion while the opponent lies safely catching his breath. The crowd chants and laughs. FILMER: “SHE’S A LEGEND!” Camera points at the ceiling for half a second from pure excitement. Hard cut. Authentic amateur phone realism, elderly adult performer only, staged professional wrestling, huge readable jump, realistic ring bounce, chaotic crowd reaction, strong viral timing, no gore, no visible injury, no text, no logos.
What if McDonald’s delivery used a flying Tesla Roadster?
10-second, 9:16, 24fps, ultra-realistic POV smartphone video, one continuous shot. Use the uploaded reference images as strict visual references: Reference 1: Chongqing Hongyadong night-view window composition. Reference 2: same female courier, hovering white Roadster, and food handoff composition. Reference 3: final McDonald’s fries, drink, and burger close-up. Use the uploaded white Tesla Roadster image as the strict vehicle reference. Keep the same silver-white body, low wedge-shaped silhouette, dark panoramic glass, black wheels, rear wing, and overall proportions throughout. Use the uploaded female portrait as the strict character reference. Keep the same young adult East Asian woman, same face, long wavy chestnut-brown hair, fair skin, makeup, and body proportions throughout. The entire video is filmed from a first-person POV inside a high-rise apartment in Chongqing at night, looking through an open window toward glowing Hongyadong, the river, bridge, and illuminated skyline. 0–2.5s Start with the Hongyadong night view matching Reference 1. One hand reaches toward the open window. The white Roadster appears from the upper-right background and flies smoothly toward the window. Soft blue-white hover light is visible beneath the car. Keep the camera almost fixed with only subtle smartphone micro-shake. 2.5–6.5s The Roadster slows down and hovers steadily directly outside the window, matching Reference 2. The passenger-side door is open. The same female courier sits inside wearing a stylish red-and-yellow McDonald’s uniform and cap. She smiles naturally and extends a simple tray toward the viewer. On the tray are only: one red McDonald’s fries carton, one white drink cup with straw, one burger box. The viewer’s two hands reach forward and receive the tray. Keep the handoff slow, clear, and physically believable. No floating food. No sudden object changes. 6.5–10s The viewer now holds the food in the foreground, matching Reference 3. The fries, drink, and burger box stay large, sharp, and stable in the center foreground. In the background, the same Roadster gently accelerates away from the window and flies over the Hongyadong skyline. End with the McDonald’s meal dominant in frame and the Roadster smaller in the distance. Visual style Ultra-realistic viral social ad, natural smartphone POV, realistic hands, realistic food textures, glossy car reflections, warm golden Hongyadong lights, deep night city atmosphere, subtle blue hover glow, realistic depth of field, believable motion and physics. Strict consistency Same woman. Same face. Same hair. Same Roadster. Same Hongyadong setting. Same McDonald’s red-and-yellow packaging. Avoid No Paris. No Eiffel Tower. No KFC. No Cybertruck. No generic sports car. No car transformation. No wings or rotors. No extra people. No duplicated hands. No duplicated food. No warped packaging. No floating objects. No chaotic camera movement. No cartoon CGI. No subtitles. No watermark. No UI.
No drone. No crew. No expensive setup.
Create a realistic 360-degree drone shot that slowly pans far away from the character, revealing the entire environment like the opening shot of a blockbuster movie.
Race car vs falling helicopter chase
15-second, 16:9 ultra-photorealistic cinematic action with synchronized sound. ONE red GT race car speeds along a closed mountain racing circuit. ONE black camera helicopter flies low above the SAME road, traveling in the SAME direction as the race car. 0–4s: Fast rear tracking shot. The red race car accelerates down the straight at extreme speed. Directly above and slightly ahead, the helicopter follows the road while filming. Suddenly the helicopter loses power. Rotor sound drops and the aircraft begins descending. 4–9s: Low front three-quarter tracking shot looking back at the approaching race car. The helicopter is clearly visible ABOVE and BEHIND the car, rapidly losing altitude toward it. The helicopter drops dangerously close. Its landing skids descend to only a few feet above the race car’s roof. The driver sees the helicopter in the mirror and violently swerves toward the open side of the track. 9–12s: Dramatic side tracking shot. The race car and helicopter are now almost level. The helicopter sinks even lower and one landing skid narrowly misses the car’s rear wing as the car slides sideways through the corner. Tires scream. Gravel and dust explode from the edge of the track. 12–15s: Wide rear shot. At the final possible moment, the helicopter regains power and climbs sharply while the race car straightens and accelerates away beneath it. The helicopter clears the car by only a few feet. End with both machines racing forward through the mountain circuit: helicopter climbing above, red race car escaping below. CRITICAL GEOMETRY: exactly ONE race car and ONE helicopter. Both always travel forward in the SAME direction. The helicopter begins ABOVE the race car, loses altitude toward the car’s ROOF, nearly hits it, then climbs away. Never place the helicopter on the road. Never place it in front of an oncoming car. Keep the same red race car and same black helicopter throughout. Realistic helicopter weight and rotor physics, believable race-car suspension and tire grip, intense speed, mountain scenery, rotor wash disturbing dust and vegetation, screaming engine, failing rotor sound, tire squeal, gravel impacts and powerful helicopter recovery.
