LogoSeadanse2.5

AI video effect prompts

Viral AI Video Effect Prompts

You saw a trend take off and want the exact prompt behind it. Copy our viral AI video prompts from 38 creator clips, or run these AI video effect prompts in our AI video maker with 40 starter credits.

Use Viral Prompts FreeZombie Match Cut effect

38 prompts38 clips31 creators credited11.2M views on Xfrom public posts on X

Turn Viral AI Video Prompts into Videos on Seedance 2.5

Press Use on any card to load its prompt into our AI video maker, add your photo, a character sheet, or a reference clip, pick your aspect ratio and length, and generate on Seedance 2.5.

0/20000
Inputs
Made with
Jump toFPV Fly-Through 5Zombie Outbreak 6Time Freeze 4Zombie Match Cut 4Magic Pen 3Hotel Lobby AI Rap Duo 3STORM Clone Dance 4Ordinary-Day Home Video 5Found Footage 438 of 38 prompts

FPV Fly-Through5

First-person flight at speed through canyons, ruins, cities and clouds.

Made with Seedance 2.015s

by @LudovicCreator22.7K viewsOriginal post

FPV dive through a storm-wrapped jungle temple

One continuous first-person flight between vines, statues and waterfalls that ends as the temple opens onto an underground sun with golden shockwaves and floating stone rings.

  • Text to video
  • FPV
  • One take
  • Fantasy
Video prompt458 chars
Immersive first-person flying camera perspective, no visible drone or flying device. The camera dives through an ancient jungle temple swallowed by storm clouds, racing between vines, stone statues, waterfalls, and glowing ruins. At the end, the temple opens into a massive underground sun, releasing golden shockwaves and levitating stone rings. Continuous POV shot, no cuts, aggressive banking, cinematic motion blur, volumetric mist, high-end fantasy VFX.

Why it works

  • It opens by removing the drone ('no visible drone or flying device'), so the view reads as pure flight.
  • The obstacles are a list the camera races between, which gives the flight its speed.
  • At under 500 characters it leaves timing and route to the model.
Made with Seedance 2.015s

by @LudovicCreator14.6K viewsOriginal post

Supersonic FPV run through a desert canyon into the upper atmosphere

A first-person flight inches above canyon rock, through narrow arches, over waterfalls and through a tunnel, then straight up through the clouds until the Earth stretches across the horizon.

  • Text to video
  • FPV
  • Canyon
  • Seedance 2.0
Video prompt792 chars
Pure first-person camera perspective, no visible drone, no cockpit, no aircraft visible. The camera flies at supersonic speed through an enormous desert canyon, hugging the terrain only inches above the rocks. Sonic booms echo through the cliffs while dust erupts behind every turn. The camera slices through impossibly narrow rock arches, skims over waterfalls, enters a dark canyon tunnel, then bursts back into daylight as multiple supersonic shockwaves ripple across the landscape. The final climb rockets vertically through thick clouds into the upper atmosphere with the Earth stretching across the horizon. Continuous one-take, authentic high-speed inertia, aggressive banking, cinematic motion blur, atmospheric distortion, volumetric dust, blockbuster Hollywood VFX, hyper-realistic.

Why it works

  • It strips every vehicle cue: no drone, no cockpit, no aircraft visible.
  • The route has a clear order (canyon, arches, waterfalls, tunnel, climb) with a change of light at the tunnel exit.
Made with Seedance 2.015s

by @umesh_ai7.5K viewsOriginal post

Stratosphere dive to a hypercar outrunning an avalanche

One unbroken aerial move: drop from the stratosphere to a futuristic car in a mountain pass, skim the ridge beside it as an avalanche follows, chase it into a tunnel and climb out to see the pass buried.

  • Text to video
  • One take
  • Avalanche
  • Seedance 2.0
Video prompt1,526 chars
An unbroken, continuous ultra-wide cinematic aerial shot begins in the stratosphere, revealing an immense frozen mountain range beneath glowing storm clouds. The camera plunges at breathtaking speed through layers of mist and snow, diving toward a sleek futuristic hypercar blasting into a narrow mountain pass. Its luminous body, adaptive aero surfaces, and electric thrusters cut through the blizzard as an enormous avalanche erupts from the peaks behind it, cascading into the valley like a towering white tsunami.
Without a single cut, the camera levels out inches above the icy terrain, skimming along the ridgeline parallel to the road while the car accelerates through sharp bends, snow exploding from beneath its wheels. The camera sweeps around the vehicle in a massive fluid arc, keeping both the hypercar and the advancing avalanche in frame as the collapsing mountainside grows closer and more violent.
The car rockets toward a dark tunnel carved into the mountain while chunks of ice and debris crash around it. At the final moment, the camera races alongside the car, swings directly in front of it, then pulls backward into the tunnel as the vehicle surges inside. The climax erupts as the avalanche slams into the tunnel entrance with overwhelming force. The camera instantly climbs through a vertical opening above the mountain, soaring into the sky to reveal the entire pass being swallowed beneath a vast, rolling ocean of snow, while the futuristic car’s glowing trail continues deep inside the mountain.

Why it works

  • It keeps both subjects in frame on purpose, the car and the advancing avalanche, during the big arc.
  • The camera's path is a run of verbs in order: plunges, levels out, sweeps, swings in front, pulls back, climbs.
  • Three paragraphs with no timestamps leave the pacing to the model.
Made with Seedance 2.530s

by @Naiknelofar7882.1K viewsOriginal post

Motocross through a ruined city, chased by an FPV camera

Six five-second blocks: a rider weaves down a wrecked street, launches off a collapsed road onto a rooftop, runs across rooftops and makes a last jump between buildings against a burning skyline.

  • Text to video
  • FPV chase
  • Motocross
  • 30 seconds
Video prompt2,688 chars · 6 beats
Create a 30-second ultra-realistic cinematic FPV action sequence following a professional motocross rider racing through a devastated abandoned city. Aggressive drone movement, realistic motorcycle physics, flying debris, dust, destruction, dramatic sunset light, detailed environments, intense speed, cinematic motion blur, 4K. No text, logos, or watermarks.
[0–5s] — THE RACE
FPV camera races directly behind a motocross rider accelerating through a ruined city street. Abandoned cars, shattered windows, collapsed concrete, and debris surround the road. The rider weaves between obstacles at high speed as dust and fragments explode from the tires.
[5–10s] — CLOSE TRACKING
The camera dives aggressively alongside the motorcycle, tracking inches from the spinning wheels and suspension. The rider powers through rubble, sparks flying as the bike clips broken metal and concrete. The camera rapidly swings back behind the rider without losing momentum.
[10–15s] — THE RAMP
The rider spots a massive collapsed section of roadway forming a natural ramp. He accelerates hard and launches into the air. The FPV camera follows upward, revealing rooftops and the devastated skyline ahead.
[15–20s] — ROOFTOP LANDING
The motorcycle lands heavily on a damaged rooftop, suspension compressing realistically. The rider immediately accelerates across the roof, jumping over cracks, fallen beams, and gaps between broken structures. Dust trails behind the rear tire.
[20–25s] — ROOFTOP RUN
The camera flies extremely close beside the bike as it blasts across multiple interconnected rooftops. The rider narrowly clears obstacles and jumps another gap while chunks of concrete fall into the streets far below. The camera rolls dynamically with the motorcycle.
[25–30s] — FINAL JUMP
The rider reaches the edge of a high rooftop and launches into a final massive jump between two buildings. The camera pulls back into a wide cinematic view, revealing a vast ruined skyline glowing with intense fire and smoke behind him. The motorcycle hangs momentarily against the burning horizon before landing on the opposite rooftop. End with a dramatic cinematic stop.
STYLE: ultra-realistic motocross action, authentic motorcycle suspension and tire physics, realistic rider movement, high-speed FPV drone cinematography, dynamic camera rolls, natural motion blur, flying dust and debris, detailed destruction, volumetric smoke, dramatic cinematic lighting, realistic fire glow, immersive scale, photorealistic textures, 4K HDR.
AUDIO: powerful motorcycle engine, tire impacts, suspension movement, flying debris, wind rush, concrete crashes, distant explosions and fire ambience. No music, no dialogue.

Why it works

  • Each block has one job: race, close tracking, ramp, landing, rooftop run, final jump.
  • The physics words go to the bike: suspension compressing on landing, sparks when it clips metal.
  • The last block pulls the camera wide for the final jump after five blocks of tight following.
Made with Seedance 2.515s

by @SebJefferies377 viewsOriginal post

Night FPV drone flight around the Statue of Liberty

Four blocks over 15 seconds: skim the harbor two to three meters up, climb the front of the statue, orbit the torch 180 degrees with Manhattan behind it, then pull back for the wide reveal.

  • Text to video
  • FPV drone
  • Night
  • Landmark
Video prompt4,386 chars
Create a **15-second ultra-realistic cinematic FPV drone sequence flying around the Statue of Liberty at night in New York City.**
The footage should feel like it was captured using a **high-end professional cinema drone**, with breathtaking scale, realistic physics, smooth but exhilarating movement, deep atmospheric perspective, and premium feature-film cinematography.
**Location:** Liberty Island, New York City, USA.
**Time:** Late night, approximately 10:30 PM.
The sky is deep navy blue with thin scattered clouds. Manhattan glows brilliantly across the dark harbor. Thousands of realistic building lights reflect across the moving surface of the Hudson and Upper New York Bay.
The Statue of Liberty is beautifully illuminated by its real architectural lighting, standing dramatically against the dark skyline.
### 0:00–0:04 — THE APPROACH
Begin extremely low above the dark harbor water.
The drone races forward only **2–3 meters above the water**, creating an intense sense of speed.
Realistic water reflections streak beneath the camera.
Small waves move naturally.
The illuminated Statue of Liberty is visible far ahead, initially small in frame.
The drone rapidly approaches Liberty Island.
Camera slowly tilts upward as the statue becomes enormous.
Maintain realistic FPV banking and aerodynamic movement.
### 0:04–0:08 — THE CLIMB
As the drone reaches the island, it performs a **smooth aggressive upward climb along the front of the Statue of Liberty.**
Fly past the pedestal.
Continue upward alongside the folds of her robe.
The enormous statue fills most of the frame.
The camera subtly rolls as the drone climbs.
Pass close enough to reveal realistic copper surface texture, weathering, seams, shadows and architectural illumination.
The Manhattan skyline gradually appears behind her.
The scale should feel MASSIVE.
### 0:08–0:12 — TORCH ORBIT
The drone reaches the statue's face and continues upward toward the raised torch.
Perform a breathtaking **180-degree banking orbit around the torch**, keeping it close in the foreground while New York City rotates into view behind it.
The golden torch illumination contrasts beautifully against the deep-blue night.
Realistic atmospheric haze surrounds distant Manhattan.
Boats appear as tiny moving lights across the harbor far below.
No impossible instant acceleration.
Maintain believable drone momentum, inertia and banking.
### 0:12–0:15 — EPIC REVEAL
After completing the orbit, the drone rapidly pulls backward and slightly upward.
Reveal the **entire Statue of Liberty from above and behind**, Liberty Island below, dark water surrounding it, and the enormous illuminated Manhattan skyline stretching across the horizon.
The camera gradually levels.
The statue becomes a powerful silhouette against New York's lights.
Finish with an enormous cinematic wide shot.
**CUT TO BLACK.**
## CAMERA + VISUAL QUALITY
Ultra-photorealistic 4K cinematic drone footage.
24fps.
Natural cinematic motion blur.
Wide-angle **18–24mm equivalent lens**.
Realistic FPV drone physics.
Smooth acceleration and deceleration.
Subtle aerodynamic vibration.
Natural banking.
Realistic parallax.
Physically accurate nighttime exposure.
Realistic architectural lighting.
Deep blacks without crushed shadow detail.
Beautiful highlight roll-off.
Realistic water physics and reflections.
Atmospheric haze around Manhattan.
No artificial neon colors.
No oversaturated HDR.
No fake buildings.
No duplicated skyline.
No warped Statue of Liberty.
No changing statue proportions.
No floating objects.
No daytime lighting.
No impossible camera teleportation.
No cinematic black bars.
## SOUND ON
Immersive nighttime New York atmosphere.
Realistic drone wind rush increasing during acceleration.
Deep harbor ambience.
Water movement below.
Very distant boat horns.
Subtle distant city ambience.
Wind intensifies dramatically during the vertical climb and torch orbit.
Add a **powerful cinematic score that gradually builds from atmospheric tension into an epic orchestral rise**, reaching its peak exactly during the final Manhattan skyline reveal.
The final result should feel like an **expensive opening shot from a major Hollywood movie filmed over New York City**, with the Statue of Liberty dominating the sequence and the final skyline reveal providing the visual payoff.

Why it works

  • Heights and angles are numbers: 2–3 meters above the water, a 180-degree orbit, an 18–24mm lens.
  • The negative list targets landmark failures: no warped statue, no changing proportions, no duplicated skyline.
  • It is built on one real landmark and asks for an orchestral score; the four-move plan works on any tall subject.
Prompt source

Zombie Outbreak6

An ordinary place — a mall, a classroom, a cinema, a highway — overrun in one continuous take.

Made with Seedance 2.525s

by @auqibhabib428.5K viewsOriginal post

Zombie attack in a mall food court, met by a security guard

Eight timed blocks of three or four seconds: an infected man attacks a shopper in the food court, panic spreads, and a security guard holds off two infected with a crowd-control barrier and gets survivors behind a storefront.

  • Text to video
  • Mall outbreak
  • Handheld
  • Timed blocks
Video prompt3,672 chars · 8 beats
Create a photorealistic live action horror short film set inside a busy modern shopping mall during normal operating hours. The scene must feel like a real emergency captured by a human handheld camera operator, not a choreographed action movie. Prioritize realistic human behavior, believable physics, natural body movement, consistent character appearance, spatial continuity, practical effects and authentic environmental detail.
MAIN CHARACTER: A male mall security guard in his late 30s wearing a realistic dark security uniform and carrying a radio. Keep his face, clothing, body proportions and equipment identical throughout the entire video.
SURVIVORS: A small group of ordinary shoppers with realistic clothing and consistent appearances. Their faces, clothing and body proportions must remain unchanged throughout the sequence.
0–4 SECONDS:
Begin immediately in the food court with shoppers eating and walking normally. Without warning, an infected man suddenly attacks a nearby shopper. The attack is chaotic and physically messy rather than theatrical. The victim falls against a table, knocking over food and drinks. Nearby shoppers freeze for a moment because they do not understand what is happening.
4–8 SECONDS:
Several shoppers back away. One person briefly tries to help before realizing the attacker is dangerous. Chairs scrape across the floor as people retreat. The security guard hears the disturbance and runs toward the food court. His reaction is confused and concerned, not heroic.
8–12 SECONDS:
The infected attacker suddenly rushes toward another shopper. The crowd finally realizes something is seriously wrong. Panic spreads naturally. People run in different directions instead of moving together. Some hide behind storefronts while others run toward exits. Shopping bags and personal belongings fall naturally as people flee.
12–16 SECONDS:
The security guard reaches the food court and sees two infected people approaching. He grabs a nearby metal crowd-control barrier for protection. He initially backs away while watching them. One infected suddenly charges. The guard swings the barrier defensively and knocks the attacker off balance. The movement must look heavy, awkward and physically difficult, never like professional choreography.
16–20 SECONDS:
A few shoppers help each other retreat behind a storefront entrance. One person pulls an injured shopper away from the food court. Nobody performs exaggerated combat moves. The guard keeps backing toward the survivors while maintaining distance from the infected.
20–24 SECONDS:
One infected suddenly charges again. The guard sidesteps at the last moment. The infected crashes into the glass railing from its own momentum. The glass cracks realistically but does not explode into an unrealistic shower of fragments. The guard immediately moves the survivors farther away.
24–27 SECONDS:
The survivors move together toward a partially closed storefront. The guard stays behind them and keeps watching the infected. Distant screaming comes from another section of the mall, suggesting the outbreak is larger than what the camera has witnessed without showing an unnecessary crowd of zombies.
27–30 SECONDS:
The group reaches temporary safety behind the storefront. The guard slowly looks through the gap toward the food court. The mall has become eerily quiet. An escalator continues running in the background while abandoned shopping bags, overturned chairs and spilled food remain scattered across the floor. A distant infected figure briefly crosses the far end of the mall. The guard quietly raises one finger, signaling everyone to stay silent.Cut naturally to black.

Why it works

  • The guard is written as ordinary, 'confused and concerned, not heroic', and his barrier swing must look 'heavy, awkward'.
  • Damage is capped on purpose: the glass railing cracks but does not shatter, and distant screams suggest a bigger outbreak without a crowd of zombies.
  • It ends on a small action, one raised finger asking for silence, rather than a big beat.
Made with Seedance 2.530s

by @azed_ai144.5K viewsOriginal post

School zombie rescue played as a third-person survival game

A student fights through her school with a broken broom handle to reach three friends in classroom 2-B and lead them out the rear exit, written as gameplay with a fixed HUD, Japanese dialogue and timed blocks.