Seedance 2.5 Halloween pumpkin trouble cat witch comedy short
👇 Create a ~20-second, 16：9 vertical, 24fps live-action Halloween fantasy comedy short. Concept: “The Cat Witch’s Prank” Overall tone: Cute Halloween, cat POV, light fantasy, playful chase comedy, and polished social-media short-film energy. The mood should feel warm, mischievous, slightly magical, and funny — never scary, violent, disturbing, or horror-driven. Keep the visuals photorealistic and cinematic, like a high-quality Halloween pet commercial. Main Character The main character is one small orange tabby cat. Appearance: Small body Round face Short orange tabby fur Amber eyes Slightly oversized black witch hat Deep purple hat band Small orange-and-black bow tie around the neck The same exact cat must remain consistent throughout the entire video. Do not change the breed, fur color, body size, face, hat, or accessories. There is only one cat in the entire film. No humans appear. Camera Style The video is primarily shown from the cat’s first-person point of view. Keep the camera low, close to the cat’s eye level, with subtle natural movement as if the cat is sneaking through the environment. Occasionally show: The cat’s orange front paws The tip or edge of the witch hat entering the top of frame The cat’s shadow on the ground Brief reflections of the cat in glass, windows, or metal decorations These details should make it clear that the viewer is experiencing the scene through the witch cat. Avoid excessive cuts. Keep movement continuous and spatial relationships easy to follow. 0–4 sec — Sneaking Toward the Pumpkins Halloween midnight. A quiet suburban front porch lit by warm amber porch lights. Three jack-o’-lanterns of different sizes sit near the front door. The porch includes subtle Halloween decorations, a few fallen leaves, candles, and wooden steps. From the cat’s low first-person POV, the camera slowly moves forward. Two small orange paws carefully step across the wooden porch. The pointed tip of the witch hat occasionally swings into the top edge of frame. The cat quietly approaches the largest jack-o’-lantern. Sound: Soft paw steps on wood, distant crickets, and subtle playful sneaking music. The cat stops directly in front of the pumpkin. 4–8 sec — The Spell The cat lifts one front paw. A tiny ring of soft purple magic appears just in front of the paw. The magic should feel delicate, cute, and mischievous — not powerful or explosive. The paw gently swipes through the air. All three jack-o’-lanterns suddenly glow purple through their eyes. The cat pauses for about 0.3 seconds. The camera slowly leans closer toward one pumpkin. The pumpkin suddenly gives a tiny shake. The cat freezes in surprise. Sound: A soft “ting,” a tiny “poof,” followed by a small squeak. 8–12 sec — The Prank Goes Wrong The largest jack-o’-lantern begins a cute magical transformation. Do not make it explode or become grotesque. Four tiny mouse-like feet appear underneath it. Two small rounded mouse ears pop out from the sides. A thin mouse tail appears behind it. The pumpkin itself stays round, recognizable, and pumpkin-shaped. The second and third jack-o’-lanterns transform the same way. Now all three pumpkin mice begin scampering quickly across the porch. From the cat POV, the camera takes one small step backward. One pumpkin mouse turns and faces the cat. Pause for about 0.3 seconds. All three pumpkin mice squeak together: “Squeak!” Then they suddenly charge toward the cat. 12–17 sec — Reverse Chase The cat immediately turns and runs. The first-person camera moves quickly but remains readable and controlled as the cat runs back across the porch. Visible in frame: Fast-moving orange paws The witch hat tip bouncing wildly near the top of frame The cat’s shadow racing across the wooden floor The camera briefly looks backward. All three glowing pumpkin mice are chasing the cat. Their tiny legs move rapidly while they squeak: “Squeak, squeak, squeak!” One pumpkin mouse runs too fast and bumps into another. Both roll once across the porch, quickly pop back upright, and immediately continue chasing. Keep the moment cute and slapstick, never dangerous. The playful Halloween music speeds up into a fast comedic rhythm. 17–20 sec — Final Twist The cat hides behind a large porch planter. The camera slowly peeks around the edge of the planter. Everything suddenly becomes quiet. The three pumpkin mice are gone. The cat relaxes. Pause for about 0.5 seconds. Then one tiny pumpkin mouse slowly rises into frame from directly below the camera, extremely close to the cat. It looks up. “Squeak?” The cat suddenly jerks backward in surprise. The witch hat falls off and lands directly in front of the camera. CUT TO BLACK. Final sound: One surprised “Meow!” Then a tiny comedic magical “poof.” Visual Direction Photorealistic live-action pet cinematography. Color palette: Warm orange pumpkin light, deep blue nighttime shadows, and subtle purple magic accents. The pumpkin mice should feel: Cute, round, tactile, believable, and lightly magical. They must not have: Sharp teeth, disturbing faces, grotesque features, blood, or horror elements. The cat’s movements must follow realistic feline behavior: Light footsteps, natural pauses, believable