  • Text to video
  • Gameplay HUD
  • Japanese dialogue
  • Rescue mission
Video prompt14,771 chars · 8 beats
Create an ultra-realistic AAA third-person survival-action game sequence about a Japanese high school student fighting through a zombie outbreak inside her school to rescue three friends trapped upstairs. The entire sequence must feel like authentic playable gameplay rather than a cinematic trailer. Her objective is not to hunt or eliminate every zombie, but to physically reach her friends, protect them after finding them, and escape the school together using defensive melee attacks, dodging, environmental distractions, barricades, stealth, stamina management, and intelligent movement through the environment.
Character A: Player Character: An 18-year-old Japanese high school student with dark hair tied into a practical ponytail, wearing a dark school blazer, white shirt, pleated skirt, knee-high socks, practical school shoes, and a small backpack. She begins frightened and exhausted but becomes increasingly determined as she gets closer to her friends.
Characters B, C, D: Friends: Three Japanese students from the same class trapped together inside classroom 2-B. Character C has an injured ankle and cannot run normally, forcing Characters B and D to support them during the escape.
Maintain identical character models, uniforms, hairstyles, injuries, weapons, school geography, zombie designs, lighting conditions, and gameplay UI throughout the entire sequence.
[2] GLOBAL SETTING
A large modern Japanese high school during late afternoon after a sudden zombie outbreak, beginning at the ground-floor entrance and progressing continuously through the main corridor, central staircase, science laboratory, preparation room, secondary staircase, second-floor corridor, classroom 2-B, rear staircase, and emergency exit.
The school remains recognizable and grounded despite the outbreak. Fluorescent lights flicker above abandoned school bags, scattered notebooks, overturned desks, club posters, shoe lockers, vending machines, classroom signs, cleaning equipment, and emergency lights, while warm orange sunset enters through large windows and contrasts with the cold fluorescent interior lighting.
Visual Style: Photorealistic AAA Unreal Engine 5 survival-action gameplay with realistic Japanese school architecture, believable human characters, highly detailed uniforms, physically simulated hair and clothing, cinematic global illumination, polished floor reflections, natural motion blur, volumetric dust, atmospheric lighting, responsive debris, grounded body mechanics, realistic facial animation, restrained environmental destruction, and smooth 60fps gameplay presentation.
Gameplay Camera: Maintain an authentic over-the-shoulder third-person gameplay camera behind Character A throughout the sequence, with controlled camera rotation around corners, slightly wider framing during group encounters, subtle impact shake during zombie attacks, tighter framing during stealth and defensive counters, and natural responses to running, crouching, dodging, blocking, and environmental interaction. Never transform into a detached cinematic camera.
Gameplay UI: Keep player health and stamina fixed in the upper-left corner, circular mini-map in the upper-right, equipped improvised weapon and healing item in the lower-left, contextual interaction prompts in the lower-right, and the current mission objective beneath the mini-map.
Initial objective:
RESCUE YOUR FRIENDS: CLASSROOM 2-B
After rescue:
ESCAPE THE SCHOOL: REAR EXIT
Combat System: Character A uses a broken wooden broom handle as her primary improvised defensive weapon. Combat revolves around dodging, blocking, pushing, brief counters, distractions, barricades, and creating enough distance to continue moving. Character A is a vulnerable student surviving through timing and intelligence, never an unstoppable action hero.
Zombie Behavior: Recently infected students and school staff move with physically believable but disturbed human locomotion, including uneven walking, sudden short lunges, aggressive grabbing, stumbling recovery, and logical reactions to sight and sound. Zombies must physically navigate corridors, classrooms, doors, and staircases without teleporting or spawning impossibly ahead of the player.
Dialogue Language: Every spoken character line must be delivered exclusively in natural conversational Japanese with authentic pronunciation, emotional delivery, and age-appropriate speech patterns. Never generate spoken English. Do not add subtitles, translations, captions, or floating dialogue text. English is allowed only inside the AAA gameplay HUD, objectives, interaction prompts, and mission-completion UI.
Audio Design: Fluorescent electrical hum, footsteps echoing through corridors, distant alarms, school bell echoes, nervous breathing, desks scraping, zombie groans, doors slamming, broken glass, broom-handle impacts, phone audio, frightened Japanese dialogue, UI feedback, and restrained survival-horror music that intensifies naturally during danger.
[3] TIMESTAMPED STORYBOARD
[00:00-00:04] THEY'RE WAITING
Begin directly in playable third-person movement as Character A enters the deserted ground-floor corridor holding the broken broom handle defensively while her phone remains on speaker, and the objective marker for classroom 2-B appears on the mini-map.
Character B's frightened voice comes through the phone.
Character B: 「今どこ？」
Character A continues toward the staircase.
Character A: 「学校の中。」
Character B answers quickly.
Character B: 「2-Bにいる。」
Character A looks toward the objective marker.
Character A: 「そこにいて。今行くから。」
A violent impact suddenly shakes a classroom door farther down the corridor, forcing Character A to stop and raise the broom handle as the combat music begins.
[00:04-00:08] FIRST INFECTED
The classroom door bursts partially open and an infected student stumbles into the corridor, notices Character A, and accelerates toward her while the gameplay camera tracks backward behind her shoulder.
The zombie lunges.
A DODGE prompt flashes.
Character A sidesteps at the final possible moment, causing the infected student to collide with the lockers before she delivers one controlled broom-handle strike that pushes it away without graphic injury.
Character B hears the collision through the phone.
Character B: 「今の音、何？」
Character A immediately runs toward the staircase.
Character A: 「聞かないで。」
[00:08-00:12] BLOCKED
Character A reaches the central staircase but immediately stops when several infected students appear below and begin moving upward toward her, making the direct route to the second floor impossible.
She quietly backs away before they fully notice her.
Character B speaks through the phone.
Character B: 「もう近い？」
Character A watches the infected climbing.
Character A: 「近かったんだけどね。」
The objective updates:
FIND ANOTHER WAY UP
Character A notices the nearby science laboratory, enters quickly, and closes the door behind herself just before the first infected reaches the corridor.
[00:12-00:16] SCIENCE LAB
Character A pushes a rolling equipment cart against the laboratory door while zombie hands begin striking the opposite side, then checks the mini-map and discovers another staircase accessible through the laboratory preparation room.
An infected staff member blocks the preparation-room exit.
Rather than fighting directly, Character A grabs a metal tray and throws it toward the opposite side of the laboratory.
The tray crashes loudly.
The infected staff member turns toward the sound.
Character A quietly slips behind it.
Character B whispers through the phone.
Character B: 「お願い、早く来て。」
Character A reaches the preparation-room door.
Character A: 「急いでる！」
She closes the door behind herself and reaches the secondary staircase.
[00:16-00:20] CLASSROOM 2-B
Character A runs upstairs into the second-floor corridor, notices several infected silhouettes wandering near the opposite end, crouches behind overturned desks, and carefully advances toward classroom 2-B without attracting them.
She reaches the door and knocks softly.
Character A: 「私だよ。」
The door immediately opens and Character B pulls her inside.
The three friends surround her with visible relief while Character C remains seated beside a desk because of the injured ankle.
Character B notices the damaged broom handle.
Character B: 「それが武器？」
Character A catches her breath.
Character A: 「今のところ役に立ってる。」
Character D looks nervously toward the hallway.
Character D: 「もう行ける？」
Character A nods while helping Character C stand.
Character A: 「うん。みんな一緒に行くよ。」
The objective updates:
ESCAPE THE SCHOOL: REAR EXIT
[00:20-00:24] STAY TOGETHER
Character A carefully opens the classroom door and checks both directions before leading everyone into the corridor while Characters B and D support injured Character C between them, making the group's movement noticeably slower than Character A's earlier solo movement.
A zombie suddenly emerges from a nearby classroom and moves directly toward Character C.
Character A immediately steps between them, blocks the zombie's reaching arms with the broom handle, redirects its momentum into an overturned desk, and pushes the desk against the classroom doorway to temporarily trap it.
Character B looks at Character A while everyone continues moving.
Character B: 「いつからそんなことできるの？」
Character A follows closely behind them.
Character A: 「今日から。」
Character B gives a tiny nervous smile before helping Character C continue toward the rear staircase.
[00:24-00:27] DON'T STOP
The previous struggle attracts several infected students from farther down the corridor, and Character A looks backward long enough to understand that fighting the entire group would be impossible.
Character A: 「止まらないで！」
Characters B and D move faster while continuing to support Character C.
Character A stays behind the group and pushes an overturned chair across the corridor, creating a temporary obstacle that forces the pursuing infected to climb around it rather than magically stopping them.
Character B turns back.
Character B: 「あなたは？」
Character A immediately begins running after them.
Character A: 「すぐ後ろにいる！」
The four students reach the rear staircase together while the zombies continue pursuing from the same corridor behind them.
[00:27-00:30] LAST ONE OUT
The group descends the rear staircase toward the emergency exit while the infected continue following through the same stairwell behind them, keeping the geography and chase direction completely continuous.
Character D reaches the bottom first and pushes open the heavy emergency door.
Warm sunset light floods inside.
Character D: 「早く！」
Characters B and D help Character C outside while Character A remains behind them with the damaged broom handle raised.
Character B reaches back through the doorway.
Character B: 「こっち！早く！」
Character A grabs Character B's hand and crosses outside.
The emergency door slams shut seconds before the pursuing zombies reach the opposite side.
All four students stand together in the courtyard, exhausted and breathing heavily.
Character B looks at Character A.
Character B: 「本当に戻ってきてくれたんだね。」
Character A looks at all three friends.
Character A: 「当たり前でしょ。」
Character B gives her a relieved smile.
The objective completes:
FRIENDS RESCUED
A new objective immediately appears:
REACH THE EVACUATION POINT
The gameplay camera naturally settles behind Character A as the four friends turn toward the distant school gate, keeping the player in control and establishing that their survival continues.
[4] GLOBAL SUPPLEMENT
The entire sequence must function as one continuous AAA playable rescue mission rather than disconnected action scenes. Character A physically progresses from the ground-floor corridor to the blocked central staircase, science laboratory, preparation room, secondary staircase, second-floor corridor, classroom 2-B, rear staircase, and emergency exit. Every cut must begin from the exact physical position, direction of travel, environmental state, and immediate objective created by the previous cut.
Character A's objective remains emotionally and mechanically clear from beginning to end: she deliberately entered the infected school because her friends are trapped inside. She never stops to hunt zombies unnecessarily, explore unrelated areas, or behave like an action hero. Every confrontation occurs because an infected person physically blocks her route or threatens one of her friends.
All spoken dialogue must remain exclusively Japanese. Characters should never suddenly switch to English. Dialogue must remain short, conversational, emotionally believable, and naturally integrated into movement rather than interrupting gameplay. Allow breathing, hesitation, facial reactions, footsteps, environmental noise, and physical actions between spoken lines. Never generate subtitles or translated dialogue on screen.
After Characters B, C, and D are rescued, all three must remain physically visible throughout the rest of the sequence. Character C continues limping because of the injured ankle while Characters B and D support them, making their reduced movement speed a genuine gameplay challenge that forces Character A to protect the rear of the group.
Zombie behavior must remain spatially logical. Infected characters respond to sight and sound, physically navigate accessible corridors and staircases, collide naturally with doors and furniture, and continue pursuing from locations where they were previously established. Never teleport zombies ahead of the group or spawn enemies directly beside the player simply to manufacture another encounter.
Maintain persistent environmental consequences throughout gameplay. The laboratory cart remains against the door, displaced desks and chairs stay displaced, thrown objects remain where they land, Character A's broom handle becomes progressively damaged, her ponytail and uniform become slightly messier, and player health and stamina remain consistent rather than resetting between cuts.
Keep the HUD permanently anchored to the screen throughout camera movement. Maintain identical character models, Japanese school uniforms, hairstyles, Character C's ankle injury, school architecture, zombie appearance rules, weapon design, lighting conditions, mini-map layout, objectives, inventory state, environmental damage, and gameplay mechanics.
Avoid gore, graphic wounds, dismemberment, sexualized school uniforms, unrealistic martial arts, superhuman strength, firearms, impossible acrobatics, teleporting zombies, duplicated characters, disappearing friends, random cinematic cutscenes, inconsistent geography, floating objects, unreadable UI, spoken English, subtitles, captions, anime rendering, or exaggerated zombie mutations.

Why it works

  • Language is split by layer: every spoken line is Japanese, and English is allowed only in the HUD, objectives and prompts.
  • Consequences persist between cuts: the cart stays against the door, the broom handle gets more damaged, and health does not reset.
  • Zombies must walk to where they appear, with no spawning ahead of the player, which keeps the route through the school believable.
Prompt source
Made with Seedance 2.527s

by @ChillaiKalan__32.9K viewsOriginal post

Highway zombie escape, written as JSON

A short JSON prompt: a woman in a yellow tracksuit with a pink bat runs from a horde down a jammed highway, fights off one infected and carries a toddler away as tankers explode.

  • Text to video
  • JSON prompt
  • Highway
  • Short prompt
Video prompt1,429 chars
{
"scene_description": "Post-apocalyptic highway zombie breakout and escape",
"subject": {
"character": "Young woman with athletic build, distinct pink highlights in short dark hair, wearing a vibrant yellow zip-up tracksuit with green side stripes",
"secondary_character": "Young toddler in casual clothes held securely in her arms",
"weapon": "Pink patterned baseball bat"
},
"environment": {
"location": "Broad sunlit highway jammed with abandoned, smoldering vehicles and overturned fuel tankers",
"hazards": "Massive chasing zombie horde, aggressive stray dogs, collapsing overhead signs, explosive fireballs, thick black smoke plumes",
"atmosphere": "Tense, chaotic, high-energy survival action"
},
"action_sequence": [
"Forward tracking camera capturing the protagonist sprinting directly away from the pursuing mob",
"Spins back to strike a lunging infected enemy with the bat followed by a martial arts roundhouse kick",
"Low-angle swooping shot as a military attack helicopter glides directly above",
"Rescues a child into her arms and continues running toward the foreground as background tankers explode"
],
"cinematography": {
"camera_movement": "Dynamic backward tracking shot, low-angle to eye-level, smooth action gimbal",
"visual_style": "High-budget Hollywood blockbuster, hyper-realistic physics, cinematic color grading, 24fps motion blur"
}
}

Why it works

  • Fields keep subject, environment, action and camera apart, and the hazards (dogs, falling signs, fireballs) get their own key.
  • The action list is only four entries, one per moment, which reads like a shot list.
  • No length or timing is given, so how the clip is paced is up to the model.
Made with Seedance 2.530s

by @john_my075.2K viewsOriginal post

Outbreak inside a Japanese cinema during a late screening

Four scenes with time ranges: a drop of blood next to a motionless man in the audience, a crash at the emergency exit, a staff member fighting back with a metal stanchion, and a groan from behind the projection-room door.

  • Text to video
  • Cinema hall
  • Four scenes
  • Twist ending
Video prompt2,929 chars · 4 beats
Create a 30-second ultra-realistic Japanese horror-action sequence set inside a modern cinema hall during a late-night screening.
Photorealistic live-action, Japanese setting, practical-effects horror, authentic human reactions, natural handheld cinematography, realistic physics, cinematic lighting, and continuous scene progression. The footage should feel like a real horror movie captured by a professional camera crew, not AI-generated.
SCENE 1 — SOMETHING IS WRONG (0–7s)
A nearly full Japanese cinema hall. Movie light flickers across the audience. A female employee in her late 20s, wearing a dark cinema staff uniform, notices a man sitting several rows ahead behaving strangely. He suddenly stops moving. A drop of blood falls onto the armrest beside him. The audience slowly turns toward him. A faint, unnatural breathing sound emerges.
SCENE 2 — THE OUTBREAK (7–15s)
Without warning, a violent crash comes from the cinema's rear emergency exit. The door buckles open as a zombie-like infected person stumbles inside. People scream and rush between the seats. Before anyone can escape, the previously motionless man suddenly convulses and transforms, revealing that the infection was already spreading inside the audience. The camera moves with the panicking crowd in a chaotic but believable handheld shot.
SCENE 3 — FIGHT FOR SURVIVAL (15–25s)
The female employee grabs a heavy metal stanchion barrier and defends herself against an approaching infected person. Two audience members join her, using cinema seats and a fallen metal sign as improvised shields. They struggle desperately, pushing the attackers backward. No superhero movements—every hit has weight, imperfect timing, exhaustion, and genuine fear. They eventually knock down the attackers. The cinema falls completely silent except for heavy breathing and the faint hum of the projector.
SCENE 4 — FINAL TWIST (25–30s)
Everyone freezes. From behind the closed projection-room door, a low, distant zombie groan echoes through the empty cinema. The female employee slowly turns toward the sound. The projector light flickers across her terrified face. The camera moves closer as something scratches against the door from the other side.
CUT TO BLACK.
VISUAL & AUDIO DIRECTION:
Ultra-realistic live-action, practical makeup and physical effects, natural skin textures, believable Japanese cinema architecture, realistic emergency lighting, subtle film grain, imperfect handheld camera movement, physically accurate collisions, and authentic crowd panic. Diegetic sound only: screams, footsteps, furniture impacts, breathing, projector hum, and distant groans. No music, narration, subtitles, watermark, random text, excessive gore, graphic injuries, animation, anime, cartoon aesthetics, CGI appearance, distorted anatomy, or unrealistic action choreography. Maintain consistent character appearances and wardrobe throughout the entire sequence.

Why it works

  • The twist is set up before it pays off: the man who stops moving in Scene 1 turns in Scene 2.
  • Sound is diegetic only (screams, breathing, projector hum), with music ruled out.
  • It ends on a sound behind a closed door, so the last threat stays off screen.
Made with Seedance 2.530s

by @Diplomeme2.3K viewsOriginal post

Outbreak in an Iranian university lecture hall

An outbreak in a lecture hall: a drop of blood on a notebook, a zombie through the ceiling, a student who turns, a professor fighting back with a bench and a groan from the hallway.

  • Image input
  • Lecture hall
  • Reference image
  • Twist ending

Inputs Reference image of the female professor

Video prompt1,538 chars
Create an ultra-realistic Iranian university horror-action sequence featuring the female professor from the reference image. She is teaching a university class when a single drop of blood lands on a student’s notebook. Everyone looks up, but the professor notices one student behaving strangely. Suddenly, the classroom ceiling cracks and a zombie crashes into the lecture hall, triggering immediate panic.
The twist: another student is already infected and suddenly transforms, revealing the outbreak has started from inside the university classroom. Students rush toward the exit while the professor grabs a heavy wooden bench and fights back. Several students join her, using chairs, desks, and other classroom furniture to defend themselves. They eventually knock down the attacking zombies and the lecture hall falls silent.
Just as everyone thinks it’s over, a faint zombie groan comes from the university hallway outside. The professor slowly looks toward the blocked classroom door. Cut to black.
Photorealistic live-action, practical-effects horror, realistic Iranian university classroom, authentic Iranian university architecture and interior details, realistic lecture hall, Iranian students and contemporary university clothing, realistic desks, chairs, notebooks, whiteboard, projector, natural handheld camera, authentic reactions, realistic physics, diegetic sound only. No music, narration, watermark, excessive gore, graphic injuries, animation, anime, cartoon, CGI look, distorted anatomy, subtitles, or random text.

Why it works

  • The professor comes from your reference image, so the words go to local detail: Iranian university architecture, students' clothing, whiteboard and projector.
  • It is three short paragraphs with no timestamps, so the pacing is left to the model.
Made with Seedance 2.548s

by @magic_ai_skill1.2K viewsOriginal post

Tokyo outbreak in two halves: Skytree deck, then Sensoji

Fifteen seconds in a dark Skytree observatory and fifteen on the Sensoji approach, ending on one continuous shot from inside a shop as a zombie breaks the window from the street and comes in.

  • Text to video
  • Two locations
  • One spoken line
  • Tokyo
Video prompt9,647 chars
30 Seconds of Extreme Live-Action Horror — Tokyo: Zombies Overrun Skytree and Sensoji. 16:9 landscape cinema. Text-only generation, no reference images.
[OBSERVATORY SURVIVOR]
Adult woman, blue jacket, white trainers, black hair, normal skin and eyes. Terrified, she rescues her fallen friend and flees downstairs.
[OBSERVATORY FRIEND]
Adult man, gray hoodie, black trousers, normal skin and eyes. He escapes with her and never becomes infected.
[OBSERVATORY PURSUER]
An adult male zombie in dark gray clothes. Gray face, cloudy white eyes, dark facial and neck veins, red liquid around his mouth from his first appearance. He runs on foot along the aisle.
[ASAKUSA SURVIVOR]
Adult man in an ochre jacket, fleeing from Sensoji into the shop. Preserve his face, clothes, normal skin and eyes.
[ASAKUSA PURSUER]
An adult male zombie in a torn black zip-up jacket. Gray face, cloudy white eyes, dark veins and red liquid around his mouth. He chases the ochre-jacketed man, breaks the shop window from outside and lunges into the room.
[SHOP CLERK]
An uninfected adult woman in a green apron. Inside the shop from her first appearance; she never goes outside. Terrified, she closes the door, ducks at breaking glass and flees deeper indoors.
[CROWDS]
Fictional adults. Across separate areas, zombies bite and grab survivors. Survivors pull companions free, evade attacks and flee. Normal skin and eyes distinguish survivors from gray-faced, white-eyed zombies. Victims clutch bitten body parts; red liquid remains on attackers' mouths and victims' sleeves. No duplicated people or infected characters recovering.
[SETTING]
The same night. The first 15 seconds show Skytree's exterior and observation deck. The next 15 show the Sensoji approach in Asakusa and a small shop beside that route. No subway, taxi or convenience-store scene. Change districts with a clear cut at 15 seconds; nobody travels between them.
Curved observatory glass overlooks dense Tokyo lights and roads below. An open central aisle leads from the windows to an emergency door, with straight descending stairs immediately beyond it. Cut from tower exterior to interior; never fly the camera through glass.
Sensoji's red Hozomon, main-hall roofs and stall-lined approach. The shop is beside the man's escape route. Looking from inside toward the frontage, the sliding glass entrance is on the left, a separate broad display window on the right. Its sill is knee-high, allowing a whole body through a large opening. The pane is intact through 23 seconds. Clerk and man are inside; the black-jacketed zombie stands on the pavement outside. Throughout 23–30, the camera remains inside, looking out through that window. The closed left door and breaking right window never exchange positions. A fictional disaster, not a real event.
[PROPS]
Rope-linked metal stanchions along the observatory aisle edges. No chairs, benches or wheeled furniture in the chase and rescue area. Collapsed stalls, fallen merchandise and overturned boxes line the Sensoji route; one canopy and its merchandise burn. Broken glass remains on the sill and indoor floor.
[LIGHTING]
Observatory ceiling lights and general indoor lamps are off. Keep the room deeply dark so Tokyo's night skyline stays clearly visible through the glass. Only weak city glow and low blue emergency guidance lights catch facial edges and the floor. Faces remain mostly shadowed; gray skin and white eyes appear where faint light reaches them. No brightly lit white hall or bright facial fill. Preserve this darkness in every connected observatory shot.
Black sky, white-and-pale-lavender Skytree and dense city lights. Sensoji is lit by red temple lamps and actual fire; the shop by its indoor lamps. Never transfer the shop's brightness to the observatory.
[TEXTURE AND LIVE-ACTION LOOK]
Japanese zombie-disaster realism. Sweat, disheveled hair, dirty rumpled clothes, skin texture and practical makeup. Fine film grain and slight light bloom. Preserve gray faces, white eyes and red liquid on mouths and bitten sleeves, without exposed wound interiors or excessive gore. No smooth CGI skin, global blur or toy-like shards.
[CAMERA AND COMPOSITION]
Vary height, distance and purpose: elevated skyline, low pursuit, face and held-arm proximity, escape viewed from lower on the stair flight, diagonal crowd fighting. Foreground rescue, middle-distance bites and distant escapes unfold at different times. Nearby people dodge immediately; distant people react later. Runners lean forward, pump their arms and push quickly off the floor.
The final 23–30 seconds are one continuous shot on the same 35mm lens. From inside, include the window frame, attacker outdoors and man indoors. Only after falling glass reveals an actual opening does the camera lower and retreat deeper inside. The attacker moves street-to-room; survivors move farther into the room. No exterior camera position or reverse angle that flips this direction.
[DIALOGUE AND SOUND]
No background music, no BGM, no musical score. No music track, instrumental accompaniment or rhythmic music bed. Dialogue, sound effects, and essential location sounds only. At 0–15 seconds, nobody in the observatory speaks. They communicate by gestures, looks and gripping an arm; only wordless breaths and screams. The only intelligible speech is the single line assigned to the ochre-jacketed man at 20–23 seconds, spoken once. Lip-sync belongs only to him. The clerk shuts the door without speaking. Others only scream or breathe without words; zombies only groan or roar. No improvised Japanese, meaningless syllables, clerk's reply, extra conversation, narration or announcements.
Synchronize running feet, metal stanchion impacts, bite victims' screams, falling stalls, flames, glass impact, cracking and falling shards. At the breach, a brief heavy impact and zombie roar erupt, then decay quickly. Do not mute location sounds.
[CAUTIONS]
No subtitles, title cards, watermarks, recording UI, filming equipment or operator's body. Preserve faces, clothes, normal/infected identities, red liquid and debris across connected shots. Never infect the observatory friend or shop clerk.
Zombies run by pushing off the floor. No chair-riding or sliding atop furniture. The final attacker is the black-jacketed zombie outside from the start. His whole body stays outside until glass falls away; then he enters through that same opening. Never replace this with an indoor zombie breaking the window and exiting. The clerk retreats farther indoors throughout. Show the connected breakage and attack over seven seconds, ending mid-attack.
(0:00–0:03) — Skytree at night. Track sideways from an elevated diagonal view, showing the white-and-lavender tower, both observation levels and dense Tokyo lights. Interior screams and metal impacts enter just before a clear cut into the dark observatory.
[CAMERA: 24mm wide / diagonal high-angle / lateral tracking]
(0:03–0:07) — Dark observatory. The blue-jacketed woman runs between stanchions; the gray-faced zombie follows, his shoulder toppling a post. Track low and sideways, revealing her fallen gray-hooded friend and a bite-and-rescue struggle by the windows. Tokyo lights remain clear behind shadowed people. Wordless screams and breath.
[CAMERA: 28mm wide / waist-height low-angle / forward tracking]
(0:07–0:11) — She pulls her friend's arm; he plants his feet and stands. They release hands and run together toward the emergency door. The zombie pursues along the same floor aisle. Neither speaks. Track sideways from faces and the held arm toward their escape. No furniture interaction.
[CAMERA: 50mm medium / eye-level / lateral tracking]
(0:11–0:15) — The gray-hooded man opens the door, letting her cross before following downstairs. From lower steps, look up and retreat down the flight. Survivors descend in the foreground; behind them, the zombie runs toward the open doorway. Distant crowds scatter elsewhere. Only hurried steps and breath.
[CAMERA: 35mm medium wide / low-angle / backward tracking]
(0:15–0:20) — Before red Hozomon, zombies grab survivors; companions pull them free. An arm bite beside fallen stalls leaves red liquid on sleeve and attacker's mouth. The black-jacketed zombie chases the ochre-jacketed man around burning fabric and boxes toward the shop. Track diagonally across fights, rescues and escapes. Wordless screams.
[CAMERA: 24mm wide / diagonal high-angle / rightward tracking]
(0:20–0:23) — From inside the shop, watch the left entrance. The ochre-jacketed man fully enters, shouting “閉めて！” once to the indoor clerk. She shuts it without replying. On the pavement outside, the black-jacketed zombie runs to the right display window and stops. Both survivors retreat indoors.
[CAMERA: 35mm medium / eye-level / interior backward tracking]
(0:23–0:26) — Start the continuous indoor shot facing the right window: man foreground left, clerk deeper inside, zombie facing inward from the pavement. He braces on pavement and strikes glass with shoulder and forearms. Cracks spread; glass falls inward, opening a large hole. Until then, his body stays outside. Survivors retreat from shards.
[CAMERA: 35mm medium wide / eye-level / locked-off indoors]
(0:26–0:30) — Without cutting, the same outside zombie brings torso and legs through the opening, enters the room and lunges onto the ochre-jacketed man. He recoils deeper indoors; the clerk flees farther inside. Broken frame behind the attacker. Lower and retreat indoors, revealing gray face and impact. Hard cut mid-attack. Screams and roars only.
[CAMERA: same 35mm medium wide / waist-height low-angle / interior backward tracking]

Why it works

  • Speech is rationed to one line, '閉めて！' from one man at 20–23 seconds, and lip-sync is given to him alone.
  • The shop is mapped from inside (door on the left, window on the right) and the last seven seconds never leave that view.
  • It uses real landmarks, Skytree and Sensoji, and labels itself a fictional disaster; swap the locations if you would rather not.