body reactions, realistic running, and natural ear and tail responses. Magic effects should stay small, elegant, clean, and playful. Avoid oversized fantasy VFX. Sound Design No human dialogue. Use: Soft cat paw sounds Small meows Mouse squeaks Light magical ting and poof effects Pumpkin rolling sounds Witch hat falling sound Playful Halloween background music Music pacing: Sneaky and quiet at first, faster during the transformation, energetic and comedic during the chase, then suddenly quiet before the final punchline. Avoid No humans. No second cat. No breed changes. No cat transforming into a human. No talking cat. No realistic horror monsters. No blood. No injuries. No sharp teeth. No disturbing faces. No horror sound design. No extreme camera shake. No pumpkin explosions. No extra limbs. No subtitles. No logos. No watermarks. #Seedance25 #AIVideo #Halloween #AICat #ImaStudio #minimaxh3
Seedance 2.5 Korean-style cozy animal comedy with short-bob heroine
Create a cinematic, ultra-realistic Korean-style short video featuring a young East Asian woman with a short black bob haircut and bangs, wearing a dark brown hoodie and casual pants. The video combines adorable animal moments, unexpected comedy, and cozy everyday life, with realistic physics, expressive characters, and professional cinematic camera work. Scene 1 – A Funny Day at the Playground: In a modern residential apartment playground, a young woman stands beside a cute gray tabby cat wearing an adorable tiger-striped animal hat. A large yellow helium balloon floats above the cat, creating a funny and unexpected situation. The woman holds a smartphone near her ear while managing the situation. Capture the cat's curious expressions and the woman's natural reactions with soft daylight, realistic fur details, and a cinematic shallow depth of field. Scene 2 – The Unexpected Accident: The woman walks near the playground while the cat and its leash move around her feet. In a sudden, funny moment, she loses her balance and falls onto the soft sandy ground. The cat remains surprisingly calm and curious, creating a humorous contrast. Use dynamic camera angles, natural body movements, realistic physics, and smooth cinematic transitions. Scene 3 – The Cat's Adventure: Show the adorable tabby cat wearing its miniature tiger hat as it walks around the playground. The yellow balloon and leash create playful visual moments. Capture close-ups of the cat's face, whiskers, paws, and fluffy striped fur. Include wide shots of the playground, apartment buildings, benches, and surrounding greenery. Scene 4 – A Cozy Evening: Transition to a peaceful residential courtyard during the evening golden hour. Warm lights glow through the apartment windows as the camera captures a quiet outdoor moment with a small fluffy dog wearing a blue ribbon. Use soft background bokeh, natural environmental details, and a warm cinematic atmosphere. Scene 5 – A Funny Dinner at Home: Inside a cozy, warmly lit modern kitchen, the same young woman sits at a wooden dining table filled with Korean-style home-cooked dishes, rice bowls, and small plates of side dishes. She eats with chopsticks while the adorable miniature cat, still wearing its tiger-striped hat, sits beside her at the table. The cat makes cute, expressive movements and watches the food with curiosity. The woman suddenly notices something unexpected and turns toward the cat with a surprised, confused expression. End with a charming comedic moment that highlights their playful relationship. Visual Style: Ultra-realistic cinematic photography, Korean slice-of-life aesthetic, 4K resolution, natural skin texture, highly detailed animal fur, expressive facial reactions, realistic lighting, smooth camera movements, shallow depth of field, soft daylight, warm evening tones, and professional film-quality color grading. Camera Direction: A combination of medium shots, wide establishing shots, low-angle animal shots, detailed close-ups, gentle tracking movements, and smooth cinematic transitions. Mood: Cute, funny, heartwarming, slightly chaotic, and wholesome. Important Consistency Rules: Keep the woman's facial appearance, short black bob hairstyle, bangs, and brown outfit consistent throughout the video. Maintain the same gray tabby cat and its tiger-striped hat in every scene where it appears. Ensure realistic animal anatomy, natural movements, consistent character appearances, and believable interactions between the characters. Output: A high-quality cinematic short film in 16:9 aspect ratio, with realistic motion, detailed environments, expressive storytelling, and no artificial-looking animation.
This 11-second AI dance shows why character consistency matters more than complicated
Replace the person in {video} with my character from {reference sheet}. Character reference sheet prompt: Create a photorealistic character reference sheet using the uploaded photos. Show six views: full-body front, side and back, plus close-up face views from the front, side and back. Preserve the character’s exact facial features, hairstyle, proportions, outfit and distinctive details across every view. Use a clean light-gray background, soft studio lighting and a consistent appearance throughout.