Time Freeze4

The world stops mid-motion and only the hero keeps moving through it.

Made with Seedance 2.015s

by @miilesus116.1K viewsOriginal post

Finger-snap time freeze in a celebrating sports bar

A woman snaps her fingers in a packed bar, a shockwave freezes beer and popcorn mid-air, she walks through and plucks a kernel, then snaps again and the celebration carries on.

  • Image input
  • Snap freeze
  • Seedance 2.0
  • 15 seconds

Inputs @ Reference Image — the main character

Video prompt2,276 chars · 5 beats
Use @ Reference Image as the main character, keeping facial features and body proportions consistent throughout. She is a 30-year-old woman. Cinematic time-freeze short film, 15 seconds, ultra-realistic, shot on Arri Alexa Mini, 35mm lens, moody sports bar interior lighting with neon accents, volumetric haze, dynamic hard shadows, shallow depth of field.
[0:00–0:03]
A packed, dimly lit sports bar, time flowing normally. Steadicam frontal medium shot tracking her walking confidently through absolute chaos. On the blurred TVs in the background, the Yankees have just clinched the World Series. The crowd is erupting in pure euphoria. As she walks, she raises her right hand and snaps.
[0:03–0:06]
At the snap, a subtle spherical shockwave bursts from her fingertips with air distortion and light refraction, expanding outward. Everything freezes mid-explosion. Thick arcs of golden beer suspend in the air, with individual glistening droplets visible. Popcorn kernels float motionless. People are frozen mid-cheer and mid-air in leaping embraces. Volumetric neon light catches the suspended dust and liquid. Absolute silence.
[0:06–0:09]
Only she moves. Her footsteps echo slightly. The camera moves backward as she walks through the frozen celebration, observing calmly. She casually ducks under an arc of frozen, splashing beer. She notices a cluster of floating popcorn, reaches out, and plucks a single kernel from the air.
[0:09–0:11]
She stops in front of a frozen woman wearing a Yankees jersey, her face locked in an ecstatic, tearful scream, arms raised for a high-five. The main character tilts her head, gently adjusts the brim of the frozen woman’s Yankees cap, nods slightly, and softly says: “perfect.”
[0:11–0:15]
She turns, tosses the popcorn kernel into her mouth, and crunches it. She faces the camera, smirks, and snaps again. A stronger reverse shockwave restores motion. The beer completes its splash, the cheering roar instantly returns, and fans land their jumps. She walks away as the camera pushes through the wildly celebrating crowd. Fade to black.
Sound: deafening bar roar/celebration → snap → deep shockwave/bass drop → absolute silence → footsteps → sharp popcorn crunch → “perfect” → snap → reverse shockwave → deafening celebration returns

Why it works

  • Five timestamped blocks each hold one action: snap, freeze, walk, a small interaction, unfreeze.
  • Sound is written as one arrow chain from the bar roar to silence and back, which lines the audio up with the beats.
  • It names a real team and event, the Yankees and the World Series; change them if you would rather avoid them.
Made with Seedance 2.520s

by @Mayz1169100.9K viewsOriginal post

Anime time freeze mid-fall in Shibuya, one continuous shot

An anime girl trips on a curb, time stops before she lands, and the same camera flies a set route around her, her console and her spilled photos before returning to its handheld spot, where time starts again.

  • Image input
  • Time freeze
  • One shot
  • Anime

Inputs Character reference sheet (the heroine)

Video prompt11,851 chars · 4 beats
Create an approximately 20-second horizontal 16:9 anime video as ONE CONTINUOUS UNBROKEN SHOT. No cuts, transitions, hidden edits, or montage.
CHARACTER LOCK
Use the supplied character reference sheet as the sole visual authority for the heroine. Every view on the sheet depicts the SAME character. Do not reproduce the reference sheet’s layout, labels, borders, or typography.
Preserve her exact facial identity, body proportions, hairstyle, clothing construction, accessories, and color placement throughout the video.
She is an energetic anime streetwear girl with vivid pink hair gathered into the short high ponytail shown in the reference, layered pink bangs, cyan-blue face-framing tips, and small star-shaped hair clips.
She wears the exact colorful over-ear headphones with cat-ear attachments and circular smiley-face earcups shown in the reference. The cat ears are part of her headphones, not biological ears. Keep the headphones securely on her head throughout.
Preserve her glossy cropped yellow jacket with oversized sleeves, pale holographic hood lining, iridescent panels, zippers, and decorative patches. Underneath, she wears the reference black cropped top with a pixel-heart graphic, black fitted shorts, decorated belt, and pink fingerless gloves.
Preserve her full-length cyan-to-lilac-and-pink gradient tights with pixel patterns and the exact chunky white sneakers with yellow, pink, and purple detailing and sculpted purple soles.
Keep the small black backpack with its pink pixel-alien motif, dangling star and heart charms, and the round smiley-face hip pouch exactly as shown. Preserve their attachment points and scale.
Do not replace her shorts and tights with a skirt. Do not add a visor, magical-girl costume, wand, or flowing ribbons.
STORY PROPS
The only added main prop is a compact retro handheld game console, small enough to fit in one hand, with a lavender-and-cyan shell, physical buttons, and a short wrist strap. Its screen shows simple abstract pixel shapes without readable text.
A thin pastel headphone cable connects one earcup to the handheld console. It hangs with natural slack and remains physically connected throughout the action.
Her round hip pouch contains a few small loose star-shaped acrylic trinkets and four or five instant-photo prints. These are separate from the decorative charms permanently attached to her outfit and backpack.
VISUAL STYLE
Premium Japanese anime with expressive linework, polished cel shading, fluid character animation, and the detailed fashion-illustration aesthetic of the reference.
Emphasize the contrast between her bright yellow jacket, pink hair, black clothing, and cyan-lilac accessories. Render the jacket’s glossy and iridescent surfaces with clean anime highlights.
No photorealism, live-action human, or 3D game-character rendering.
LOCATION
A crowded Shibuya street in Tokyo during a bright summer late afternoon.
Dense pedestrian traffic, storefronts, vending machines, traffic lights, taxis, bicycles, and layered urban architecture establish a lively city.
Warm sunlight catches her pink hair, yellow jacket, holographic trim, and chunky sneakers. Her yellow silhouette remains easy to track against the crowd.
OPENING CAMERA — Y2K ANIME HANDYCAM
The camera feels like an early-2000s MiniDV camcorder operated by a friend running beside her.
Use irregular handheld shake, imperfect framing, slight exposure shifts, subtle highlight bloom, mild chromatic fringing, and gentle tape-era softness. Keep the anime artwork and facial expressions readable.
The opening is spontaneous and energetic, with no polished stabilization.
0–6 SECONDS — RUNNING THROUGH SHIBUYA
The unseen camera operator runs alongside the girl, struggling to keep her between waist-up and full-body framing.
She hurries through the packed sidewalk while holding the small handheld console in one hand. Its wrist strap is not secured around her wrist. The headphone cable hangs loosely between the console and her headphones.
Her ponytail bounces, blue-tipped side strands flick across her cheeks, and oversized yellow sleeves shift with each stride. The backpack and attached charms bounce naturally.
She slips between pedestrians, darts around a bicycle, and sidesteps someone carrying shopping bags. One pedestrian briefly crosses part of the foreground without fully concealing the scene.
Clearly late but still playful, she glances back toward the camera, laughs breathlessly, and calls:
“Hurry! We’re gonna be late!”
She turns forward and speeds up.
6–8 SECONDS — THE ACCIDENT
The camera moves slightly ahead into a front three-quarter angle while maintaining the running handheld motion.
She briefly glances down at the handheld console.
Distracted, she fails to notice a low curb. The thick front sole of her chunky sneaker catches its edge.
Her confident expression instantly becomes wide-eyed shock.
She pitches forward and instinctively reaches out to break her fall. The console slips from her hand and travels a short distance ahead of her, within the slack length of its connected headphone cable.
The movement flips open her loosely fastened hip pouch. Several loose star-shaped trinkets and a few instant-photo prints spill outward.
Her jacket sleeves billow, her ponytail swings forward, and the headphone cable forms a broad curved loop beside her outstretched arm. Her headphones stay firmly in place.
At the exact instant before she reaches the pavement:
TIME STOPS COMPLETELY.
She hangs halfway through the fall, both arms extended, one knee bent beneath her and the other leg trailing behind.
Her startled face, ponytail, jacket folds, backpack charms, and shoelaces are perfectly motionless.
The console hangs just ahead of her hand. The connected headphone cable is suspended in its curved shape. Star trinkets and photographs remain scattered through the surrounding air.
Every pedestrian, bicycle, and vehicle stops at the same instant.
Only the camera moves.
8–17 SECONDS — CONTINUOUS FROZEN-TIME FLIGHT
There is NO transition and NO replacement camera.
From its existing position, the same viewpoint continues moving. Human handheld instability gradually evolves into fast, precise FPV-style flight.
Follow one coherent, continuous route through the frozen scene:
Rise above the girl just far enough to reveal the frozen Shibuya crossing. Pedestrians remain locked mid-step, a taxi is stationary in the intersection, and shopping bags hang motionless.
Immediately descend toward her face. Pass close to her wide eyes, pink bangs, cyan-blue side strands, star hair clips, and cat-ear headphones.
Curve around the outside of one smiley-face earcup, revealing its layered colorful construction. Do not pass through her hair, headphones, or body.
Sweep past her shoulder and along the outside of the yellow sleeve. Show the glossy jacket folds, holographic trim, pink fingerless glove, and outstretched fingertips through changing camera perspective.
Continue briefly past the backpack’s pixel-alien motif and its suspended charms, then bank toward the floating handheld console.
Arc closely around the console to reveal its physical buttons, pastel casing, short wrist strap, and abstract pixel display. The console and screen image remain completely frozen.
Thread through the clear opening formed by the broad curve of the headphone cable, staying safely between the cable and her arm. The cable does not shift or widen for the camera.
Fly between the suspended star trinkets and instant-photo prints, creating strong foreground parallax.
The photos show casual anime snapshots of the same girl at arcades, hanging out with friends, and exploring colorful city streets. No readable captions, logos, or text appear on the photos.
Bank around the final photograph to reveal her full frozen pose again.
Keep the camera path spatially coherent and the important details readable. Do not add extra maneuvers that require teleportation or impossible object intersections.
17–20 SECONDS — RETURN AND TIME RESUMES
Make one fast sweeping arc around the heroine, showing the complete frozen arrangement: her falling pose, bright yellow jacket, suspended console, curved cable, star trinkets, and photographs against the motionless crowd.
Descend to street level and physically return to approximately the same imperfect handheld angle and height used before the freeze.
As soon as the camera reaches that position:
TIME RESUMES.
No flash, transition, glitch, or cut.
Every person and object continues from its exact frozen position.
The girl catches herself with one gloved palm against the pavement and lands awkwardly on one knee without injury.
The console lands beside her, remaining within the slack length of the headphone cable. Her headphones stay on. The star trinkets bounce, and the photographs flutter onto the pavement.
Her ponytail, jacket, and backpack charms settle naturally.
The pedestrians immediately continue walking as though nothing unusual happened.
At the same moment, the precise camera movement becomes shaky human handheld footage again. Her friend instinctively lowers the camera toward her.
Still crouching, she quickly scoops up the console and looks toward the lens. Embarrassment flashes across her face, then turns into a mock-annoyed, playful glare.
She holds up the console and says:
“You saw nothing!”
She immediately reaches for a scattered photograph.
End while the handheld camera is still shaking.
CAMERA CONTINUITY
One viewpoint, one uninterrupted shot, and one continuous spatial path.
No camera teleportation, hidden cuts, lens jumps, full-frame wipes, or transitions concealed by passing objects.
Unrestricted camera flight begins only after time stops. Return physically to the street-level filming position before resuming handheld movement.
FROZEN-TIME RULE
During frozen time, absolutely nothing except the camera may move.
No blinking, breathing, hair movement, fabric flutter, rotating charms, drifting photographs, cable motion, animated console screen, pedestrians, or vehicles.
Use true three-dimensional parallax around stationary subjects. Do not simulate this sequence with a flat freeze-frame zoom.
View-dependent reflections on the glossy jacket may change naturally as the camera moves, but the lighting, clothing geometry, and all objects remain fixed.
CHARACTER AND PROP CONTINUITY
Maintain the exact reference face, pink ponytail, cyan-blue side strands, star clips, cat-ear headphones, yellow jacket, black pixel-heart top, black shorts, gradient pixel tights, pink gloves, sneakers, backpack, and smiley hip pouch.
No duplicate character, extra limbs, wardrobe changes, disappearing accessories, or newly appearing objects.
The headphone cable has one consistent length and remains attached to the same earcup and console throughout. It never clips through her body.
The freeze is an unexplained cinematic event, not a power she activates. No transformation, magical effects, or futuristic interface.
AUDIO
Opening: busy street ambience, running footsteps, breathless dialogue, and light rattling from her accessories.
Frozen time: street sound abruptly drops away, with restrained flight sound emphasizing the moving viewpoint.
Time resumes: the city ambience returns instantly, together with the landing, scattered objects, and her final line.
OVERALL FEEL
Playful Y2K streetwear energy with a spectacular frozen-time detour and an embarrassed comic payoff.
Her yellow jacket, pink hair, cat-ear headphones, and pixel accessories remain the visual anchors throughout the dense Tokyo setting.
No slow motion before the freeze. No montage. No black frames. No magical-girl wand, ribbons, or tennis equipment. No photorealism. No subtitles, promotional overlays, reference-sheet typography, logos, or watermarks.

Why it works

  • The frozen-time rule is total: no blinking, breathing, hair or cable movement, and only the camera moves.
  • Time resumes when the camera gets back to its pre-freeze height and angle, which joins the two halves without a cut.
  • The headphone cable has a fixed length and forms a loop the camera threads through without the cable moving.
Prompt source
Made with Seedance 2.517s

by @Mayz116950.3K viewsOriginal post

Anime time freeze, second version: heart wand and ribbons

The same 20-second freeze with a heart-themed heroine: she trips in Shibuya, the camera flies through the open center of her heart wand and through a ribbon loop, and she lands when time resumes.

  • Image input
  • Time freeze
  • Remix
  • Anime

Inputs Character reference sheet (the heroine and her wand)