Creature breaks out of cinema screen
THIS IS ONE SINGLE CONTINUOUS UNBROKEN SHOT FROM FIRST FRAME TO FINAL BLACKOUT. No cuts, transitions, fades, dissolves, or hidden edits. Any cut is a generation failure. 12-second ultra-realistic 9:16 smartphone video inside a packed luxury cinema. Shot on iPhone 17 Pro, handheld micro-shake, HDR, realistic low-light detail, deep focus, natural motion blur. No text, captions, logos, UI, or watermarks. SCENE: Warm amber cinema lighting, cream seats, dark central aisle, enormous curved screen. Audience packed on both sides, quietly watching. CHARACTER: Young South Asian man, 22–24, dark hair, light stubble, black turtleneck and black jeans. Face from. Seated beside the center aisle throughout. Camera always remains physically in his hand until the final impact. 0–3s: Camera faces the cinema screen. An ancient flooded underground ruin fills the movie screen. Through dark water, an enormous skeletal humanoid creature slowly rises, dripping, with elongated arms and a distorted skull. It walks directly toward the screen. 3–5s: Without any visual boundary, the creature physically steps OUT of the movie and onto the real cinema carpet. Water splashes onto the aisle. It is nearly ceiling-height. Audience members gasp, laugh nervously, and raise their phones. The creature continues slowly toward them. 5–7s: Camera naturally swings toward our character. His expression changes from amazement to nervous disbelief. He looks at the creature, then back at the camera. Camera smoothly swings back toward the aisle. The creature is now terrifyingly close. 7–9s: Low handheld angle as the creature passes between the seats. Wet footprints appear on the carpet. Everyone films it. Phones glow across the auditorium. Heavy footsteps vibrate the seats. Nobody runs yet. 9–10s: The creature suddenly FREEZES. Its head violently rotates toward our character. Its jaw tears open impossibly wide and releases a deafening supernatural ROAR. The physical pressure wave hits the entire auditorium simultaneously. Lights flicker, seats shake, people scream and recoil. 10–12s: The character's hand involuntarily jerks from the shock. The phone remains in his loose grip while the camera violently whips through random real fragments: screaming audience, ceiling lights, blurred creature, seats, glowing phones, carpet, creature lunging closer. No editing, only uncontrolled physical camera movement. His grip finally fails. Phone tumbles, spins once, violently strikes the seat. INSTANT BLACK SCREEN from physical impact. Complete silence. No fade. PHYSICS: Creature crosses from screen into reality seamlessly with identical perspective, lighting, scale, shadows, and floor contact. Water physically transfers from the movie world onto the real carpet. Every audience reaction follows the same roar pressure wave naturally. LIGHTING: Cool blue-green screen light mixed with warm amber cinema sconces. Creature becomes increasingly illuminated by real auditorium lighting after leaving the screen. Violent light flicker only after the roar. AUDIO: Subterranean water ambience, audience murmurs, wet footsteps, rising tension, sudden deafening roar with microphone clipping, synchronized screams, chaotic handling noise, phone impact, then absolute silence. STYLE: Photorealistic viral found-footage realism, imperfect smartphone cinematography, believable human reactions, realistic physics, terrifying scale, immersive spatial audio, no artificial cinematic cuts.
Seedance 2.5 Arcane-style Muay Thai fight from three reference images
Create a 30-second cinematic Muay Thai fight sequence using the three reference images strictly. REFERENCE LOCK: [1] FIGHTER #1 (BLACK SHORTS) — preserve the exact angular face, messy black hair, shirtless muscular build, white hand wraps, black shorts, and bare feet. [2] FIGHTER #2 (WHITE SHORTS) — preserve the exact angular face, messy brown hair, shirtless muscular build, white hand wraps, white shorts, and bare feet. [3] THE ARENA — preserve the exact traditional Muay Thai stadium, warm overhead spotlights, hanging banners, and weathered ring with red and white ropes. VISUAL STYLE: Premium Arcane-inspired 2D hand-painted digital artwork, hard-edge brushwork, thick painterly strokes, dramatic lighting, subtle 2.5D depth. Gritty underground Muay Thai atmosphere. Realistic weight and impact. No music, only heavy breathing, skin-on-skin strikes, canvas footsteps, and arena ambience. 0–5 SECONDS — RING STARE-DOWN Inside the warm, spotlight-lit arena [3], Fighter #1 [1] and Fighter #2 [2] stand in the center of the ring. Low-angle camera tracks upward along Fighter #1 [1]: bare feet → black shorts → taped fists → focused eyes. Both fighters raise their guards into classic Muay Thai stances. CUT. 5–10 SECONDS — OPENING STRIKES Fighter #2 [2] lunges forward with a fast jab-cross combo and a heavy low kick. Fighter #1 [1] checks the kick with his shin and counters with a snapping body kick. Sweat and dust burst off their skin under the overhead lights [3]. 10–15 SECONDS — CLINCH & KNEES Both fighters lock into a tight neck clinch. Fighter #2 [2] drives a brutal knee into Fighter #1's [1] ribs. Fighter #1 [1] twists their grip, off-balancing Fighter #2 [2], and lands two rapid counter-knees. 15–20 SECONDS — ELBOW EXCHANGE They break the clinch and trade razor-sharp close-range elbows. Slash from the right — Fighter #1 [1] leans back, grazing past his chin. Spinning back-elbow from Fighter #1 [1] — Fighter #2 [2] blocks high and stumbles back against the ropes [3]. 20–25 SECONDS — AERIAL ASSAULT Fighter #2 [2] rebounds off the ropes with a flying knee. Fighter #1 [1] sidesteps at the last millisecond and pivots hard on his bare foot. Close-up of Fighter #1's [1] eyes locking onto the opening. 25–30 SECONDS — THE KNOCKOUT SMACK! Fighter #1 [1] unleashes a devastating roundhouse kick directly to Fighter #2's [2] jaw. Fighter #2 [2] crashes onto the dusty canvas [3] as Fighter #1 [1] stands tall, chest heaving under the spotlights.
Knight rebuilds ruined cathedral with particles
Moonlit ruin→slipped blade→seal shatter→particle cathedral. Character reference, SERAINE from @[image 1]: chin length silver white short hair, a long fringe covering an eye, grey blue eyes, a cold and severe expression; silver plate armor covered in gilded scrollwork relief, a white cape, and a pale white kite shield slung on her back with a gold edged cross on its face. No horns on her head. Keep her face, hair, armor patterning, shield and sword design consistent for the whole shot. Light system (global rule, runs by the second): from 00:00 to 00:05, the only cold light in frame is blue moonlight falling through the collapsed dome, and the only warm light is a thin layer of white gold light on her blade; there is no other light source. At 00:05, the moment the seal shatters, the white gold light on her blade goes out and the sword turns back to cold steel. From that same moment, the particles pouring out of the seal give off white gold light themselves and become the main light source in frame; the particles come from the shattered seal, not from the sword. At 00:23 the rose window finishes assembling and light floods through its tracery into the nave. From 00:26 the particles solidify and the light settles into a soft, steady warm gold. Apart from the moonlight, the blade, the particles and the rose window, no other light source appears in frame. Brightness (global rule, runs by the second): from 00:00 to 00:05 it is very dark overall, with only the outline of the figure, the silhouettes of the broken pillars and the reflections off the standing water readable. The instant of the shatter at 00:05 must be clearly brighter than any frame before it. From 00:05 to 00:23, as the architecture grows back layer by layer, the frame gets brighter stage by stage, each stage brighter than the last. From 00:23 to 00:26 is the brightest stretch of the piece and must be clearly brighter than 00:05 to 00:23. From 00:26 to 00:30 it settles into a steady warm light, clearly darker than the brightest stretch but clearly brighter than the opening. The camera starts low behind her, with a corner of the kite shield on her back always pinned along the left edge of the frame. Deep night, the nave of a gothic cathedral left as a ruin: the dome has collapsed into a huge hole, the stone pillars are only broken stumps, vault ribs lie scattered across the floor, and the floor is slick, covered in rubble and standing water. Set into the flagstones at the exact center of the nave is a round gilded holy seal about 2 meters across. From the first frame SERAINE holds the sword in her right hand, the tip pointing down at her side, the whole sword clearly visible. She walks down the nave toward the seal. At 00:03 she steps on a slick piece of rubble, her foot slides out and she pitches forward, and by instinct she drives the sword down to catch herself; the tip sinks right into the center of the seal and the blade stands straight up there. Her hand then slips off the hilt and she drops to a knee at the edge of the seal. From then on the sword stays standing straight up in the center of the seal, never falling, never moving, and nobody touches it again. The camera drops to floor level with that downward thrust. The seal cracks outward in radiating lines from the point where the tip went in, and at 00:05 the whole seal shatters, the fragments breaking into countless tiny particles that pour out of the seal in every direction and wash across the entire nave floor like a tide. From then on she stays kneeling beside the seal until she stands at the end. The camera hugs the floor and pushes forward in the direction the particles flow. The particles reach the foot of every broken pillar and stack upward from the break, extending each pillar section by section. The camera climbs alongside a single growing pillar all the way up to the height of the dome. In midair the particles weave vault rib after vault rib, the ribs cross and close, patching the hole in the dome bay by bay. The camera glides along the closing ribs toward the high wall at the west end of the nave, where in the round opening the particles assemble a huge rose window petal by petal, finishing at 00:23. The camera slowly backs away from the rose window and descends, pulling back along the central axis of the nave until the whole rebuilt cathedral opens up in frame: rows of stone pillars, the closed dome, the rose window. As the particles settle they solidify into real grey white stone and gilded ornament. The seal reassembles whole in its original place, and the sword is still standing straight up in its exact center. SERAINE rises to her feet beside the sword, her back to the camera, head tilted up toward the rose window; she and the upright sword are just 2 small silhouettes in the vast cathedral. Frame requirements: the camera moves continuously through the whole piece with no edit cuts of any kind. All the architecture grows on from the existing breaks in the ruin and never appears out of thin air. Once the particles solidify they are real stone and metal, not translucent light. The sword's length and design stay the same the whole time, and after the shatter the blade stays cold steel and never turns into a column of light. The outfit follows the character sheet strictly, and the armor patterning, the cape and the shield design do not change
Robot transforms bed morning routine