Video prompt10,299 chars · 4 beats
Create an approximately 20-second horizontal 16:9 anime video as ONE CONTINUOUS UNBROKEN SHOT. No cuts, transitions, hidden edits, or montage.
CHARACTER LOCK
Use the supplied character reference sheet as the sole visual authority for the heroine. All views on the sheet depict ONE character, not multiple people. Use the sheet only as a design reference; do not reproduce its layout, borders, labels, or typography.
Preserve her exact facial identity, body proportions, silhouette, hairstyle, outfit, accessories, and prop designs throughout.
She is a cheerful, energetic heart-themed anime heroine with short pink hair, blue-tipped side strands, large golden-amber eyes, a white-and-pink visor with a heart emblem, and the distinctive oversized bow and decorative hair accessories shown in the reference. Her hair must remain short: no long ponytail or new hairstyle.
Preserve her exact pink-and-white ruffled cropped top, chest bow and heart ornament, layered pink-and-white short skirt, blue accent details, decorative arm accessories, asymmetric leg accessories, and chunky heart-decorated platform sneakers. Preserve the heart-shaped accessory on her back and the long pink-and-blue ribbon-like elements shown in the reference.
Use the reference heart-shaped wand and its ribbon-like cord as her main handheld prop. Preserve its recognizable heart outline, handle, colors, and proportions. It is not a tennis racquet.
The only added accessory is a small white-and-pink zippered pouch attached at her hip, visually secondary to her existing outfit. It contains a few heart-shaped trinkets and small instant-photo prints.
VISUAL STYLE
Premium Japanese anime with clean expressive linework, polished cel shading, lively facial animation, and fluid movement. Match the reference illustration’s proportions and visual aesthetic. Saturated pink, warm white, magenta, and small cyan-blue accents define the heroine.
No photorealism, live-action rendering, or 3D game-character appearance.
LOCATION
A crowded Shibuya street in Tokyo on a bright summer late afternoon. Pedestrians fill the sidewalks and crossing. Storefronts, vending machines, traffic lights, taxis, buses, and bicycles create a dense urban background.
Warm sunlight catches her pink hair, glossy heart ornaments, layered skirt, and colorful ribbons. Her pink-and-white silhouette remains clearly readable against the busy city.
OPENING CAMERA — Y2K ANIME HANDYCAM
The footage feels as if a friend is running beside her with an early-2000s MiniDV camcorder, translated into anime.
Use irregular handheld shake, imperfect framing, slight exposure changes, subtle highlight bloom, mild chromatic fringing, and gentle tape-era softness. Keep faces and character details readable. The opening feels spontaneous and physically human, not like a stabilized commercial.
0–6 SECONDS — RUNNING THROUGH SHIBUYA
The unseen camera operator runs beside the heroine, struggling to keep her between waist-up and full-body framing.
She hurries through the crowded sidewalk, holding her heart-shaped wand in one hand and a small pink heart charm in the other. Her hip pouch bounces with her steps. Her short pink hair, blue-tipped strands, bows, and long ribbon elements react naturally to her movement.
She slips between pedestrians, dodges a bicycle, and steps around someone carrying shopping bags. A passing pedestrian briefly obstructs part of the frame without concealing a cut.
She is clearly late, but still playful and full of energy. She glances back toward her friend while running, laughs breathlessly, and calls:
“Hurry! We’re gonna be late!”
She faces forward again and accelerates.
6–8 SECONDS — THE ACCIDENT
The camera moves slightly ahead into a front three-quarter angle, still shaking as the operator runs backward.
The heroine briefly tosses the small heart charm a few centimeters above her palm. Distracted, she fails to notice a low curb.
The thick sole of her platform sneaker catches the edge.
Her expression snaps from cheerful confidence to wide-eyed shock. Her body pitches forward. The heart-shaped wand slips from her fingers, and the small heart charm flies toward the foreground.
The sudden movement flips open her loosely fastened hip pouch. A few heart-shaped trinkets and four or five instant-photo prints spill into the air. Her long ribbon elements sweep into broad open curves around her.
At the exact instant before she reaches the pavement:
TIME STOPS COMPLETELY.
She hangs halfway through the fall, both arms reaching forward, one knee bent and the opposite leg extended behind her. Her startled expression is perfectly frozen.
The wand hangs nearby, its heart-shaped opening clearly visible. The small heart charm remains close to the camera. The scattered trinkets, photographs, ribbon loops, hair strands, and clothing are locked in midair.
Every pedestrian and vehicle stops at the same instant.
Only the camera continues moving.
8–17 SECONDS — CONTINUOUS FROZEN-TIME FLIGHT
There is NO transition and NO replacement camera.
The same viewpoint continues from its existing position. Its handheld instability smoothly evolves into a fast, precise FPV-style flight through the frozen three-dimensional scene.
Follow one coherent, uninterrupted route:
First, rise just above the heroine and reveal the frozen crossing around her: pedestrians suspended mid-step, a taxi stopped in the intersection, and shopping bags hanging motionless.
Immediately curve back down toward her face. Pass close enough to see her wide golden-amber eyes, pink fringe, blue-tipped side strands, and the heart emblem on her visor. Move around the outside of her hair and bow without intersecting them.
Continue past her shoulder and outstretched hand, revealing the layered outfit, heart ornaments, and a platform sneaker hanging just above the pavement. Keep the framing action-focused and never look beneath her skirt.
Bank toward the suspended heart-shaped wand. Briefly arc around its glossy pink outline, then fly through the clearly open center of the heart without touching the frame.
Continue through one large, stationary loop formed by her pink-and-blue ribbon. The ribbon does not move or open to accommodate the camera.
Pass the foreground heart charm and fly between the scattered trinkets and instant-photo prints, creating strong depth and foreground parallax.
The photographs show small casual anime snapshots of the same heroine with friends, colorful city outings, summer skies, and playful everyday moments. They have plain borders with no captions or readable text.
Bank around the edge of the final photograph to reveal the heroine’s entire frozen pose again.
Keep this flight quick but visually legible. Maintain stable geography and prioritize a continuous camera path over adding extra maneuvers.
17–20 SECONDS — RETURN, TIME RESUMES, COMIC PAYOFF
Make one fast sweeping arc around the frozen heroine, briefly showing the whole suspended arrangement: her startled pose, floating wand, heart charm, scattered photographs, and curved ribbon elements, surrounded by the motionless Shibuya crowd.
Descend to street level and physically return to approximately the same handheld side angle and camera height used before the freeze.
The moment the camera reaches that position:
TIME RESUMES.
No flash, transition, pulse, or cut.
Every person and object continues from its exact suspended position.
The heroine completes the fall but catches herself with one palm against the pavement, landing awkwardly on one knee without injury.
The wand clatters beside her. The heart charm and small trinkets bounce across the ground. The photographs flutter down. Her ribbons, short hair, and skirt settle naturally.
Pedestrians resume walking as though nothing unusual happened.
At the same instant, the precise camera movement returns to imperfect human handheld shake. Her friend instinctively lowers the camera toward her.
Still crouched, she quickly picks up the nearest heart charm and looks up at the camera. Her embarrassment lasts only a beat before she adopts a mock-stern, playfully defensive expression.
She holds the charm toward her friend and says:
“You saw nothing!”
She immediately reaches for her fallen wand.
End while the handheld camera is still shaking.
CAMERA CONTINUITY
One viewpoint, one uninterrupted spatial path, one continuous shot. No teleportation, lens jumps, hidden wipes, or full-frame obstructions used to disguise edits.
The camera becomes freely mobile only during frozen time, then returns to the original street-level filming position before handheld movement resumes.
FROZEN-TIME PHYSICS
During the frozen interval, absolutely nothing except the camera moves. No blinking, breathing, drifting hair, fluttering ribbons, rotating props, falling photographs, moving pedestrians, or vehicles.
Maintain true three-dimensional parallax around stationary subjects. This must not look like a flat freeze-frame image being zoomed or warped.
The freeze is an unexplained cinematic event, not a spell cast by the heroine. No transformation sequence, magical blast, or new costume.
CHARACTER AND PROP CONTINUITY
Keep the same face, short pink hair with blue tips, visor, bows, heart accessories, pink-and-white outfit, platform sneakers, and reference wand in every frame.
All props must be present before the freeze and remain in consistent positions during it. No duplicated heroine, extra limbs, newly appearing objects, or changing ribbon attachments.
AUDIO
Opening: lively street ambience, running footsteps, breathless dialogue, and light accessory rattles.
Frozen time: street ambience abruptly drops away; restrained camera-flight sound emphasizes the movement.
Time resumes: city sound returns immediately with the landing, prop impacts, and final dialogue.
OVERALL FEEL
Bright, chaotic, playful summer anime energy. A heart-themed pink heroine rushing through Shibuya becomes the center of an impossible frozen-time camera flight, then returns to an embarrassing everyday mishap.
No slow motion before the freeze. No cuts. No black frames. No outfit changes. No tennis equipment. No orange-haired replacement character. No smartphone interface. No subtitles, promotional overlays, reference-sheet text, logos, or watermarks.

Why it works

  • It keeps the first version's structure and swaps only the character and props, a quick way to remix a prompt that already works.
  • The props are picked to be flown through, the heart-shaped wand and a ribbon loop, and neither may move for the camera.
  • The negative list names specific wrong turns: no tennis equipment, no orange-haired replacement character, no smartphone interface.
Prompt source
Made with Seedance 2.015s

by @rovvmut_43.5K viewsOriginal post

Finger-snap time freeze on a city sidewalk

One paragraph: a man walks toward the camera, snaps to freeze pedestrians and pigeons, takes a sip of a frozen woman's coffee, puts it back and snaps the city back into motion.

  • Image input
  • Snap freeze
  • One paragraph
  • Seedance 2.0

Inputs @ Image1 — the lead character

Video prompt2,392 chars
@ Image1 is the lead character - maintain identical facial features, proportions, and identity throughout. He is wearing round clear glasses, a black hoodie and brown cargo pants. Create a 15-second ultra-photorealistic cinematic time-freeze sequence with an ARRI Alexa Mini look, 50mm lens, natural daylight, strong shadows, shallow depth of field, and subtle film grain. The scene opens on a busy city sidewalk in full motion, captured in a front-facing steadicam shot tracking backward as he walks calmly toward camera through a natural flow of pedestrians while pigeons cross the bright sky above; without breaking stride, he casually raises his hand and snaps his fingers. Instantly, a bright spherical shockwave erupts from his hand, with visible air distortion and refracted light, expanding rapidly across the environment, freezing dust and debris mid-air, locking pedestrians in place mid-motion, and suspending pigeons mid-flight as reflections ripple across surrounding glass buildings before everything falls into complete stillness and silence, with fine particles hanging in the sunlight. The camera continues gliding backward as he walks through the frozen world, his footsteps the only sound, subtly observing his surroundings before approaching a frozen woman with striking red hair in a flowing green dress, her hair suspended mid-motion as she holds a takeaway coffee. He steps in, briefly studies the moment, removes the cup from her hand, takes a síp, then carefully places it back exactly as it was. He then turns his attention to the camera, raises his hand again, and snaps his fingers, releasing a more powerful shockwave that rapidly spreads outward, reversing the effect as motion instantly resumés - pedestrians continue walking, pigeons scatter into flight, the woman carries on unaware still holding her coffee, and debris drops naturally to the ground as full city sound returns. He pauses for a brief moment, then continues walking forward, gradually blending into the crowd while the camera keeps tracking backward, maintaining distance as he disappears into the flow, ending in a smooth fade out. Sound design follows the sequence: city ambience, sharp snap, deep expanding shockwave rumble, complete silence, isolated footsteps, subtle ambient echo, sip, second snap, stronger reverse shockwave, city noise returning, brief pause, then footsteps continuing.

Why it works

  • The camera is one backward-tracking steadicam from start to end, so there is nothing to cut.
  • The coffee goes back 'exactly as it was', which is why the woman carries on unaware after the unfreeze.
  • A sound-design line at the end lists every audio beat in order, from city ambience to the footsteps after the second snap.
Prompt source

Zombie Match Cut4

She aims the gun, can't pull the trigger, and the embrace hard-cuts into the couple's happy past.

Made with Seedance 2.560s

by @ShamiWeb3486K viewsOriginal post

Zombie girlfriend fixes his collar, then bites: a 60-second two-parter

In a Beijing apartment at night, Su Yi, now a zombie, straightens Chen Hao's collar, the scene dissolves into a tender memory, and a kiss on his neck dissolves back into the bite before his one deadpan line.

  • Text to video
  • 60 seconds, 2 parts
  • Dissolve
  • Dark comedy
Video prompt16,326 chars
"THE LAST GESTURE" — 60s Zombie Horror/Dark-Comedy Prompt (16:9)
Context: Photorealistic zombie horror drama with an abrupt deadpan dark-comedy payoff. Shot on ARRI Alexa 35 with spherical prime lenses. A compact modern apartment in Beijing at night, recently lived-in and freshly disturbed — displaced furniture, one overturned chair, scattered small objects, a messy table, curtains pulled out of place; the architecture stays intact and contemporary. Mood: an ordinary home turned hostile, where one familiar gesture briefly reopens something the room has already lost. Visual grade: low-key cinematic underexposure, cold 5600K window spill against a single 3000K practical lamp, natural skin texture, natural motion blur, restrained depth of field, subtle fine 35mm film grain. The grade stays fixed across every shot.
The apartment exists in two timeline states. PRESENT state: night, cold blue-grey spill from the frame-left window, one dim warm lamp making small amber pockets, furniture disturbed. MEMORY state: the identical apartment before the outbreak — furniture restored, surfaces tidy, soft even 3000–3500K domestic practical light wrapping the faces, no horror shadows, skin alive and healthy. Room geometry, furniture placement and camera angle stay identical between the two states; only lighting, tidiness and mood change.
Camera language: PRESENT state is nervous handheld throughout — visible operator shake, reactive reframing, small lateral corrections, short push-pulls, slight roll during physical struggle, unstable breathing drift. MEMORY state is the deliberate opposite — smooth, quiet, stable, almost no shake. Every shot below declares which mode it uses.
Location this video: the living room, with a small dining area, kitchenette and hallway mouth carrying background depth.
Characters
Chen Hao (male lead) — Chinese, late 20s, lean-athletic build, angular face with a strong jaw, dark brown eyes, short black hair with natural volume and a slight fringe, warm light-tan skin, naturally attractive in an unpolished, boy-next-door way. Wardrobe: oversized washed charcoal t-shirt, lightweight dark grey open overshirt, relaxed dark jogger pants, dark low-top canvas sneakers. Chen Hao is the only man in the video.
Su Yi (female lead) — Chinese, mid-20s, strikingly beautiful, tall and slim with a model's proportions and posture, long dark slightly wavy hair with soft wispy bangs, thin round metal-frame eyeglasses, warm brown eyes, fair porcelain skin, delicate refined features — the epitome of elegant beauty. Wardrobe (MEMORY state): cream cropped cardigan over a white inner top, charcoal-grey pleated mini skirt, white socks, white shoes. Su Yi exists only inside the MEMORY state — clean, healthy, unharmed.
Continuity lock — glasses: Su Yi wears the same thin round metal-frame eyeglasses in every single shot of the video, PRESENT and MEMORY alike, with zero exceptions. She never appears without them. In MEMORY state the glasses are clean and sit straight; in PRESENT/zombie state the same glasses are dirty and crooked but still on her face. Do not remove or omit the glasses in any shot, including Shots 5 and 6.
Su Yi Zombie — the same face, same glasses (dirty, crooked, but present), same hairstyle base and same body proportions as Su Yi. Wardrobe override: the identical cream cardigan, white inner top and charcoal-grey pleated skirt now torn and distressed, skin pale and unhealthy, dark hollow eye sockets, glasses dirty and crooked, hair messy and matted, blood smears and grime across face, hands and clothing.
Su Yi and Su Yi Zombie are the same woman at two points in time — the same person before and after.
Chen Hao Zombie (late reveal, Shot 13 only) — same face, same short black hair, same build as Chen Hao. Wardrobe unchanged but now torn at the shoulder/collar where he was bitten, skin gone pale and grey, eyes clouded and hollow, a fresh bloody bite wound visible at the base of his neck through the torn overshirt, posture slack and stiff like Su Yi Zombie's.
Spatial anchor: Chen Hao starts in midground-left, backing diagonally toward camera past the overturned chair. Su Yi Zombie comes in from midground-right, out of the direction of the dining area. The kitchenette and the hallway mouth stay visible in deep background behind both. All blocking stays on one side of the room — the camera never crosses to the opposite wall.
Not a destroyed post-apocalyptic ruin — the apartment is recently messy, still intact and modern. Not an outdoor scene. Su Yi and Su Yi Zombie are never in frame at the same time.
Generation note: the 12 shots below are split into two 30-second halves so each can be generated as its own clip and stitched together. Each shot lists two timestamps — "clip time" (seconds within that 30s generation) and "full video" (seconds in the final 60s edit). PART 1 ends on the tender embrace (the emotional high point); PART 2 opens with the kiss and turns into the horror payoff.
PART 1 — "Before" (0:00–0:30) — PRESENT struggle → dissolve into MEMORY tenderness
Shot 1 — 0:00–0:05 (clip time) / 0:00–0:05 (full video)
Camera: PRESENT mode. Handheld medium-wide from inside the living room, backing away with Chen Hao while drifting slightly sideways, 32mm. Shake reacts to Su Yi Zombie's unpredictable movement. Foreground: blurred edge of the overturned chair at frame-left. Midground: Chen Hao backing diagonally toward camera as Su Yi Zombie surges in from the right. Background: kitchenette and hallway mouth in low cold light. Act: Su Yi Zombie lunges at Chen Hao with uncontrolled hungry aggression — mouth open, shoulders jerking, rapid uneven steps, hands clawing forward, matted hair swinging across the crooked glasses. Chen Hao backs away fast, eyes wide, forearms raised defensively, never turning his back. Constrain: Su Yi Zombie reads as frightening from the very first frame — feral, unstable, no lingering human softness yet. Audio: uneven aggressive footsteps, furniture scraping the floor, wet snarling breath, Chen Hao's ragged breathing, clothing friction.
Shot 2 — 0:05–0:10 (clip time) / 0:05–0:10 (full video)
Camera: PRESENT mode. Handheld medium two-shot from slightly below eye level, anchored between Chen Hao at left-midground and Su Yi Zombie at right-midground, 40mm. Small backward and lateral corrections follow the struggle; slight roll when Su Yi Zombie jerks sideways. Act: Su Yi Zombie lunges again. Chen Hao catches her and grips both shoulders firmly, arms fully extended to hold the snapping mouth away from his neck. Su Yi Zombie thrashes violently — torso twisting, mouth snapping forward, head thrown side to side. Chen Hao's jaw clenches, brows pull together, breathing fast, hands tightening — visibly straining. Constrain: Chen Hao's hands stay clamped on both shoulders with arms extended for the whole shot — the mouth never reaches his neck here. Real time speed. Audio: palms slapping onto shoulders, shoe friction on the floor, strained exhales, snarling, fabric stretching.
Shot 3 — 0:10–0:14 (clip time) / 0:10–0:14 (full video)
Camera: PRESENT mode. Handheld chest-up two-shot, slow unstable micro push-in, 50mm. Shake gradually reduces but never fully disappears. Act: Su Yi Zombie's thrashing tapers into small spasms, then heavy breathing, then near stillness; the mouth slowly closes and the vacant eyes lock directly onto Chen Hao. Chen Hao stays tense and confused, hands still clamped down, not trusting the sudden calm. Constrain: the momentum drains progressively and visibly — the stillness stays uncanny and vacant, never a warm human expression. Audio: the chaotic sound suddenly drops away; heavy breathing from both, faint clothing settle, low cold room tone.
Shot 4 — 0:14–0:19 (clip time) / 0:14–0:19 (full video)
Camera: PRESENT mode dissolving into MEMORY mode. Tight handheld two-shot, very slow shaky push-in toward the space between Chen Hao's chest and Su Yi Zombie's rising hand, 50mm. Act: Su Yi Zombie slowly raises one hand toward Chen Hao's neck; he stiffens instantly, eyes tracking the hand. Instead of biting, she grips the edge of the dark grey overshirt collar and with strangely precise small finger movements straightens it and smooths the fabric near his upper chest. Chen Hao's expression shifts from fear into stunned recognition. On the last beat the image cross-dissolves into the MEMORY state while the hand keeps moving. Constrain: the dissolve preserves the exact camera angle, body position, hand placement, eyeline and spacing — only the woman, the lighting and the room's tidiness change, and the handheld shake dissolves into perfectly stable movement across the same beat. Audio: fabric brushing under fingertips, two sets of breathing, the cold room tone slowly warming.
Shot 5 — 0:19–0:24 (clip time) / 0:19–0:24 (full video)
Camera: MEMORY mode. Stable near-locked intimate chest-up two-shot, Chen Hao slightly left of center and Su Yi right of center, tiny gentle inward drift, 50mm. Act: Su Yi calmly finishes straightening the collar, glances down to inspect it, makes one final small correction, then lightly pats Chen Hao's upper chest once. He watches her with relaxed affection. Both exchange tiny genuine smiles. Constrain: zero handheld shake in this shot — the frame stays smooth and quiet. No dialogue. Audio: peaceful domestic room tone, soft fabric movement, one quiet pat.
Shot 6 — 0:24–0:30 (clip time) / 0:24–0:30 (full video)
Camera: MEMORY mode. Stable gentle floating reframe widening to include both shoulders and upper torsos, slowly drifting to a soft three-quarter side angle, 40mm. Act: Su Yi steps forward and wraps both arms naturally around Chen Hao. He immediately returns the embrace, posture going completely relaxed. She settles her head against the side of his neck and shoulder. Dialogue (Su Yi, soft, warm): "You worry too much. I'm right here." Dialogue (Chen Hao, quiet, half-smiling): "I know. Just... don't ever disappear on me." Constrain: realistic arm placement and real body contact — the embrace has weight and clothing compression, and the camera stays smooth with no shake. Audio: soft fabric, clothing compressing, quiet steady breathing, warm room tone.
PART 2 — "After" (0:30–1:00) — the kiss dissolves into the bite → English resolution
Continuity note: PART 2 must open on the exact final frame of PART 1 — same camera angle and distance, same MEMORY-state lighting, Su Yi and Chen Hao mid-embrace with her head resting at his shoulder, her glasses clean and visible, both perfectly still for one held beat before Shot 7 begins. Do not restart, re-establish, or re-frame the scene — treat Part 1's last frame as Part 2's first frame.
Shot 7 — 0:00–0:05 (clip time) / 0:30–0:35 (full video)
Camera: MEMORY mode dissolving into PRESENT mode. Tight three-quarter two-shot from just behind Chen Hao's shoulder giving a clear view of Su Yi's face beside his neck, 50mm, almost no movement. Act: Still embracing, Su Yi gently presses her lips to the side and base of Chen Hao's neck in one soft affectionate kiss, then keeps her face resting there. His eyes soften and half-close, comfortable and safe. Hold this exact pose for one full beat, then the frame cross-dissolves back into the PRESENT state — Su Yi becomes Su Yi Zombie and the warm clean apartment becomes the cold messy night apartment, while the head, mouth, arms and shoulder orientation stay in exactly the same place. Constrain: the contact point and head angle are identical before and after the dissolve — for the first instant after the change the zombie's mouth at his neck still reads as a continuation of the kiss, and his relaxed embrace stays momentarily unchanged. The stable camera dissolves back into nervous handheld across the same beat. Audio: very quiet intimate ambience, one soft kiss; then the warm ambience drains into a cold hollow room tone.
Shot 8 — 0:05–0:09 (clip time) / 0:35–0:39 (full video)
Camera: PRESENT mode. Tight handheld two-shot holding the same angle from behind Chen Hao's shoulder, 50mm, with one sharp jolt punctuating the impact. Act: Su Yi Zombie's mouth opens wider against the exact same neck and shoulder area. One tiny half-beat passes where Chen Hao has not yet realized what changed — then she bites hard into the base of his neck and upper shoulder. His eyes snap wide open. Constrain: the bite stays non-graphic — it sells through fabric pull, shoulder compression and his performance, with minimal blood and no explicit wound detail. Audio: sudden hard bite and fabric impact, one sharp involuntary inhale from Chen Hao.
Shot 9 — 0:09–0:14 (clip time) / 0:39–0:44 (full video)
Camera: PRESENT mode. Tight handheld two-shot, strong reactive shake returns, small reframing follows Chen Hao's recoiling and twisting body, 40mm. One continuous take, no cut away. Act: Su Yi Zombie keeps biting and gnawing at the same shoulder and neck area in repeated hungry motions, arms tightening and pulling him closer while feeding. His whole body jerks, mouth opens involuntarily, brows pull up and inward, gasping and grimacing in real pain while shoving at her shoulders and upper arms. Constrain: she never performs one bite and stops — the feeding is continuous, frantic and animalistic, staying non-graphic through fabric pulling, body jolts and head movement with limited blood. Audio: irregular hungry biting, fabric tearing and pulling, Chen Hao's pained gasps and broken breathing, without excessive wet gore sound.
Shot 10 — 0:14–0:19 (clip time) / 0:44–0:49 (full video)
Camera: PRESENT mode. Handheld tight-medium drifting slightly toward Chen Hao's face while Su Yi Zombie stays attached at foreground-right, 50mm, occasional foreground occlusion from their arms. Act: Chen Hao clenches his jaw, winces hard, tries to twist his shoulder free and pushes against her. She drags herself straight back onto the bite every time he opens even a small gap. His efforts weaken progressively; his expression travels from shock to sharp pain to frustration to the realization that this is completely hopeless. Constrain: Su Yi Zombie stays purely predatory here — the feeding never pauses and she never softens or recognizes him again. Audio: continued biting, scuffling feet, strained pushes, exhausted grunts.
Shot 11 — 0:19–0:24 (clip time) / 0:49–0:54 (full video)
Camera: PRESENT mode. Handheld tight-medium three-quarter profile on Chen Hao, movement settling slightly but still breathing, 50mm. Act: Chen Hao finally stops seriously resisting. His hands slowly loosen and drop, his body goes slack with exhausted resignation, and he stares into empty space ahead — the pain still tight in his face, now mixed with pure disappointment, like a man realizing the beautiful moment he just believed in meant absolutely nothing. Constrain: Su Yi Zombie stays continuously biting and feeding at the shoulder through this entire beat. Audio: steady feeding sounds, Chen Hao's long shaky exhale, low cold room tone.
Shot 12 — 0:24–0:30 (clip time) / 0:54–1:00 (full video)
Camera: PRESENT mode. Handheld tight-medium hold on Chen Hao's face in asymmetric composition, center-left, 50mm. Handheld breathing remains but turns less reactive, an awkward observational hold. Act: Chen Hao gives one last small painful wince as she keeps biting, exhales, and his expression flattens into tired irritated acceptance. Without turning dramatically toward camera and without comedic exaggeration, he mutters the only line of the film. Su Yi Zombie keeps biting through and immediately after the line. Hold one short uncomfortable beat, then hard cut to black. Constrain: Chen Hao's delivery stays completely deadpan — no comedic mugging, no direct stare into the lens, and the line stays clearly audible over the biting. Dialogue (Chen Hao — flat, drained, deadpan, breath-short, tired irritation, zero theatrics): "Tā mā de. Zhēn de yǒu jiāngshī." (Subtitle-free; approximate meaning: "God damn it. There's actually a zombie.") Audio: continuous biting and fabric pulling, then his flat line clearly audible over it. Cut to black.
Negative prompt: No music. No subtitles. No on-screen text or logos. No 3D or CGI animation style. No other people. No explicit gore, open wounds or blood spray. No smooth gimbal movement in any present-timeline shot.

Why it works

  • It is 60 seconds written as two 30-second generations: every shot carries clip time and full-video time, and Part 2 must open on Part 1's last frame.
  • The camera tells the timelines apart, nervous handheld in the present and near-locked in the memory, and each dissolve swaps both at once.
  • One loose end: a zombie look for Chen Hao is written for a 'Shot 13' that the twelve-shot list never reaches.
Prompt source
Made with Seedance 2.517s

by @Ciri_ai27.5K viewsOriginal post

Zombie chase down an alley that blows out into a reed field

Sixteen timed shots: a zombie in a torn suit chases a woman down a striped-shutter alley, a white light blowout carries her into a golden reed field, and there he is human again and catches her in an embrace.

  • Image input
  • 16 shots
  • Five references
  • Light blowout

Inputs <<<image_1>>> — the alley location · <<<image_2>>> — the man as a zombie · <<<image_3>>> — the same man, human · <<<image_4>>> — the woman · <<<image_5>>> — the reed field

Video prompt15,583 chars · 2 beats
One continuous video alley chase flows directly into the golden field reveal via a hard light-blowout transition. Sixteen shots, two locations, one unbroken timeline, screen-left-to-right travel maintained through both acts.
SCENE CONTEXT
A reanimated corpse in a ruined business suit pursues a sprinting woman down a shuttered warehouse service alley. She gains ground, he loses it on a toppled bin. At her breaking point she stops, turns, and faces him — and in a burst of blinding light she bolts past him out of the alley's open end into open golden grassland, and he is revealed as himself again, human, chasing not to catch her but to reach her. He catches up, she turns into him, and the piece ends in a still embrace. The two settings are physically distinct (alley → field); the transition is carried by a hard light-blowout cut, not a dissolve.
ACTIVE REFERENCES
<<<image_1>>> (LOCATION — ALLEY): roller shutter wall of horizontal pastel blue-grey slats crossed by one cobalt-blue stripe and one mustard-yellow stripe, rust bleeding down from bolt seams, a maroon plastic wheelie bin with a hinged lid standing against the shutter, a dark vertical steel support beam and a chipped yellow bollard pole at the right edge, cracked stained concrete underfoot. 100% matches the reference.
<<<image_2>>> (ZOMBIE — MAN, Act 1): adult male of Korean descent, ashen grey-green mottled skin, disheveled dark straight hair dulled with grey dust, grime-stiffened charcoal suit jacket torn at the left shoulder seam over a white dress shirt with rust-brown chest spatter and top buttons torn away, one clouded pale-blue eye and one bloodshot brown eye, deep dried laceration across the forehead above the right brow and a shorter one down the left cheek, cracked lips, jagged yellowed broken teeth, dried blood ringing the mouth and collar. 100% matches the reference.
<<<image_3>>> (HUMAN — MAN, Act 2 reveal): same Korean man, dark straight hair now clean and styled, warm skin tone restored, clean-shaven jaw with light stubble, navy tailored two-piece suit or navy blazer, white dress shirt, black leather derby shoes, silver wristwatch. No wounds, no grime, no blood.
<<<image_4>>> (WOMAN, both acts): Korean woman, dark hair parted center and worn loose over the shoulders, warm fair skin, cream ribbed knit zip-up cardigan over a white top, khaki/tan straight-leg trousers, off-white low-top sneakers. 100% matches the reference in Act 1; same wardrobe carries into Act 2.
<<<image_5>>> (LOCATION — FIELD): wide field of tall golden reeds swaying in the wind, low sun glowing near the horizon with soft warm lens flare, hazy mountain silhouette in the background. 100% matches the reference.
LOCATION MAP
Act 1: a single continuous alley running screen-left to screen-right, striped roller shutter as the full backing wall, maroon wheelie bin two-thirds down the run, steel beam and yellow bollard marking the open far end. Act 2: the alley's open end gives directly onto the reed field — she exits the alley mouth and is already running through grass on the far side of the light-blowout cut. Direction of travel stays screen-left to screen-right across both acts without exception.
FORMAT MODE
Sixteen shots across 30 seconds. Hard cuts unless a transition is specified. Shot 12 carries the one authored transition effect (light blowout) that bridges the two locations. Every cut is authored below — the camera does not cut on its own.
OPTICS
Anamorphic spherical hybrid, T2.3 equivalent, for the alley act. Field act shifts to a softer spherical look with warm flare artifacts. Focal lengths vary by shot as specified. Shallow depth of field on close and medium work throughout; deep focus only on the long-lens and wide shots as noted.
SHOT TIMELINE
SHOT 1 (00:00.0–00:01.6) — GROUND-LEVEL FEET
Camera locked off 20cm above the concrete, 24mm wide, square to the alley. Her off-white sneakers hammer through frame left to right at full sprint, kicking dust bursts off the concrete. Two beats later his scuffed grey-toed derby shoes drag through the same path, the left foot turning inward on each step. Rack focus from the near concrete grit to the receding legs. Cuts hard on his trailing foot leaving frame.
SHOT 2 (00:01.6–00:03.2) — OVER-SHOULDER CHASE CAM
Handheld at 1.5m directly behind her right shoulder, 35mm, pushing forward at her pace with a live 3cm vertical bounce on each footfall. Her loose dark hair whips back into the lens edge and separates into strands. The striped shutter wall streaks past on the left in horizontal blur. She snaps a look back over her shoulder and he enters the lower right of frame behind her, closer than expected. Exits on a fast whip pan left.
SHOT 3 (00:03.2–00:04.8) — REVERSE FACE DOLLY
Camera retreating ahead of her at 1.5m, 50mm, medium close-up front-on, matched exactly to her pace. Mouth open mid-scream, chest heaving, eyes flicking off-camera behind her. Six-degree dutch tilt on entry drifting back to level across the shot. The wall stripes converge and blur behind her head. Exits as her swinging hand crosses the lens and smears the frame.
SHOT 4 (00:04.8–00:06.2) — ZOMBIE LOW-ANGLE PUSH-IN
Camera low at 0.6m looking up, 40mm. He lurches directly toward lens, head lolling, jaw grinding, his right hand entering the top of frame reaching past camera. Digital scale-in punch of roughly fifteen percent stacked with a 4Hz handheld vibration. The clouded pale-blue eye catches the bounce light off the shutter slats and goes flat white for two frames. Cuts hard.
SHOT 5 (00:06.2–00:07.8) — WIDE LATERAL TWO-SHOT
Lateral tracking dolly at 1.6m, 32mm, full-figure wide, moving right at the runners' pace. Both figures in frame against the striped wall, gap holding at roughly one and a half strides. The maroon wheelie bin sits ahead in their path at the right of frame. No effects, no shake — the clean breathing shot of the piece. She breaks left around the bin as the shot ends.
SHOT 6 (00:07.8–00:09.0) — BIN ARC AND WIPE
Camera arcs thirty degrees around the wheelie bin at 1.2m, 28mm, as she cuts past it on the outside. The bin's maroon body sweeps into the foreground and fills the frame edge to edge in a soft-focus wipe. The next shot emerges out of the maroon as the bin clears.
SHOT 7 (00:09.0–00:11.2) — THE NEAR-MISS. SIGNATURE VISUAL EFFECT.
Speed drops to approximately twenty percent. High three-quarter angle at 2.2m looking down across her shoulder, 75mm, tight on the cream jacket's shoulder seam. His grey-green fingertips enter frame and graze the wool without gripping — the fabric compresses under the fingertips, then releases and springs back. Individual hair strands hang suspended in the air around her neck. Dust motes float motionless in the pale bounce light. Speed ramps smoothly back to one hundred percent across the final four frames as the fabric snaps free and her shoulder pulls away.
SHOT 8 (00:11.2–00:12.6) — BIN TOPPLES
High angle at 4m looking almost straight down, 24mm. He shoulder-checks the wheelie bin as he passes. The hinged lid flaps open, the bin tips, and it rolls across the concrete behind them trailing grit. Camera tilts down and drifts right, losing them at the bottom of frame. Cuts hard.
SHOT 9 (00:12.6–00:14.0) — LOOK-BACK SNAP
Handheld medium close-up at 1.6m, 50mm, snap-panning right to follow her head turn as she checks behind her. Heavy rolling-shutter smear across the pan. Her hair crosses her face and pulls clear. She registers the widening gap. The smear carries directly into the next frame.
SHOT 10 (00:14.0–00:15.6) — LONG LENS COMPRESSION
Static locked-off telephoto from far down the alley, 135mm, both figures small in frame and stacked against each other by the compression. The wall's horizontal stripes flatten into hard graphic bands. The gap between them is now visibly three strides, clearly wider than it was. Lowest energy point in the edit — no camera movement, no effects, deep haze.
SHOT 11 (00:15.6–00:17.2) — THE STUMBLE
Static low wide at 0.9m, 35mm. His lead foot catches the rolled bin. He pitches forward and drops onto one hand, palm scraping concrete, then hauls himself back up without his head ever leaving her — the eyes track her the whole way down and back. She is already at the right edge of frame, three strides clear and still accelerating.
SHOT 12 (00:17.2–00:19.5) — THE TURN AND THE BLOWOUT
She reaches the open mouth of the alley and, instead of running on, stops dead and turns to face him full-on — chest heaving, no longer fleeing. Camera holds level at 1.8m, 40mm, on the two of them squared off, the open daylight of the alley's end flaring hard behind her. He keeps closing. On his final stride the daylight behind her blooms into total white, overexposing outward from her silhouette until the frame is pure blown-out white for the final two frames. No dissolve — the white is a hard wall the alley disappears into.
SHOT 13 (00:19.5–00:21.5) — FIELD EMERGENCE
The white resolves directly into the reed field, mid-motion, no fade. Wide shot, low sun glowing behind the subjects with soft warm lens flare, hazy mountain silhouette in the background. She is already running left to right through tall golden reeds, laughing now instead of screaming, glancing back over her shoulder. The man — now fully human, clean, in a navy blazer — runs behind her, reaching one arm toward her with a joyful smile. Side-tracking camera at chest height, warm backlight, shallow depth of field, romantic cinematic color grade.
SHOT 14 (00:21.5–00:24.0) — SIDE-TRACKING RUN
Camera closer now, continuing the side-track through the reed field. She laughs brightly, hair blowing in the wind; he reaches his hand further toward her, both fully smiling, sun flare intensifying as they move through a patch of brighter backlight. Warm glowing atmosphere, soft golden bokeh, smooth tracking camera motion, no cuts within the shot.
SHOT 15 (00:24.0–00:27.0) — THE CATCH AND TURN
Camera holds at 1.6m, 65mm, as he finally reaches her — his outstretched hand meets hers or her arm, and instead of a struggle it becomes a pull-in: she turns into him mid-stride, momentum carrying them into a stumbling half-spin that settles into an embrace, both slightly out of breath, both laughing. Eye contact held for a full beat before her face tips into his shoulder — lashes wet, laughing through what reads as both relief and joy. Warm rim light separates both figures from the golden grass behind them. Slow 4% push-in across the shot.
SHOT 16 (00:27.0–00:30.0) — WIDE RESOLUTION
Camera pulls back to 3m, 32mm, wide two-shot of the embrace standing still in the swaying grass, low sun and mountain silhouette centered behind them, lens flare soft across the upper frame. Neither figure moves except gentle breathing, the wind through the grass, and her hand resting on his back. Grain settles. Frame is still when it ends.
PERFORMANCE
Woman (Act 1): brows drawn tight, eyes wide and wet with catch-light, jaw dropped in a raw open scream, neck tendons standing with each gasp, chin driving forward on the acceleration beats. Woman (Act 2): the same fear releases into breathless laughter and open relief the instant she crosses into the field.
Zombie (Act 1): jaw hanging slack and grinding, broken teeth bared through split lips, the clouded eye fixed and unblinking while the bloodshot eye tracks her continuously, guttural throat movement on each lunge, no facial expression beyond the mechanical. Man (Act 2): warm, open, smiling, fully human — no trace of the mechanical carried over.
PHYSICS
Feet strike concrete with visible dust puffs on every footfall in Act 1; in Act 2 feet part the reeds with a soft rustle and no dust. Her cream jacket snaps and flares with each stride throughout both acts, the open zip flapping. Her loose hair separates into distinct strands and trails behind her head, never tying or clumping. In Act 1 his suit jacket flaps stiff and heavy, the torn shoulder seam gaping wider as he runs, and dried blood on his collar stays static and matte — no new blood, no fresh spatter. In Act 2 his navy blazer moves naturally with his stride, unmarked. The wheelie bin has real plastic weight when struck, the lid hinging loose, the roll decelerating naturally across the concrete.
LIGHTING
Act 1: low cool ambient at 4200K, no direct key, both faces lit by bounce off the pale shutter slats, faint cobalt and mustard color spill from the painted stripes, subjects slightly backlit at the rim throughout, heaviest haze at Shot 10. Act 2: warm low-sun backlight at roughly 3200K, strong practical lens flare, soft golden ambient fill, no cool tones remaining anywhere in frame.
COLOR GRADE
Act 1: desaturated teal-cyan base pulled from the blue-grey slats, the mustard stripe and maroon bin plastic the only warm accents, blacks retained rather than crushed. The Shot 12 blowout carries the grade from cool to a clipped, textureless white. Act 2: grade flips to warm amber and gold, gentle highlight bloom, soft filmic contrast, no crushed blacks.
WARDROBE
Zombie / Man: Act 1 — charcoal wool suit jacket stiff with grime and grey dust, torn open at the left shoulder seam, white dress shirt with rust-brown spatter across the chest and right shoulder, top three buttons torn away, collar askew, dust-caked trousers, grey-scuffed black derby shoes. Act 2 — navy blazer, white dress shirt, clean and unmarked, same build and facial identity as Act 1 minus the wounds and grime.
Woman: cream wool-blend zip jacket with the zip open to mid-chest and the collar flaring with motion, white ribbed top underneath, tan straight-leg trousers, off-white low sneakers, long dark hair worn loose and untied throughout both acts.
AUDIO
Act 1: ragged panicked breathing building to a scream; low guttural groaning on each of his lunges; two distinct sets of footfalls on concrete, hers light and fast, his heavy and uneven; the hollow plastic clatter and roll of the toppled bin; a faint metallic resonance off the shutter door. Shot 12: the ambient sound drops out under the white blowout. Act 2: wind moving through reeds, her breathless laughter, both sets of footfalls softened on grass, no dialogue, near silence under the final still embrace.
STYLE
Photoreal cinematic horror-thriller shifting into romantic drama. Fine 35mm-style grain throughout, pore-level naturalistic skin detail on both faces, practical prosthetic makeup register on the zombie rather than digital creature rendering in Act 1. No stylization beyond the color grade in either act.
OUTPUT SETTINGS
16:9, 24 fps, 30 seconds total. Real-time speed throughout except the specified slow-motion ramp in Shot 7.
POSITIVE LOCKS
<<<image_2>>>/<<<image_3>>> and <<<image_4>>> identities, wardrobe, and (in Act 1) wound placement stay 100% consistent with their references in every frame. The man's facial identity is continuous across the wound/no-wound transition — same bone structure, same hairline — only the condition of the skin and wardrobe changes at the blowout. Her hair stays loose and untied throughout both acts, never tied back. Direction of travel is screen-left to screen-right in every shot without exception across both locations. The gap between them narrows through Shot 7 and widens from Shot 8 through Shot 11 before closing again in the field act. His hand grazes the jacket in Shot 7 and never achieves a grip until the intentional catch in Shot 15. The white blowout in Shot 12 is the only location transition in the piece — no other dissolves or fades occur. Cuts occur only at the sixteen specified timestamps and nowhere else.

Why it works

  • Five tagged reference images split the jobs: the alley, the man as a zombie, the same man human, the woman and the field.
  • Travel is screen left to right in every shot of both acts, which is what lets the blowout read as one continuous run.
  • The gap between them is tracked shot by shot: narrowing through Shot 7, widening through Shot 11, closing in the field.
Made with Seedance 2.530s

by @AIwithSynthia23.4K viewsOriginal post

Zombie lover in a farmhouse kitchen, short version

In four paragraphs, a survivor lowers and drops her pistol, opens her arms to her infected lover, he lunges at her, and a smash cut lands on golden wheat-field memories of the two of them.

  • Text to video
  • Smash cut
  • Short prompt
  • Korean horror
Video prompt1,380 chars
Create a short 16 9 photorealistic Korean cinematic zombie-horror scene in an abandoned farmhouse kitchen. A crying female survivor from Korea with messy dark hair, dried blood on her forehead, and a dirty olive-green hooded anorak aims a black pistol at her infected lover. She is trembling, tears running down her face, but cannot pull the trigger.
The infected man has tangled dark hair, pale dirty skin, milky-white eyes, dark veins, torn clothing, and an increasingly feral expression. He snarls loudly and screams directly at her. She slowly lowers the pistol, drops it beside her, and opens her arms toward him. He suddenly screams again, lunges forward and violently grabs her face.
Hard smash-cut to warm golden-hour memories of the same couple happily together in a vast wheat field — laughing, holding each other, walking hand-in-hand through the golden grass, and sitting together beneath the sunset. Make the contrast between the cold present and beautiful memories emotionally powerful.
Photorealistic live-action cinema, natural skin and tears, realistic infected eyes and facial expressions, handheld camera, practical-effects horror, 35mm film grain, believable movement and lighting. No gunshot, no gore, no dialogue, no slow motion, no extra plot, no CGI look, no animation, no distorted anatomy, no duplicate characters, no subtitles, no text, no watermark.

Why it works

  • At about 1,400 characters it gives the story and the look, and leaves shot choice and timing to the model.
  • It names what the cut is for ('the contrast between the cold present and beautiful memories'), not only that there is one.
  • The closing negative list rules out a gunshot, gore and slow motion, the three things this story most invites.
Made with Seedance 2.530s

by @YourAlphaMom1.3K viewsOriginal post

Zombie lover match-cut into sunset memories, beat by beat

A woman lowers her pistol and opens her arms to the man she loves after he has turned, and the moment he reaches her cuts to the same embrace on a sunlit hillside, across nineteen timestamped beats.