Cinematic, realistic, funny morning routine video: I’m sleeping peacefully in bed when a futuristic robot rolls up beside me, its screen clearly displaying “6 AM,” and blasts an alarm. Half-asleep and annoyed, I punch the robot hard in a playful slapstick moment. The unharmed robot instantly pulls out a remote and transforms my bed into a wildly dancing, shaking sofa, jolting me fully awake. It then quickly and perfectly brushes my teeth, cleaning every tooth, before instantly preparing and handing me a fresh hot coffee, which I drink. Fast-paced visual comedy, expressive reactions, smooth transformations, realistic physics, cinematic camera movement. No dialogue, conversation, voiceover, or spoken communication. Only sound effects and upbeat background music. My image is attached here and it is @img1 . Use it completely and perfectly.
Ballerina with Peg-Leg in Urban Alley
Cinematic shot of an East Asian girl with braids dancing in a wet Tokyo alley. Wardrobe: kimono top, pink tutu, one wooden peg-leg, one ballet shoe. 0-16s: Graceful ballet movements and spins. Setting: damp asphalt with reflections, moody lighting, neon signs. Realistic textures.
Divine Ritual and Mythological Encounter
Cinematic mythological film. 0-4s: Woman in a gold saree performs Bharatanatyam in a stone temple with volumetric sunbeams. 4-8s: Close-up of feet with bronze anklets stomping in dust. 8-12s: Men with long hair perform a fierce fire dance at night. 12-16s: Extreme close-up of chanting lips with smoke. 16-20s: A massive blue-skinned four-armed deity emerges from a glowing temple portal. High-fidelity, warm firelight, dramatic shadows, epic scale.
A short story about Revenge. Video: Seedance 2.5 in Dreamina Characters
Style: Hyper-realistic cinematic Western revenge film, grounded gunfight choreography, realistic body mechanics, dramatic facial acting, dusty frontier atmosphere, sweeping cinematic camera, emotional storytelling, ultra-detailed costumes, practical effects, realistic physics. Duration: 30 seconds Aspect Ratio: 16:9 IMPORTANT Keep the SAME characters exactly as the reference sheets. — RED: long crimson coat, red hat covering his eyes, twin revolvers, calm, controlled, emotionally scarred. — SKULLY: black suit, skull mask, black hat, twin revolvers, cold and merciless. Everything must feel like a real Western film. No exaggerated action. Realistic gun handling. Heavy emotional acting. Fast cinematic editing during the flashback. ------------------------------------------------ [00:00–00:03] Extreme close-up of RED's face beneath the brim of his hat. Only his mouth and clenched jaw are visible. Wind blows his red coat gently. Dust drifts across the empty main street. Cut to an extreme close-up of SKULLY's skull mask. His fingers slowly tighten around his revolver grip. Silence except for wind and creaking wooden signs. ------------------------------------------------ [00:03–00:10] Rapid flashback montage. Fast multishots with quick cuts and bright white flash transitions. — RED is violently pinned to the muddy ground by three ruthless outlaws. — His hands claw desperately through the dirt. — SKULLY drags RED's terrified wife toward a waiting horse. — Close-up of her tear-filled face. — She stretches her hand toward RED. — She screams: "REEEEED!!" — RED screams back, struggling violently. — Boots stomp him back into the dirt. — Close-up of their fingertips almost touching but never reaching. — SKULLY mounts the horse with her. — Horse gallops away through clouds of dust. — RED continues screaming helplessly as the flashback burns away into white. ------------------------------------------------ [00:10–00:17] Back to the present. Wide cinematic shot of the deserted Wild West main street. Citizens quietly gather on wooden porches, balconies and outside the saloon. Bartender freezes while cleaning a glass. A child is pulled indoors. The sheriff watches silently. Dust devils roll between RED and SKULLY. Their coats sway gently. No one speaks. ------------------------------------------------ [00:17–00:21] Alternating close-ups. RED's hand hovers beside his revolver. SKULLY smiles beneath the skull mask. Close-up on his mouth. He quietly says: "You will meet your wife soon." Silence. Only wind. A loose tumbleweed rolls between them. Extreme close-up of both trigger fingers. ------------------------------------------------ [00:21–00:25] The duel. Ultra-fast multishots. Boots planting. Coats whipping. Hands drawing revolvers. Extreme close-up of spinning cylinders. Simultaneous gunshots. Massive muzzle flashes. Echoes across the town. Silence immediately follows. One second passes. SKULLY slowly drops to his knees. His revolver slips from his hand. He collapses face-first into the dusty street. ------------------------------------------------ [00:25–00:30] Ground-level POV beside the fallen SKULLY. His vision blurs. RED slowly walks into frame. His long crimson coat fills the screen. He stops over SKULLY. Camera tilts upward from the ground as RED points his revolver directly down at him. Close-up of RED's expression. Calm. Cold. Without emotion. He quietly says: "Not yet." A final gunshot. Cut instantly to black. Audio No music. Only cinematic Western sound design: constant desert wind, distant crows, creaking wooden buildings, leather movement, boots on dirt, horse snorts in the flashback, muffled crowd breathing, clothing rustle, revolver hammers cocking, realistic gunshots with long canyon echoes, shell casings hitting the ground, SKULLY collapsing into dust, and complete silence after the final shot. Negative prompts No slow motion except subtle emphasis during the flashback. No superhero movement. No unrealistic gunplay. No floating cameras. No modern objects. No blurry characters. No cartoon effects. No excessive blood. Prioritize emotional facial acting, cinematic framing, realistic pacing, and frequent shot changes with strong visual storytelling.