  • Text to video
  • Match cut
  • Timestamped beats
  • 30 seconds
Video prompt10,131 chars
Create a 30-second, horizontal widescreen, ultra-realistic live-action short film at 24 fps about a woman confronting the man she loves after he has become a zombie.
She aims a pistol at him, struggles with the decision, then lowers the weapon and opens her arms. He rushes into her, and their physical contact cuts directly into an embrace from their happy past.
The opening confrontation lasts approximately eleven seconds. The remaining time shows their relationship through several intimate memories, ending with them together at sunset.
CHARACTERS
WOMAN: A woman around 30 with expressive brown eyes, brown hair loosely tied back, natural skin texture, and a recognizable, emotionally expressive face.
In the present, she wears a worn olive-green jacket, a dark-green scarf, and dark trousers. Her face looks exhausted. Her eyes are reddened and wet, with visible tears on her cheeks.
She holds a black pistol in her RIGHT hand, initially supported by her LEFT hand.
In the memories, she has the SAME face and hair colour, with cleaner, more relaxed styling. She wears a light cream blouse and dark trousers. Her expressions are warm and spontaneous.
MAN: A man around 30 with short dark hair, light beard stubble, and a clearly recognizable face.
In the present, he has cloudy pale eyes, an unhealthy grey complexion, dirt and superficial marks across his face, dishevelled hair, and an aggressive expression. He wears a worn dark-grey short-sleeved shirt and dark trousers.
In the memories, he has the SAME facial structure, hairstyle, and beard, with healthy skin and normal eyes. He wears a pale-blue shirt and dark trousers.
Preserve both identities across the present and the memories. The healthy man must immediately register as the same person seen in the kitchen.
PRESENT-DAY LOCATION
A neglected kitchen in an old house: stained pale walls, worn wooden cupboards, open shelves, scattered household utensils, a sink, and a large weathered window.
The woman stands near the entrance. The man stands several steps opposite her near the kitchen counter.
Establish their positions through matching eyelines and a shot over the woman’s shoulder.
In the side view used immediately before the memory transition, the woman is on SCREEN LEFT and the man approaches from SCREEN RIGHT.
VISUAL LANGUAGE
Use cinematic live-action photography with realistic skin, restrained performances, shallow depth of field in close-ups, and controlled, subtle camera movement.
Present-day scenes have muted olive, grey, and brown tones. Window daylight illuminates the faces unevenly, leaving believable shadows in the room. Tears catch small highlights.
The memories take place on a grassy hillside in late-afternoon and sunset light. Tall golden grass moves in the breeze, with distant rolling hills and a wide sky.
Warm backlighting outlines their hair and shoulders. Faces remain readable, with natural skin texture, soft background blur, and occasional gentle sunlight flare.
The shift from the dim kitchen to the open, sunlit field is immediate and emotionally clear.
Use deliberate cuts between faces, shared actions, and wider views. Give each expression enough time to register.
00:00–00:01.7 — THE WOMAN AIMS
Open immediately on a close-up of the woman aiming the pistol toward the man just outside the frame.
Her extended hands and pistol occupy the soft foreground. Focus rests on her tearful eyes.
Her brows tighten. Her lips press together as she struggles to maintain her aim.
00:01.7–00:03 — REVEAL HIM
Cut over her shoulder.
Her head, arms, and pistol remain softly blurred in the left foreground. The man stands near the kitchen counter in the background.
His head tilts slightly. His cloudy eyes stare at her, and his arms hang at his sides.
Keep their distance and opposing positions readable.
00:03–00:04 — SHE HOLDS HER AIM
Cut to a frontal medium shot of the woman.
Both arms remain extended. She keeps looking at him, breathing visibly through the tension.
The doorway and window establish the same room behind her.
00:04–00:05.2 — HIS FACE
Cut to a close-up of the man.
His upper lip lifts, his teeth become visible, and his expression grows more threatening.
His face remains recognizably human beneath the signs of infection.
00:05.2–00:06.5 — HER DECISION
Cut tighter on the woman’s face.
A tear travels down her cheek. Her lower lip trembles slightly.
She continues looking directly at him. Her expression gradually shifts from fearful determination to grief and tenderness.
00:06.5–00:09.5 — SHE OPENS HER ARMS
Return to the frontal medium shot.
She slowly lowers the pistol from its aiming position.
Her LEFT hand releases its supporting grip. She separates her arms and opens them wide, inviting him into an embrace.
The pistol stays in her RIGHT hand, lowered and directed away from him as that arm moves outward.
Hold briefly on her open arms and tearful face. Her posture is vulnerable and deliberate.
00:09.5–00:10.1 — HE RUSHES FORWARD
Cut to the man’s close-up.
He opens his mouth and suddenly lunges toward her. His face rapidly approaches the lens with natural motion blur.
00:10.1–00:10.7 — SHE CLOSES HER EYES
Cut to the woman’s face.
She closes her eyes and stays in place, her arms still open outside the close framing.
Her expression remains sorrowful and calm.
00:10.7–00:11 — CONTACT
Cut to a medium side view of the kitchen.
The woman stands on SCREEN LEFT. The man rushes in from SCREEN RIGHT.
His arms reach around her shoulders as their bodies meet. She begins closing her arms around him, keeping the pistol-bearing hand outside their bodies.
Cut at the moment they come together. The consequences of the attack remain outside the image.
00:11–00:12.6 — MATCH CUT INTO THEIR PAST
Cut directly to the SAME couple in the sunlit field, already meeting in a loving embrace.
Maintain the woman on SCREEN LEFT and the man on SCREEN RIGHT at the cut. Match the placement of their shoulders and the direction of their movement so the embrace visually continues across the change of location.
They are healthy and wearing their memory outfits. These scenes take place before his infection.
His arms wrap around her shoulders; hers close around his torso. They lose their balance playfully and sink together into the tall grass.
She settles onto her back while he supports himself beside and above her.
00:12.6–00:14.5 — HIS SMILE
Cut to a close-up of the healthy man leaning over her.
He laughs softly with an open, affectionate expression.
Her hand enters the frame and rests against his cheek. He turns his face slightly into her palm.
Warm sunlight outlines his hair. His attention stays on her.
00:14.5–00:16.3 — HER ANSWERING LOOK
Reverse to a close-up of the woman lying in the grass.
Part of his shoulder and face occupies the soft foreground.
She smiles up at him, eyes narrowing naturally with happiness. A loose strand of hair crosses her cheek.
Let this expression echo the earlier close-up of her tearful face in the kitchen.
00:16.3–00:18.3 — A SMALL SHARED JOKE
Cut forward to the couple sitting beside each other in the grass.
She notices a dry grass stem caught in his hair, reaches up, and carefully removes it.
She holds it between them with an amused smile. He looks at the stem, then back at her, and they both laugh.
Keep the gesture simple and clearly visible.
00:18.3–00:20 — THEIR HANDS FIND EACH OTHER
Cut to a close view of their hands as they stand side by side.
Her hand brushes his. He turns his palm toward hers, and their fingers interlace.
Their joined hands remain connected as they begin moving forward. The camera briefly follows the hands against the softly glowing grass.
00:20–00:22.5 — RUNNING THROUGH THE FIELD
Cut to a wider rear view.
The woman runs on the LEFT and the man on the RIGHT, holding the same joined hands.
They run away through the tall grass toward the warm light. She looks across at him briefly, and he returns the glance.
The camera follows with gentle human movement. Nearby grass passes through the foreground, and their clothing moves naturally with their strides.
00:22.5–00:24.3 — TOGETHER AT SUNSET
Cut forward to them seated on the hillside, backs toward the camera.
She sits on the LEFT and he sits on the RIGHT. Her arm rests around his shoulders.
He leans comfortably toward her as they watch the low sun over the distant hills.
Their movement settles. The pace becomes quieter.
00:24.3–00:27 — A LAST TENDER GESTURE
Cut closer from behind and slightly to the side.
He rests his head against her shoulder. She gently kisses the top of his head, holds the contact for a brief moment, then looks back toward the sunset.
He remains tucked against her. Their closeness feels familiar and unperformed.
00:27–00:29.6 — FINAL WIDE VIEW
Cut to a very wide view of the hillside.
The embracing couple appears small near the lower centre of the frame, surrounded by tall grass beneath a broad pink-and-gold evening sky.
Hold this peaceful composition. The story ends within the memory.
00:29.6–00:30 — END
Cut directly to black for the remaining fraction of a second.
CONTINUITY
The emotional progression is: she aims → sees his aggression → struggles to shoot → lowers the weapon → opens her arms → he rushes forward → she closes her eyes → their bodies meet → the contact continues through a cut into a loving embrace from the past.
The kitchen sequence remains one continuous present-day event. The pistol stays in her RIGHT hand throughout that sequence and remains unfired.
The memories use the same two people in their established healthy appearances and light clothing. Their intimacy is shown through reciprocal looks, touch, shared laughter, joined hands, and the final embrace.
The transition into the past occurs through a cut on matching physical action. Every memory retains the same field, consistent clothing, and warm light progressing toward sunset.
Keep the image free of captions, subtitles, titles, logos, watermarks, and interface elements.
AUDIO
Natural scene sound only. Music will be added separately.

Why it works

  • Beats are timed to a tenth of a second (00:10.1–00:10.7), so the lunge, the closed eyes and the contact each get their own cut.
  • Screen positions are locked across the match cut, her on the left and him coming in from the right, so the kitchen embrace carries straight into the field.
  • Each character has a present and a memory wardrobe, with the SAME face written in capitals, which keeps the healthy man readable as the zombie.
Prompt source

Magic Pen3

Point a marker at real things and they turn into flat 2D doodles, first person.

Made with Seedance 2.515s

by @Strength04_X22.9K viewsOriginal post

Magic pen turning street traffic into 2D cartoons

A 15-second vertical phone vlog in one shot: the vlogger points a marker, says 'Biu!', and a bus, a bicycle, a scooter and a pedestrian each turn into flat cel-shaded cartoons before the whole road does.

  • Text to video
  • Magic pen
  • 9:16 vertical
  • Phone POV
Video prompt4,139 chars · 5 beats
Magic Pen Street Transport Vlog · 15 Seconds · Vertical 9:16
[STYLE]
Live-action + flat 2D anime sticker compositing, first-person street transport magic vlog. 8K photorealistic detail with the authentic look of a real smartphone rear-camera recording. Strong visual contrast between the photorealistic city and flat cartoon vehicles. One continuous handheld smartphone shot. No cuts. No scene transitions.
[CAMERA]
Raw, unstabilized handheld smartphone footage throughout: natural walking shake, subtle arm movement, occasional autofocus hunting, realistic exposure shifts between bright sky and building shadows, and slight rolling-shutter jelly during fast pans. Natural smartphone HDR colors with soft, realistic tones. The camera quickly whip-pans to follow the marker and each target.
[LIGHTING]
Late-afternoon sunlight coming consistently from one direction. Every real object and every animated character casts a soft contact shadow matching the direction of the sunlight.
[SETTING]
One continuous walk through a Korean city neighborhood. A wide urban road with a pedestrian sidewalk, bicycle lane, city buses, parked scooters, and a crosswalk. Everyday street life, with background pedestrians and vehicles moving naturally.
[PEN — MAGIC RULES]
The vlogger's real hand, with photorealistic skin and visible pores, holds a black marker and remains visible throughout the entire video, serving as the visual guide connecting every beat.
Every transformation follows the exact same sequence:
Marker points → Vlogger says “Biu!” → blue hand-drawn sketch lines wrap around the target → ink spreads → the target transforms into a lively flat 2D cel-shaded anime character with bold cartoon outlines.
The transformed character must preserve the original object's exact size, position, speed, direction, and perspective, remaining perfectly anchored to the real street environment. Characters retain flat sticker-style shading and are never relit by the real-world lighting. Anything they touch reacts with realistic physical interaction.
[00:00–00:03]
First-person view while walking beside a busy road. The vlogger raises the black marker into frame and points at a passing city bus.
“Biu!”
Blue sketch lines wrap around the bus. The bus transforms into a giant flat 2D cartoon vehicle with bold anime outlines while maintaining its exact size, speed, direction, and perspective.
It continues driving normally through traffic.
SFX: Bus engine + marker whoosh + soft cartoon movement sound.
[00:03–00:06]
The marker quickly swings downward toward a bicycle passing along the bicycle lane.
“Biu!”
The bicycle transforms into a cute hand-drawn cartoon bicycle. It continues moving at the same speed, its wheels spinning rapidly, with small sketch particles appearing behind it.
SFX: Bicycle wheel sound + light sketch rustling.
[00:06–00:10]
The marker follows the bicycle toward a parked scooter near the sidewalk.
“Biu!”
A rapid blue sketch outline covers the scooter. It transforms into a flat 2D cartoon scooter while preserving its exact position, proportions, and perspective.
Its wheels spin once and its small cartoon headlight blinks twice.
SFX: Soft engine hum + tiny mechanical click.
[00:10–00:12]
The marker quickly swings toward a pedestrian crossing the road.
“Biu!”
Blue sketch lines trace around the person. The same person, in the exact same pose, proportions, and position, becomes a vibrant anime cel-shaded illustration.
They continue walking naturally and briefly wave once.
SFX: Fast doodling sound + soft sparkle chime.
[00:12–00:15]
The vlogger stops walking and gives one final, louder:
“Biu—!”
The marker spins upward toward the road. A swirling blue ink trail spreads across the vehicles and pavement, transforming the entire road surface into a hand-drawn anime street while the surrounding buildings remain completely photorealistic.
A small handwritten “THE END!” doodle — the only text appearing anywhere in the video — pops up above the road.
Freeze frame.
SFX: Sharp throwing whoosh + spreading watercolor-ink bloom + felt-tip marker squeaking as it writes “THE END!”

Why it works

  • Every transformation follows one written sequence (point, 'Biu!', blue sketch lines, ink spreads, transform), so the beats repeat cleanly.
  • Cartoon objects keep the real object's size, speed, direction and perspective and are never relit, which keeps them anchored in the street.
  • 'THE END!' is declared the only text anywhere in the video.
Made with Seedance 2.520s

by @Strength04_X7.6K viewsOriginal post

Magic pen turning an airport into 2D cartoons

The street template moved into an airport terminal over 20 seconds: a suitcase, a luggage cart, a departure display and a plane outside turn into cartoons, and the last 'Biu!' turns the sky above the glass roof.

  • Text to video
  • Magic pen
  • Remix
  • Airport
Video prompt3,763 chars · 5 beats
Magic Pen Airport Vlog · 20 Seconds · Vertical 9:16
[STYLE]
Live-action + flat 2D anime sticker compositing, first-person airport magic vlog. 8K photorealistic detail with the authentic look of a real smartphone rear-camera recording. Strong visual contrast between the photorealistic airport and flat cartoon objects. One continuous handheld smartphone shot. No cuts. No scene transitions.
[CAMERA]
Raw, unstabilized handheld smartphone footage throughout: natural walking shake, subtle arm movement, occasional autofocus hunting, realistic exposure shifts between bright terminal lighting and window sunlight, and slight rolling-shutter jelly during fast pans. Natural smartphone HDR colors with soft realistic tones. The camera quickly whip-pans to follow the marker and each target.
[LIGHTING]
Bright natural daylight enters through large airport windows, mixed with soft indoor lighting. Every real object and animated character maintains soft contact shadows matching the environment.
[SETTING]
One continuous walk through a modern Korean airport terminal. Large glass windows, check-in counters, luggage carts, digital displays, passengers, and polished floors create a premium realistic environment. Background travelers move naturally.
[PEN — MAGIC RULES]
The vlogger's real hand, with photorealistic skin and visible pores, holds a black marker and remains visible throughout the entire video.
Every transformation follows the exact same sequence:
Marker points → Vlogger says “Biu!” → blue hand-drawn sketch lines wrap around the target → ink spreads → the target transforms into a lively flat 2D cel-shaded anime character with bold cartoon outlines.
The transformed target preserves its exact size, position, speed, direction, and perspective, remaining perfectly anchored in the real airport environment. Characters retain flat sticker-style shading and are never relit by the real-world lighting.
[00:00–00:04]
First-person view walking through the airport terminal. The vlogger raises the marker and points toward a large rolling suitcase beside a traveler.
“Biu!”
Blue sketch lines wrap around the suitcase. It transforms into a lively flat 2D anime suitcase while keeping its exact size and position.
It rolls forward a few steps.
SFX: Airport ambience + rolling wheels + marker whoosh.
[00:04–00:08]
The marker quickly swings toward an airport luggage cart moving through the terminal.
“Biu!”
The cart transforms into a flat 2D anime luggage cart while maintaining its exact size, speed, direction, and perspective.
Its wheels spin rapidly as it continues moving.
SFX: Wheel movement + metallic rolling sound.
[00:08–00:12]
The marker follows toward a large digital departure display.
“Biu!”
The display transforms into a flat 2D anime screen. Its visual elements become animated hand-drawn symbols that move across the screen.
SFX: Electronic beeps + sketching sound.
[00:12–00:16]
The marker swings toward an airplane visible through the terminal window.
“Biu!”
Blue sketch lines wrap around the airplane. It transforms into a giant flat 2D anime airplane while preserving its exact size, position, direction, and perspective.
It slowly moves across the distant runway.
SFX: Distant aircraft engine + magical ink transition.
[00:16–00:20]
The vlogger stops and gives one final louder:
“Biu—!”
The marker spins upward toward the airport glass ceiling. A blue ink spiral spreads across the glass, transforming the visible sky outside into a hand-drawn anime sky while the airport terminal remains completely photorealistic.
A small handwritten “THE END!” doodle appears among the illustrated clouds.
Freeze frame.
SFX: Sharp throwing whoosh + spreading watercolor bloom + felt-tip squeak writing “THE END!”

Why it works

  • The style, camera and pen-rule blocks are nearly word for word the street version's; only the setting and the targets change.
  • Targets move from near to far (suitcase, cart, display, a plane on the runway) and the last one is the sky.
Prompt source
Made with Seedance 2.015s

by @Shorelyn_737 viewsOriginal post

Magic pen at a bus stop: a pigeon, a bus-turned-cat, a schoolgirl

A first-person shot at a Japanese bus stop where the pen has to touch each subject: a pigeon becomes a cartoon bird, an arriving bus becomes a giant orange tabby cat, and a girl on the bench turns anime and waves.

  • Text to video
  • Magic pen
  • First person
  • Seedance 2.0
Video prompt3,407 chars
10-second cinematic mixed-media transformation video, horizontal 16:9, one continuous first-person POV shot with no cuts.
The viewer sees through the eyes of a person standing beside a modern Japanese city bus stop, holding a slim black magic pen in one hand. The environment is fully photorealistic, while every object touched by the pen transforms into clean, colorful 2D comic-style animation. Smooth handheld movement, seamless transformations, playful energy, precise object tracking, realistic daylight, no subtitles, no text, no logo, no watermark.
**0:00–0:03 — Bird Transformation**
The continuous POV shot begins facing the sidewalk near the bus stop. A real pigeon suddenly flies toward the camera and briefly hovers directly in front of the person. The hand quickly raises the magic pen and gently taps the bird in mid-air. At the exact moment of contact, a small blue ink flash spreads across its body, instantly transforming it into a cute 2D comic-style white bird with bold blue outlines, expressive eyes, and exaggerated flapping wings. The cartoon bird circles playfully around the camera and flies toward the street.
**0:03–0:06 — Bus Transformation**
Without cutting, the camera naturally follows the cartoon bird toward the road as a real city bus approaches the bus stop. The person extends the magic pen and taps the front of the moving bus as it arrives. A bright orange illustrated ripple spreads from the point of contact. The entire bus smoothly transforms into a giant, adorable 2D comic-style orange tabby cat with green eyes. The cat continues moving forward at the same speed and along the same path as the bus, casually walking beside the curb. Its tail sways through the air while small hand-drawn motion marks appear around its paws.
**0:06–0:10 — High-School Girl Transformation**
The camera turns smoothly from the giant cartoon cat toward the bus-stop bench. A realistic Japanese-style high-school girl is sitting there in a neat school uniform, holding her bag on her lap. The person walks closer and gently taps her shoulder with the magic pen. A clean blue-and-white comic outline rapidly traces around her body, transforming her into a polished 2D anime and comic-style high-school girl while preserving the same pose, uniform, hairstyle, and facial identity. She looks toward the camera, smiles, stands up, and waves happily. The cartoon bird lands beside her while the giant orange cat looks back from the street.
**Ending frame:** The continuous POV camera holds briefly on the comic-style girl waving, the cartoon bird beside her, and the giant orange cat walking past, all perfectly integrated into the realistic Japanese city environment.
**Visual style:** High-quality photorealistic live-action environment mixed with crisp 2D cel animation, clean blue and white outlines, bold solid colors, accurate shadows, consistent scale and perspective, fluid first-person camera movement, seamless magical transformations, whimsical Japanese commercial aesthetic.
**Important constraints:** One continuous unbroken shot, no cuts, no hidden transitions, no teleporting camera, no scene changes, no object transforming before physical contact with the magic pen, the pen must visibly tap each subject, preserve the realistic environment, no 3D cartoon rendering, no flickering, no deformed hands, no extra fingers, no subtitles, no words, no logos, no watermark.

Why it works

  • Contact is the trigger: nothing may transform before the pen visibly taps it.
  • The bus does not just turn flat; it becomes a cat that keeps the bus's speed and path.
  • Earlier transformations come back in the last frame (the bird lands by the girl, the cat looks back), which ties the three beats together.

Hotel Lobby AI Rap Duo3

Two people trade bars in a studio or a hotel lobby, face-swapped onto a reference clip or written from scratch.

Made with Seedance 2.021s

by @Just_sharon786.3K viewsOriginal post

Two-person rap performance at a red car

A rapper stands at the rear of a red car while a man sits on the trunk nodding along, each written with a vocal profile, a signature tic and an eye-contact rule.