Where silence meets the moonlight, and the wind tells a story.
Cinematic slow-motion portrait of a woman in a massive, voluminous black dress with heavy layered folds, standing in profile on a vast white salt flat at dusk. A glowing crescent moon sits behind her against a steel-blue, overcast sky. Strong wind continuously whips her long, dark, wavy hair out to the right in flowing strands, while the heavy fabric ripples and sways at the hem. She stays almost still, gazing calmly to the left with a subtle head turn. Static locked-off camera with a very slight, slow push-in. Moody, desaturated teal-blue and charcoal palette, soft diffused light, high-fashion editorial, photoreal, shallow haze on the horizon, 24fps, 9:16. **Negative prompt (if supported):** > camera shake, fast movement, walking, morphing fabric, face distortion, extra limbs, text, watermark, flicker **Tips:** - Since the source is a Midjourney still, use it as the start frame (image-to-video) and keep the prompt focused on motion (wind, hair, fabric) rather than re-describing everything. - For Midjourney's video feature, "low motion" suits this best, with a short motion prompt like: *"wind blowing hair and dress fabric, subtle head turn, locked camera."* - If the hair moves too chaotically, add "gentle, consistent wind direction left to right." I can also write shorter versions tailored to Runway, Kling, Veo, or Midjourney if you tell me which one you're using.
Animals driving rally car chaos
Create a 10-second, vertical ultra-realistic smartphone recording inside a rally car driven by a dog. ONE continuous shot at normal speed. CAMERA AND CAR A fixed wide-angle view from high beside the front passenger-side windshield pillar, looking diagonally across the cabin. Keep the dog driving, the cat in the front passenger seat, the raccoon between the front seatbacks, the steering wheel, and the gear lever visible together. The recording device is outside the frame. The viewpoint stays attached to the car, vibrating and jolting with the chassis. A worn rally hatchback with turquoise front bucket seats, a bright-green steel roll cage, a simple dashboard, and ONE tall sequential gear lever mounted on the centre tunnel beside the driver. Ordinary daylight through the windows, natural exposure, slight wide-angle distortion, motion blur, and mild phone-video compression. Real fur, skin, upholstery, metal, and glass. ANIMALS DRIVER: a Chinese Crested dog with a mostly hairless pink-grey body and a fluffy white crest. It sits upright in the driver's bucket seat, secured by a fitted harness. The steering wheel and gear lever are within comfortable reach of its front paws. Its expression is intensely focused on the road. FRONT PASSENGER: a brown tabby cat secured in the passenger seat. Wide eyes, ears angled back, front paws braced on the seat cushion. It looks toward the road and occasionally toward the dog, opening its mouth for loud, protesting meows. REAR PASSENGER: a raccoon secured in the centre rear seat, its masked face visible between the front seatbacks. It braces its front paws against the back of the driver's seat while its head and shoulders bob over the bumps. All three retain realistic animal bodies, paws, faces, and proportions. CONTINUOUS ACTION Start immediately with the car ALREADY travelling very fast along a rough gravel rally road. Roadside trees and banks stream rapidly past the windows. High engine revs, tyre noise, and constant suspension movement establish speed immediately. They get shaken violently the whole way; the entire vehicle is bouncing and jolting—it’s total chaos. The dog makes quick, controlled steering corrections. Its fluffy crest shakes with the vibration. Twice during the recording, the dog keeps its LEFT front paw on the steering wheel, moves its RIGHT front paw onto the same gear lever, makes one short, decisive shift, and returns that paw to the wheel. Each shift produces a mechanical click and a brief drop in engine pitch before the revs rise again. The car pounds across successive ruts and bumps. Strong vertical jolts and brief sideways body roll move all three animals in response to the SAME vehicle motion. Their harnesses restrain their torsos while their heads, ears, and fur continue moving with inertia. The cat repeatedly meows in alarm, pressing its paws into the cushion. The raccoon sways and braces behind the driver. The dog remains determined, watching the road and correcting the steering. Near the end, a larger bump produces one pronounced suspension compression and rebound. All three animals bounce in their seats; the cat lets out its longest meow while the dog immediately steadies the wheel. End at 10 seconds with the car still racing forward at high speed. TONE AND SOUND The humour comes from the dog's serious driving and the passengers' believable animal reactions. Live sound only: a hard-revving engine, gear changes, gravel striking the underside, suspension thumps, cabin rattles, and clearly audible cat meows. No human speech, music, laugh track, or cartoon sound effects. CONTINUITY One fixed interior viewpoint, one lens, one uninterrupted recording. No cuts, exterior shots, slow motion, or zooms. The dog alone operates the controls. The animals remain in their assigned seats throughout. Preserve the same wheel, gear lever, cabin, and animal anatomy. No human hands, duplicated paws, morphing faces, captions, logos, watermarks, or interface.
1970s Shibuya horror short
【風格】 1970年代日本都市傳說恐怖電影（Japanese Urban Legend Horror Movie），經典日系恐怖片風格，昭和時代澀谷街景，高密度群眾演出，電影級分鏡，強烈戲劇張力，歡樂復古流行音樂逐漸扭曲變調為詭異恐怖配樂，緊湊快節奏剪輯，陰森都市怪談氛圍，9:16直式畫面。 【音樂】 開場：歡樂復古1970年代日本流行樂。 中段：音樂逐漸失真、降速、出現不和諧音。 結尾：低沉詭異弦樂、女性怪笑聲、電子故障音（Glitch）。 【總時長】15秒，中文對話 【場景】1970年代東京澀谷區熱鬧商店街 → 暗巷 → 都市傳說恐怖事件 【鏡頭1（00:00 - 00:03）】 熱鬧的1970年代澀谷街頭。 大量打扮前衛新潮的年輕男女逛街聊天。 地面散落報紙、飲料罐與垃圾。 突然一名穿著女僕咖啡廳制服的粉紅色長髮可愛女店員從狹窄暗巷衝出來。 她身穿低胸黑色與白色相間的女僕裝，搭配白色長手套、黑色網襪與高跟鞋，臉色慘白、滿臉恐懼。 一邊奔跑一邊尖叫： 「救命啊！有鬼！」下一秒她重重跌倒在地，當場昏迷。 周圍路人紛紛停下腳步圍觀。 【鏡頭2（00:03 - 00:06）】 一名醉醺醺的平頭流氓推開圍觀群眾。 肩膀露出龍紋刺青。 他粗暴地大喊： 「滾開！」人群驚慌後退。 流氓蹲到女店員旁邊。 不耐煩地用力搖晃她肩膀： 「小姐！妳醒醒！」 【鏡頭3（00:06 - 00:08）】 女店員虛弱睜開眼睛。 顫抖地說： 「有鬼……」她艱難抬起左手。 指向旁邊漆黑暗巷。 接著再次失去意識。 流氓順勢轉頭望向暗巷。 背景音樂突然停止。 【鏡頭4（00:08 - 00:10）】 暗巷深處。 垃圾堆旁飄浮著一名白衣長黑髮女鬼。 長髮遮住半張臉。 蒼白雙眼死死瞪著鏡頭。 周圍空氣微微扭曲。 遠方傳來低沉詭異笑聲。 【鏡頭5（00:10 - 00:12）】 流氓臉部超特寫。 他震驚得不自覺張開嘴巴。 頭緩緩歪向右側。 醉意瞬間消失。 額頭冒出冷汗。 【鏡頭6（00:12 - 00:14）】 女鬼也做出完全相同的動作。 嘴巴慢慢張開。 頭歪向右側。 接著發出尖銳而詭異的笑聲： 「呵呵呵呵呵……」下一秒。 女鬼突然高速朝鏡頭飛撲而來。 長髮與白衣瞬間佔滿整個畫面。 【鏡頭7（00:14 - 00:15）】 畫面突然發生強烈電子故障。 Glitch！ 黑白噪訊瘋狂閃爍。 尖銳雜音爆發。 畫面瞬間切成全黑。 只剩女鬼最後一聲詭異笑聲在黑暗中迴盪。
here's how to reverse-engineer higgs genjutsu in 2 mins: - get a reference video - create
Apply the source video1 motion to the character in the reference image1, keeping their clothing, the framing and the timing.