  • Image input
  • Two references
  • Performance tics
  • Seedance 2.0

Inputs @[Image 1](image_1) — the female rapper · @[Image 2](image_2) — the man on the car

Video prompt2,628 chars
A confident white female rapper @[Image 1](image_1). Blaze hair styled in double buns pulled up high. Pink Zipp Republic jersey, layered gold chains and cross necklaces, gold star drop earrings, small cross tattoo under the left eye. Standing firmly beside the car's rear quarter panel, shifting weight from heels to toes in time with the beat. Using both hands to express the rhythm, with gestures like cutting the air, pointing fingers, or stacking bars in mid-air.
Vocal profile: A white woman speaking in a London accent with a Nigerian lilt. Low chest-voice range, crisp consonants, punching the end of each bar with forceful delivery. Between bars, drop the volume and mumble low.
Signature tic: At the hook line's entry, thrust both hands forward simultaneously for a double point. But at the moment a bar lands clean, drop the shoulders back, and half a beat before the next line starts, lift the chin slightly.
Eye contact: Stare straight into the lens, blinking slowly and deliberately. During transitions, flick the gaze momentarily to the man sitting atop the car, then back to the lens. Match the character's appearance 100% to the reference image. Do not change anything except the character's appearance.
@[Image 2](image_2) - Already referenced from the image. White male, early to mid-30s. Short tight black hair, trimmed mustache connected to a full black beard, broad nose, thick lips, sturdy build with broad shoulders, prominent tattoos on both forearms. Wearing a red Zipp Republic 2.0 jersey with a white outline graphic and "2.0" numbering on the chest. Layered gold Cuban link chains, black jeans, black low-cut trainers, gold wristwatch.
Seated on the red car's trunk lid, both feet planted firmly on the rear bumper for balance. Upper body swaying loosely in a relaxed state to the beat, nodding the head low with each snare hit then lifting it slightly. Lightly tapping the thigh with one hand to mark the rhythm.
Psychological engine: A calm intimidation and ease, not showboating the performance but surrendering to the groove itself.
Signature tic: Each shoulder roll catches the light on the chains, while the chin stays steady during nods. At the moment the rapper lands a punchline, turn the face toward her, hold the gaze for one beat, then return to his own rhythm.
Eye contact: Calm eyes half-closed, with an unhurried, languid gaze. Naturally shifting between the rapper and the lens.
Match the character's appearance 100% to the reference image. Do not change any other elements. All appearing characters are Western/Caucasian only. Both perform singing and rapping with a Nigerian accent.

Why it works

  • Each performer's tic is tied to the music: a double point when the hook comes in, a head turn toward her when she lands a punchline.
  • Eye contact is scripted per person: she holds the lens, he drifts between her and the camera.
  • No lyrics are written, and the accent is given two ways (London with a Nigerian lilt, then 'both' with a Nigerian accent), so check what comes back.
Made with Veo 38s

by @ZHO_ZHO_ZHO56.2K viewsOriginal post

Newton and Einstein rap battle on a sci-fi stage

One paragraph: Newton raps about gravity and apples in a British accent, Einstein answers about relativity in a German accent, on a neon stage with holographic equations and a crowd of scientists.

  • Text to video
  • Veo 3
  • Rap battle
  • One paragraph
Video prompt706 chars
A high-energy rap battle between Isaac Newton and Albert Einstein on a futuristic sci-fi stage. The camera alternates between close-ups and dramatic wide shots as they diss each other with sharp lyrics. Newton, in a classic 17th-century outfit, raps with a British accent about gravity and apples. Einstein, with wild hair and a German accent, fires back about relativity and space-time. Their lip-sync is perfectly timed to the beat, and their facial expressions are intense and animated. The background pulses with neon lights and holographic equations, reacting to the rhythm. The crowd of AI-generated scientists cheers them on in sync with the music. It feels like a rap battle from another dimension.

Why it works

  • Each rapper gets a costume, an accent and a topic in one clause, about as much as a single short exchange can carry.
  • It was written for Veo 3, not Seedance, and casts real historical figures; no lyrics are given, so the words are the model's.
Prompt source
Made with Seedance 2.529s

by @OriSilver5K viewsOriginal post

One-line face swap onto a two-person rap clip

A single sentence: copy @Video1 exactly, with @Image1 as the guy on the left and @Image2 as the guy on the right.

  • Reference video
  • Face swap
  • One sentence
  • Two people

Inputs @Video1 — the original clip · @Image1 — the guy on the left · @Image2 — the guy on the right

Video prompt169 chars
Exactly the same clip as @Video1 but left guy singing is @Image1 right guy in clip is @Image2 every movement every gesture everything including sound is exactly the same

Why it works

  • Left and right are each tied to one image, the one detail a two-person swap cannot skip.
  • It asks for the sound to be copied too, so the original track comes along with the clip.
  • At 169 characters it leans entirely on the reference clip.
Prompt source

STORM Clone Dance4

You stand still in the middle while a crowd of identical yous moves around you.

Made with Seedance 2.530s

by @AIwithkhan262.8K viewsOriginal post

STORM trend: a woman in black stands while the crowd moves

With uploaded images setting the woman and the crowd, she walks to the center, takes one drag, exhales toward the lens and stops, and the crowd behind her starts moving in sharp synchronized jerks.

  • Image input
  • STORM trend
  • Crowd in sync
  • Reference images

Inputs Uploaded images — the woman, crowd, wardrobe, framing and atmosphere

Video prompt1,593 chars
Create a ultra-realistic cinematic fashion video using the uploaded images as exact visual references for the woman, crowd, wardrobe, framing, and atmosphere.
A young woman with the same short dark bob wears a fitted black turtleneck, black trousers, and black shoes, standing confidently at the center of a large crowd.
Start with her walking slowly toward the middle of the crowd, calm and expressionless, with a lit cigarette naturally held between her lips.
She reaches the center, stops, and faces directly into the camera.
She takes one slow drag from the cigarette, then removes it from her mouth.
She slowly exhales a visible stream of smoke toward the camera, with the smoke naturally drifting across her face.
She lowers her hand and becomes completely still, staring directly into the lens.
The entire crowd behind her remains frozen for a brief moment.
Suddenly, everyone behind her begins moving in perfect synchronization — sharp head turns, shoulder jerks, folding forward and snapping upright.
The movement should feel rhythmic, strange, powerful, and physically realistic.
The woman does not move at all while the entire crowd moves behind her.
Slowly push the camera closer to her, keeping her sharply focused while the crowd has subtle depth-of-field blur.
Dark, cold, mysterious lighting with deep shadows, realistic skin, hair, clothing, cigarette ember, and physically accurate smoke.
Photorealistic live-action quality, cinematic composition, natural human movement; no animation, CGI look, duplicated woman, distorted bodies, excessive smoke, text, subtitles, or logos.

Why it works

  • The contrast is stated outright: 'The woman does not move at all while the entire crowd moves behind her.'
  • The crowd's moves are named (head turns, shoulder jerks, folding forward and snapping upright) instead of 'dancing'.
  • There are no timestamps; the order of the sentences carries the pacing.
Made with Seedance 2.526s

by @mrdejie38.7K viewsOriginal post

STORM trend: lead in white inside a crowd of black suits

A timed music-video shot: the lead smokes, walks into a formation of 40–60 men in black suits, turns to face the camera at the center and stands while they break into synchronized chest-clutching and bowing.

  • Image input
  • STORM trend
  • Timed beats
  • Reference image

Inputs Reference image of the lead (face, body, hair and outfit)

Video prompt3,876 chars
Photoreal cinematic music-video shot, 30 seconds, 24fps, shallow-to-medium depth of field, overcast natural daylight, slight film grain. Use the provided reference image as the exact face, body, hair, and outfit of the lead. The lead is the only person wearing white.
**Lead character ( I use myself as the reference):**
**Background dancers:** 40–60 men of mixed East Asian and mixed ethnicities, various hair lengths.
**0.00–3.00s — Tight frontal close-up**
Camera locked on the lead’s face and upper torso, eye-level, slight handheld micro-sway. He faces camera wearing black rectangular sunglasses. Cigarette in his mouth. He exhales a slow, thick cloud of white smoke that drifts across the lens. At 1.0s he raises his right hand (gold rings visible) and takes the cigarette between index and middle finger, bringing it to his lips for another short drag. At 2.2s he lowers the cigarette, still looking slightly down-camera with a frown. Background: out-of-focus wall of men in black suits standing motionless.
**3.00–4.20s — Head and body turn**
He drops his gaze and turns his head and torso to his right (screen left), presenting the back of his head and the embroidered white jacket. Camera stays tight, following the turn.
**4.20–9.50s — Tracking from behind as he walks into the formation**
Camera pulls back to a medium-wide over-the-shoulder / behind-the-back tracking shot. The lead walks slowly and steadily forward into the dense ranks of black-suited men. The men remain completely still, eyes forward or slightly down, like a living wall. Camera dollies backward at the same pace he walks, keeping him centered. We see more and more rows of black-suited men filling the frame. The brick building and arch become visible in the far background.
**9.50–13.80s — He threads through the crowd**
He continues walking, now weaving between two columns. A few men shift half a step to let him pass but otherwise stay frozen. Camera continues tracking backward and slightly rising so we see more of the formation. By 13s he is deep inside the group.
**13.80–16.50s — He turns and settles in the middle**
He stops, pivots 180 degrees so he now faces the camera again, standing in the exact center of the formation, one row back from the front line. He takes another drag, smoke curling. Camera has now become a locked wide shot looking slightly up the steps, the entire group framed symmetrically, brick arch behind them. He is the only figure in white; everyone else is black.
**16.50–20.80s — Stillness, then first twitch**
Hold the wide composition. All black-suited men remain statue-still. The lead stands motionless except for the cigarette. At ~20.5s the men begin a tiny, almost invisible shift of weight.
**20.80–22.50s — Choreo ignition**
On a sharp musical beat the entire black-suited group explodes into the same synchronized motion from the reference: they clutch their chests / ties with both hands, twist their torsos, and snap their heads in different directions. Movement is sharp, aggressive, and perfectly timed across the whole crowd. The lead does not participate; he stays upright, smoking, watching them.
**22.50–26.20s — Full collapse / bow wave**
The dancers drop into a deep, chaotic-looking but actually tightly choreographed bow: they fold at the waist, heads whipping down then snapping back up, hair flying, some looking skyward with mouths open, others clutching their jackets. The motion travels through the ranks like a ripple. Camera remains the same wide locked-off shot so we see the whole staircase of bodies convulsing around the single still white figure in the center. He takes one more slow drag.
**26.20–30.00s — Hold and settle**
The dancers continue the bowing / clutching / head-whip cycle two more times, then freeze in a mixed pose—some still bent, some half-risen, all looking strained.

Why it works

  • One color rule separates him from everyone: the lead is the only person wearing white.
  • Beats are timed to fractions of a second, down to a first twitch at about 20.5 seconds before the choreography starts at 20.8.
  • The 'Lead character' heading is left empty with a note that the author used himself, so your reference image carries the whole look.
Prompt source
Made with Seedance 2.530s

by @john_my0729K viewsOriginal post

STORM trend remake: swap the central man for your own face

A character-replacement prompt: the uploaded reference video controls motion, camera, timing and cuts, and <image_1> replaces the central man with a woman in an acid-lime blazer.

  • Reference video
  • Character swap
  • Reference video
  • STORM trend

Inputs Uploaded reference VIDEO — motion, camera, timing and cuts · <image_1> — the woman who replaces the central man

Video prompt6,845 chars · 5 beats
Use the uploaded reference VIDEO as the exact motion, choreography, camera, framing, timing, pacing, shot duration and cut-point reference.
Use <image_1> as the STRICT identity reference for the MAIN WOMAN.
The final video should feel like the SAME original video, but with the original central man completely replaced by the woman from <image_1>.
Do NOT recreate or preserve the original man's face or identity.
IDENTITY <image_1>
The central character is the exact same woman shown in <image_1>.
Preserve her:
- exact face
- facial structure
- jawline
- cheekbones
- nose
- eyes
- eyebrows
- lips
- ears
- skin tone
- natural facial proportions
- hairstyle
- hair length, volume and texture
- any facial features visible in the reference
- natural body proportions
Maximum identity preservation throughout the entire video.
She must remain recognizably the same person from the uploaded photograph in every shot and every angle.
NO face morphing.
NO identity drift.
NO changing facial structure.
NO different hairstyle.
NO unwanted hair changes.
NO beautification or artificial facial retouching.
NO age changes.
NO facial distortion.
CHARACTER REPLACEMENT
Replace ONLY the original central man with the woman from <image_1>.
Keep the exact position, scale, body movement, gestures, walking path, head movements, smoking actions and timing from the reference video.
Everything the original central character does in the reference video should be performed by the <image_1> woman at the same moment and with the same choreography.
Do not reinterpret or redesign the choreography simply because the replacement character is female.
WARDROBE
Dress <image_1> in the same character-reference outfit, adapted naturally for the woman:
Acid lime green single-breasted blazer, notch lapels, two buttons, buttoned closed, slightly oversized shoulders.
Plain white cotton shirt underneath.
Narrow black knitted tie, neatly knotted.
Straight black trousers.
Plain black leather ankle boots, low heel.
No jewellery.
No watch.
No belt.
No bag.
No phone.
No logos or patterns.
The lime blazer must remain exactly throughout the entire video.
IMPORTANT:
Do not let the wardrobe replacement change the original choreography, body positioning, composition or pacing.
SMOKING ACTION
Preserve the smoking sequence from the reference video exactly.
At the beginning, the main woman holds a cigarette naturally between her fingers and brings it to her mouth.
She lights the cigarette using the same hand movements, timing and body positioning as the reference video.
She takes a drag and naturally exhales smoke.
Preserve the original:
- hand position
- cigarette position
- lighter movement
- timing
- head movement
- facial expression
- smoke timing
- arm movement
The smoking action should look physically natural and cinematic.
Do not exaggerate the smoke.
Do not create excessive smoke.
No smoking action should occur from any other character.
All other characters remain unchanged.
EXACT ORIGINAL ACTION / TIMING
0–4 SECONDS:
Match the reference video exactly.
The central woman is shown smoking.
She brings the cigarette toward her mouth, lights it, takes a drag and exhales.
Match the original close framing, facial positioning, hand movements and timing.
4–11 SECONDS:
Match the reference video exactly.
The woman turns away from camera and moves forward into the crowd using the exact same movement and timing as the original central character.
Preserve her body position, walking speed, direction and camera framing.
11–17 SECONDS:
Match the original choreography exactly.
The woman moves back toward the central position and performs the same gestures and movements as the original central character.
Do not invent new actions.
17–22 SECONDS:
The woman reaches and holds the same central position as the reference.
She becomes calm and composed.
Maintain the exact same camera composition.
22–30 SECONDS:
The crowd begins the same synchronized chaotic movement from the reference.
The central woman remains calm and controlled in the center while the surrounding crowd moves violently and rhythmically around her.
Match the exact timing, intensity and choreography of the original crowd movement.
The central woman must remain clearly identifiable and visually separated from the crowd.
CAMERA
COPY THE REFERENCE VIDEO'S CAMERA LANGUAGE EXACTLY.
Preserve:
- camera position
- camera height
- camera distance
- lens perspective
- focal length feel
- framing
- depth of field
- camera movement
- push/pull movement
- tracking
- camera shake if present
- zoom
- subject scale
- transitions
- shot duration
- exact cut points
Do NOT create new shots.
Do NOT add new camera movements.
Do NOT change the original video's pacing.
The output should have the same visual rhythm and edit structure as the reference video.
CROWD
Keep the original crowd concept and choreography from the reference video.
Large group of young men surrounding the central woman.
Preserve their:
- positions
- density
- formation
- movements
- synchronized choreography
- timing
- background behavior
Do not replace the crowd with women.
Do not introduce new characters.
Do not allow anyone else to become visually dominant over the main woman.
Only the central character is replaced.
LOCATION
Preserve the same environment and architecture as the reference video.
Keep the same:
- concrete/architectural setting
- stairs
- background structure
- crowd arrangement
- environmental depth
- overall composition
Do not redesign the location.
Do not introduce new buildings or props.
LIGHTING / COLOR
Preserve the reference video's overall cinematic lighting and color grading.
Keep the dark, moody atmosphere.
The acid lime blazer should be the distinctive color of the central character.
Keep natural skin tones.
Realistic shadows and highlights.
No artificial glow.
No haze.
No bloom.
No excessive smoke.
No fantasy lighting.
VISUAL STYLE
Photorealistic cinematic live-action footage.
Natural human motion.
Natural body physics.
Realistic fabric.
Realistic hair.
Realistic skin.
Realistic cigarette smoke.
Realistic lighting.
No CGI appearance.
No animation.
No cartoon.
No plastic skin.
No face distortion.
No duplicated main character.
CRITICAL INSTRUCTION
REFERENCE VIDEO = controls EVERYTHING about the original video's motion.
<image_1> = controls ONLY the identity and appearance of the replacement central woman.
Replace the original central man with the woman from <image_1> while preserving the original video's actions, choreography, camera, timing, pacing, framing and cuts as closely as possible.
The result should look like the original video was filmed with <image_1> instead of the original actor.
ONE MAIN CHARACTER ONLY.
NO IDENTITY DRIFT.
NO EXTRA VERSION OF <image_1>.

Why it works

  • A closing line splits the jobs: the reference video controls everything about motion, and <image_1> controls only identity and appearance.
  • It says not to redesign the choreography because the replacement is a woman, the drift it expects.
  • The new outfit is itemized down to what is absent: no jewellery, watch, belt, bag or phone.
Made with Seedance 2.530s

by @Mayz116910K viewsOriginal post

Five-sentence character swap onto a reference video

Five short sentences: replace the main subject in @ video with the person in @ image, and keep the background, camera motion, timing and lighting.

  • Reference video
  • Character swap
  • Short prompt
  • Reference video

Inputs @ video — the clip to copy · @ image — the new subject

Video prompt383 chars
Replace the main subject in @ video with the subject from @ image. Use @ image as the sole character identity reference. Strictly preserve the subject's appearance and facial features from @ image maintaining the same recognizable identity consistently throughout every frame. Preserve the background, camera motion, timing, and lighting from @ video. Keep everything else unchanged.

Why it works

  • It shows the minimum a swap needs: which input supplies the identity and which supplies everything else.
  • Being this short, it leans on the model and on the reference clip for all the detail.
Prompt source

Ordinary-Day Home Video5

A shaky camcorder clip of an ordinary day that reads as somebody's real memory.

Made with Seedance 2.527s

by @Sheldon0565.3M viewsOriginal post

A summer day in a Seoul neighborhood, as a friend's home video

Thirty seconds of small moments on an older consumer camera: a rooftop drink, an alley walk, a missed basketball shot, a corner shop, a sudden shower he walks into, and a wave before the recording cuts off.

  • Text to video
  • Home video
  • Seoul
  • 30 seconds
Video prompt6,870 chars
Create a 30-second, 1080p ultra-realistic documentary-style personal home video showing an ordinary summer day in the life of a young Korean man. The footage should feel spontaneous, intimate, imperfect, and genuinely observed rather than performed.
MAIN SUBJECT
The same young Korean man in his early 20s throughout the entire video.
Naturally handsome, realistic skin texture, minimal styling, relaxed expression, slightly tired but peaceful eyes.
He has naturally messy medium-length dark hair with a few strands falling over his forehead and very subtle stubble.
He wears a loose washed-black T-shirt, relaxed olive-beige trousers, worn white sneakers, and a simple silver wristwatch.
Keep his face, identity, body proportions, hairstyle, clothing, watch, and overall appearance completely consistent from beginning to end.
LOCATION
A quiet older residential neighborhood in Seoul during a warm summer afternoon.
Narrow concrete alleys, low-rise homes, rooftop terraces, external staircases, potted plants, laundry lines, parked bicycles, utility poles, overhead wires, mature trees, concrete walls, small residential courtyards, and distant city sounds.
The neighborhood should feel authentic and lived-in.
No crowds, tourist attractions, advertisements, recognizable brands, or commercial activity.
CAMERA / VISUAL STYLE
Authentic casual personal-video footage captured with an older consumer digital camera.
Handheld camera operated by a friend walking nearby.
Natural camera shake, imperfect framing, occasional autofocus changes, slight exposure adjustments when moving between sunlight and shade, soft image detail, mild motion blur, subtle digital noise, slightly muted colors, imperfect white balance, and natural compression.
The camera operator occasionally reacts a little late, cuts off part of the subject, or briefly loses focus.
No stabilization, gimbal movement, drone shots, cinematic camera choreography, dramatic lighting, slow motion, modern commercial color grading, or polished cinematography.
The footage should feel like someone simply decided to record his friend during an ordinary day.
00:00–00:05 — MORNING ROOFTOP
He sits casually on a small rooftop beside an old plastic chair.
A cold bottled drink rests beside him.
He looks quietly across the neighborhood while the wind moves his hair and T-shirt.
He takes a sip, notices something happening in the distance, and smiles faintly.
He briefly notices the camera and gives a subtle nod before looking away.
The camera takes a moment to find focus on his face.
00:05–00:10 — WALKING THROUGH THE ALLEY
He gets up and walks downstairs into the neighborhood.
He walks casually through a narrow concrete alley with his hands in his pockets.
He passes parked bicycles, potted plants, laundry hanging from balconies, and old residential walls.
The camera follows several steps behind him.
He occasionally looks back toward the camera but never deliberately poses.
His footsteps remain naturally synchronized with his movement.
00:10–00:14 — SMALL EVERYDAY MOMENT
He notices an old basketball resting near a wall.
He picks it up, casually bounces it twice, then takes a simple shot toward a nearby hoop.
The shot misses.
He laughs quietly, shakes his head, and leaves the ball where he found it.
The camera briefly loses focus during the movement and recovers naturally.
No exaggerated athletic movement.
00:14–00:19 — LOCAL SHOP
He walks to a tiny neighborhood shop and buys a cold drink.
He exchanges a few natural words with the shopkeeper but the conversation is not clearly audible.
He steps outside, opens the bottle, takes a drink, and leans casually against the wall.
He watches bicycles and pedestrians passing in the distance.
The camera remains handheld and slightly imperfect.
00:19–00:23 — SUMMER RAIN
A sudden summer shower begins.
He looks toward the sky with mild surprise.
Instead of immediately running for shelter, he smiles and slowly walks into the rain.
The rain becomes heavier.
His hair becomes wet and falls naturally across his forehead.
He eventually starts running down the alley, laughing genuinely.
He briefly spins around while running, then continues toward a covered walkway.
His clothes become visibly damp.
Maintain realistic rain interaction, wet fabric, wet hair, reflections, and foot contact with the ground.
00:23–00:27 — QUIET MOMENT
He reaches the covered walkway and catches his breath.
Rain falls heavily behind him.
He wipes water from his forehead and looks quietly toward the street.
For a moment, everything becomes still.
He notices the camera again.
He gives a small genuine smile, not a posed expression.
00:27–00:30 — WALKING AWAY
The rain becomes lighter.
He walks away down the wet residential lane.
The camera follows from behind.
Reflections shimmer across the concrete.
He turns his head once, gives a tiny wave toward the camera, smiles, and continues walking.
The camera remains pointed toward the empty street for a brief moment.
At approximately 00:29, the recording abruptly cuts to black mid-motion.
No fade-out.
PHYSICAL REALISM
Maintain believable real-world physics throughout.
Hands, fingers, feet, clothing, hair, rain, bottle, basketball, and background objects must behave naturally.
No extra fingers, fused hands, duplicated limbs, distorted anatomy, floating objects, teleportation, disappearing objects, or sudden transformations.
The bottle remains a separate physical object and never intersects with his face.
The basketball behaves naturally and remains where it lands.
Parked bicycles and background objects remain stationary unless physically moved.
His feet remain properly connected to the ground while walking and running.
Keep the environment and subject consistent between shots.
AUDIO
Natural environmental audio only.
Footsteps on concrete, distant traffic, birds, leaves moving in the wind, bicycles, faint neighborhood conversations, shop sounds, bottle opening, basketball bouncing, rain hitting concrete, water dripping from rooftops, and subtle camera-handling noise.
No music.
No narration.
No soundtrack.
No artificial sound effects.
No spoken dialogue is necessary.
FINAL FEEL
The result should feel like a forgotten personal recording of an ordinary summer day.
Not a commercial.
Not a fashion film.
Not a music video.
Not a professional cinematic production.
The emotional appeal should come from small human moments: sitting alone, wandering through familiar streets, missing a basketball shot, drinking something cold, getting caught in the rain, laughing, and walking home.
Quiet, masculine, youthful, nostalgic, slightly melancholic, warm, spontaneous, and deeply human.
Prioritize natural behavior, consistent identity, believable physics, imperfect handheld framing, authentic environmental details, and the feeling that the camera happened to be there.

Why it works

  • It writes the operator's mistakes in: reacting a little late, cutting off part of the subject, losing focus.
  • The moments are kept small on purpose; the shot misses and he leaves the ball where he found it.
  • The ending is specified as an abrupt cut at about 00:29, mid-motion, with no fade.
Made with Seedance 2.510s

by @RishuaVR3.7M viewsOriginal post

Late morning in an Indonesian alley, early-2000s camcorder look

Ten seconds in four short blocks: a young woman fixes her hair on a terrace step, pets a cat in the alley, sits with iced tea, then smiles at the camera before the recording cuts off.

  • Text to video
  • Home video
  • DV camcorder
  • ASMR audio
Video prompt3,721 chars
Main subject: young Indonesian woman, early 20s, natural everyday appearance, faded mustard-yellow oversized cotton t-shirt, brown high-waisted linen wrap skirt reaching just below the knee, simple brown leather slide sandals, thin gold stud earrings, dark wavy hair in a messy low bun with loose wispy strands framing her face. Realistic warm brown skin texture, minimal natural makeup, warm and approachable demeanor. Maintain consistent identity, clothing, hairstyle, and appearance throughout the entire video.
Location: Authentic Indonesian residential neighborhood (gang) during a calm late morning. Narrow paved alleys, low-rise houses with open front terraces, terracotta roof tiles, potted frangipani and bougainvillea, laundry on bamboo drying racks, parked motorbikes along plastered walls, overhead utility wires, banana and coconut trees casting dappled moving shadows, quiet tropical residential atmosphere. No modern stores, billboards, cafés, or crowds.
Visual Style: Ultra-realistic documentary realism. Genuine candid behavior. Natural body language. Unscripted slice-of-life feeling. Strong environmental authenticity. Rich real-world details and believable human motion.
Camera Style: Early-2000s consumer DV camcorder aesthetic. Friend casually recording everyday moments. Heavy handheld shake, imperfect framing, frequent autofocus hunting, lens breathing, exposure pumping when moving between harsh tropical sun and deep alley shade, occasional motion blur, subtle rolling shutter, mild digital compression artifacts, warm faded colors, soft contrast, slight sensor noise. No stabilization. No cinematic camera moves. No modern color grading.
00:00–00:02
Outside a small house entrance with an open terrace. She sits on the low terrace step adjusting her loose hair bun with both hands raised. A light warm breeze moves loose strands of hair. She smiles naturally while the camera struggles to lock focus.
00:02–00:05
The camera follows her into a narrow alley lined with potted plants and plastered walls. She notices a neighborhood cat resting on a warm concrete ledge, crouches down, and gently pets it. Framing drifts off-center as the operator tries to keep up. Morning sunlight flickers through banana leaves overhead.
00:05–00:08
Small front terrace with a glass of sweet iced tea on a wooden stool. She sits relaxed on a plastic chair watching the quiet gang, casually brushing hair behind her ear. A motorbike passes slowly in the distant background. Handheld side angle with natural camera drift.
00:08–00:10
She turns toward the camera with a warm, genuine smile, gives a gentle nod of acknowledgement, then turns to walk down the tree-lined alley holding her iced tea. Recording cuts abruptly to black mid-motion as if the camcorder was switched off.
Audio: Hyper-detailed environmental ASMR binaural audio — crisp micro-sounds captured up close with high sensitivity. Crisp rustle of cotton fabric and linen as she moves, deep rhythmic cat purring vibrating close to the microphone, ice cubes gently clinking against the glass condensation, soft scuff of leather sandals on textured concrete pavement, dry banana leaves brushing together in the warm breeze, rhythmic faint sound of a sapu lidi broom sweeping fallen leaves nearby, water dripping into a terracotta gentong, distant gentle crowing rooster, faint muffled low-frequency hum of a passing motorbike down the alley. Pure natural foley textures, zero music, zero studio effects, zero narration, zero speech.
Goal: Authentic Indonesian neighborhood life captured like a forgotten home video from the early 2000s — candid, imperfect, realistic, warm, and deeply believable, with immersive environmental ASMR soundscape.

Why it works

  • The audio list is close-up and local: cat purring, ice against the glass, a sapu lidi broom sweeping, water dripping into a gentong.
  • Camera faults are listed one by one (autofocus hunting, lens breathing, exposure pumping between sun and shade) instead of 'make it look old'.
Made with Seedance 2.015s

by @Goodmanprotocol112.5K viewsOriginal post

Beach day with friends as rapid phone-memory jump cuts

A woman from your reference photo laughs, splashes, snacks and builds a sandcastle with friends across six 2.5-second blocks, cut every one to two seconds like scrolling through phone clips.

  • Image input
  • Jump cuts
  • Phone footage
  • Seedance 2.0

Inputs Reference photo of the main woman

Video prompt2,868 chars · 6 beats
Super casual real smartphone home video footage of a sunny beach day outing with friends. Natural mobile phone camera recording with slight authentic handheld shake, normal frame rate with smooth natural motion, rapid-fire montage with constant quick jump cuts every 1–2 seconds like scrolling through phone memories. Unpolished authentic phone recording of a mixed group laughing, playing in the sand, snacking and clicking pictures. Pure raw home video feel with no cinematic polish or heavy effects.
Use the provided reference photo as the STRICT ONLY visual reference for the main woman. Maintain her exact appearance with zero deviation: [her described features]. Generate a mixed group of friends of all ages around her on the beach.
0–2.5s:
Shaky handheld rapid cuts of the main woman laughing with friends near the shoreline. Wind gently moves her hair. Quick flashes of feet in warm sand and waves washing over them.
2.5–5s:
Abrupt jump cuts showing a close-up of her smiling with sunglasses pushed up onto her head, followed by the group splashing water and taking casual selfies together.
5–7.5s:
Fast handheld footage of her sitting on a beach towel chatting animatedly with friends. Snacks and cold drinks are scattered around. Natural sunlight creates subtle lens reflections and realistic exposure changes.
7.5–10s:
Quick cut to a close-up of her giving a big warm smile toward the camera and waving playfully. The footage immediately jumps to the group building a small sandcastle together while laughing naturally.
10–12.5s:
Abrupt edit showing everyone gathered beneath a beach umbrella, casually passing chips and cold drinks around. Friends cheerfully toast with cans while talking over one another.
12.5–15s:
Final rapid transition showing the main woman relaxed and sitting among her friends, softly smiling while looking toward the ocean. The video ends naturally with a gentle phone sway as if someone simply stopped recording.
VISUAL STYLE:
- Natural smartphone video quality.
- Slight realistic handheld shake.
- Smooth normal frame-rate motion.
- Authentic casual interactions and physics.
- Realistic sunlight and exposure adaptation.
- Stable main character consistency.
- Unpolished home phone recording aesthetics.
- No professional stabilization.
- No cinematic color grading.
- No beauty filters.
- No artificial effects.
- No AI artifacts or glitches.
IMPORTANT GENERATION REQUIREMENTS:
- Consistent identity throughout the video.
- Realistic human anatomy and hand interactions.
- Natural walking and body movement.
- Authentic beach ambience and environmental audio.
- Physically correct lighting and shadows.
- Rapid memory-style jump cuts every 1–2 seconds.
- No duplicated people or objects.
- No facial distortions.
- No impossible movements.
- Preserve the casual smartphone home-video feeling from beginning to end.

Why it works

  • The cut rhythm is a number, a jump cut every 1–2 seconds, and it is repeated in the requirements list.
  • Each block names two or three quick moments, which gives the fast cutting something to cut between.
  • The placeholder '[her described features]' is left in; replace it with a description of your reference person.
Made with Seedance 2.541s

by @MrLarus6.1K viewsOriginal post

Rainy-season afternoon in a Japanese neighborhood, in fragments

Seven separate MiniDV fragments of one afternoon (shoes on at the door, a windy walk, a drink under clouds, a stray cat, getting a bicycle ready, riding it, rain on a porch), with the weather worsening from one to the next.

  • Image input
  • Home video
  • Fragments
  • Rainy season

Inputs One character reference image

Video prompt2,964 chars
Create an ultra-realistic candid home-video sequence using one character reference image.
SUBJECT:
A naturally pretty young Japanese woman in her early 20s. Keep her face, body, hairstyle, outfit and appearance perfectly consistent throughout. Natural fair skin, realistic skin texture, minimal makeup, slim but believable proportions, relaxed everyday personality.
SETTING:
A quiet Japanese residential neighborhood during the rainy season narrow streets, small homes, porches, plants, bicycles, utility poles, overhead wires and ordinary lived-in details.
VISUAL STYLE:
Early-2000s consumer MiniDV home-video footage. Imperfect handheld operation, slight camera shake, awkward framing, occasional reframing, autofocus hunting, exposure changes, subtle motion blur, faded colors, soft contrast and light digital compression. No stabilization, cinematic camera moves or modern color grading.
The video should feel like several separate fragments captured during the same ordinary afternoon, not one continuous shot.
SEQUENCE:
At the entrance of her home, she sits down, puts on her shoes, adjusts her bag and gets ready to go out.
Cut to her walking through the residential street. A cool breeze gradually becomes stronger, naturally blowing her hair across her face as the sky darkens.
Cut to another quiet moment. She drinks from a small bottle and casually looks up at the cloudy sky. A distant low rumble of thunder can be heard.
Cut to her crouching beside a stray cat near some plants. She gently reaches toward it, gives it a little food, pets it briefly and smiles while keeping her attention on the cat. Autofocus shifts imperfectly between her face, hand and the animal.
Cut to another fragment in the damp alley. She stands beside a simple city bicycle, naturally adjusts or prepares it while the weather continues to change.
Cut to her riding the bicycle slowly through the residential street. Wet pavement reflects the grey sky while ordinary residents occasionally pass in the background.
Final fragment: the rain has become heavier. She sits quietly under a residential porch with slightly wet hair, gently dries or brushes back her damp hair and watches the rain outside. Plants move in the wind, rain falls steadily beyond the shelter, and distant thunder mixes with the neighborhood ambience. The recording ends abruptly like an old camcorder being switched off.
AUDIO:
Only authentic location sound light wind gradually becoming stronger, rustling leaves, distant traffic, footsteps, bicycle sounds, neighborhood ambience, cat sounds, distant thunder, increasing rainfall and dripping water. No music, narration or cinematic sound design.
GOAL:
Make it feel like forgotten Japanese family footage from the early 2000s intimate, spontaneous, imperfect, warm, mundane and deeply believable. Prioritize natural human motion, varied everyday moments, changing weather, environmental detail and consistent identity over cinematic beauty.

Why it works

  • It asks for 'several separate fragments', not one continuous shot, which suits the home-video look.
  • Weather is the thread through the cuts: a breeze, a darkening sky, thunder, wet pavement, then heavier rain.
  • No length or timestamps are given, so how long each fragment runs is up to the model.
Prompt source
Made with Seedance 2.530s

by @j_smeaton992K viewsOriginal post

Watching tennis courtside, filmed on a DV camcorder

A woman in a straw hat watches a grass-court match for 30 seconds, reacting to the play, then notices the camera at 00:24 and gives a shy smile before turning back.

  • Text to video
  • Home video
  • DV camcorder
  • Tennis
Video prompt2,027 chars
Create a 30-second ultra-realistic early-2000s consumer DV home-video recording at the Eastbourne tennis tournament.
Main subject: the exact same 25-year-old Korean woman throughout the entire video. Preserve her identity, facial features, skin tone, age, hairstyle, long dark-brown hair, beige wide-brim straw hat with a dark-brown band, light-green off-shoulder crossover pleated dress, pearl earrings, and thin gold bracelet. No changes to her appearance or clothing.
She sits near the grass court watching a live tennis match. Her behavior is completely natural and spontaneous: she follows the ball, leans forward with excitement, smiles, laughs softly, claps once or twice, fixes windblown hair, adjusts her dress, and casually touches her hat. At 00:24–00:27 she notices the camera, looks surprised for a moment, smiles shyly, touches her hat as if greeting a friend, then turns back to the match.
The setting is a warm sunny afternoon with occasional clouds, realistic spectators, grass court, fencing, and distant players. Background remains slightly out of focus.
Make it feel like a friend suddenly recording her with a consumer DV camcorder in the early 2000s — NOT cinematic, commercial, smartphone, or professionally filmed.
Include authentic handheld shake, imperfect framing, autofocus hunting, brief focus shifts, auto-exposure changes, slight overexposure, motion blur, rolling-shutter artifacts, lens breathing, digital compression, sensor noise, faded colors, soft contrast, and authentic DV texture.
Camera naturally shifts between slightly off-center front, medium close-up, wider shot, and side profile. Never use a gimbal or perfect stabilization.
Audio: authentic tennis-court ambience only — wind, racket impacts, crowd murmurs, occasional applause, seat creaks and subtle environmental sounds. No music, narration, voiceover, artificial sound effects, subtitles, transitions, logos, or fade-out.
End abruptly at 00:30 while she smiles briefly and turns her attention back to the tennis match.

Why it works

  • Her behavior is a list of small actions (follows the ball, claps once or twice, fixes windblown hair) instead of one pose.
  • Only one moment is timed, 00:24–00:27 when she notices the camera, and the rest is left loose.
  • It names a real event, the Eastbourne tournament; use a generic grass court if you would rather not.

Found Footage4

A tourist's or hiker's camera roll that turns into something else halfway through.

Made with Seedance 2.029s

by @minchoi81.2K viewsOriginal post

Tourist camera roll from a dinosaur island preserve

Seven two-second clips from one day, morning to dusk: through the gates in a jeep, giants on the savanna, a river crossing, a predator behind the fence, a stampede, a horned giant at golden hour and a distant roar.

  • Image input
  • Found footage
  • Seven clips
  • Seedance 2.0

Inputs The attached gate photo (Shot 1)

Video prompt4,275 chars
Format: Found-footage vacation video, 15 seconds, seven shots with hard cuts — a tourist's camera roll from one full day, morning to dusk.
Image reference: Use the attached gate photo as the exact visual reference for Shot 1 — match its architecture, proportions, torch placement, lighting, and signage precisely.
Main subject: Young woman, blond hair in a loose low bun shedding strands, oversized faded olive t-shirt, denim shorts, white canvas sneakers, disposable-camera strap on her wrist. Realistic skin texture, minimal makeup, wide-eyed tourist joy. Maintain perfect visual consistency — identity, hair, clothing, proportions — across every shot and location.
Location: A vast island dinosaur preserve. Jungle mountains, open savanna paddocks, wooden observation structures, tall electrified perimeter fences, dirt tour roads. Dinosaurs are photorealistic animals — muted natural hides, birdlike motion, visible breathing weight. Filmed like wildlife, never like monsters.
Camera style: Mid-range smartphone, handheld by her friend. Heavy shake, autofocus hunting, motion blur on pans, compression grain, exposure pumping. Each cut is abrupt and mid-motion, like real unedited camera-roll clips.
00:00–00:03 — THE GATES (morning, filmed from inside a moving open-top tour jeep). The camera films forward over the seats as the jeep approaches the colossal timber gates from the reference image — flared stone-and-wood towers, flaming torches, jungle pressing in on both sides, the road still wet from morning rain. The doors groan open. She spins around from the front seat into frame, gripping the roll bar, and half-shouts over the engine: "Oh my god, this is wild! I can't believe we're actually here!" — as the jeep passes under the archway into shadow and out again.
00:03–00:05 — SAVANNA (late morning, wide shot from a raised wooden platform). She stands small at the rail against an open plain where a herd of long-necked giants wades through heat haze, one rearing slowly against the treeline. The camera zooms in shakily, overshoots, corrects.
00:05–00:07 — RIVER CROSSING (midday, low angle through the jeep's open side). Past her shoulder as the jeep fords a shallow river: a crested duck-billed dinosaur drinks twenty meters away, lifting its head to watch them pass, water dripping from its jaw. She raises her disposable camera; the phone catches its tiny click.
00:07–00:09 — JUNGLE FENCE (afternoon, handheld walking shot). Dim under the canopy beside a tall humming fence. She peers through the wire — and a small crested predator peers back from the ferns, head tilting in quick birdlike jerks. She backs up a slow half-step, whispering "okay... hi." Autofocus argues between wire and eye.
00:09–00:11 — THE HERD RUNS (late afternoon, chaotic pan from the jeep at speed). A flock of small striped dinosaurs floods across and around the vehicle. The camera whips left-right, blurred bodies streaking past, her whooping off-mic, dust on the lens.
00:11–00:13 — QUIET GIANT (golden hour, static shot at a wooden overlook). The frame finally settles: silhouetted against low sun, she stands at a rail as an immense horned dinosaur grazes just below, close enough to hear grass tearing. She isn't laughing anymore — just watching.
00:13–00:15 — DUSK ROAR (blue hour, half-framed and accidental). Filmed as the phone was being lowered, frame tilted: a colossal, deep bellow rolls in from the darkening jungle far off. Everyone freezes and turns toward it. Cut to black mid-turn.
Audio: Ambient only, shifting per location — jeep engine and gate groan, her excited line over the motor, savanna wind, river wading, fence hum, thundering small feet, grass tearing, and the single distant bellow that swallows everything. No music. No narration.
Style & quality boosters: Raw consumer camera-roll aesthetic, abrupt mid-motion cuts, heavy handheld instability, natural motion blur, golden-hour lens flare, coherent animal mass and physics — ground tremor, water displacement, dust. No text overlays, no watermarks.
Goal: One unforgettable day inside the world's most famous dinosaur preserve, told through seven imperfect tourist clips — opening on the gates everyone recognizes, and closing on the sound that reminds them what lives beyond the fences.

Why it works

  • Each shot carries a time of day as well as a place, so the light moves across the reel.
  • Dinosaurs are to be filmed 'like wildlife, never like monsters', which keeps it tourist footage.
  • The gate comes from your photo, and the prompt calls it the gate 'everyone recognizes', aiming at a famous film without naming it.
Made with Seedance 2.530s

by @YourAlphaMom8.2K viewsOriginal post

Ghost crawls out of the TV, then gets a makeover and a wedding

Friends flee a ghost crawling out of the television, one man stays, and the found-footage scare turns into a montage of cleanup, park dates, a proposal and a small wedding.

  • Text to video
  • Horror to romance
  • Montage
  • 30 seconds
Video prompt5,477 chars
Create a 30-second, 1080p ultra-realistic handheld home-video style clip that begins like a frightening paranormal incident and gradually turns into a bizarre wholesome romantic montage.
OVERALL STYLE
The footage must look like real consumer-camera video, not polished cinema.
Use a realistic handheld recording aesthetic: natural hand shake, imperfect framing, occasional autofocus delay, slight exposure shifts, soft digital detail, mild motion blur, faint noise, realistic indoor lighting, and believable human reactions.
The first part should feel like spontaneous scary footage captured in a living room at night.
The later part should become a quick but clear montage of sweet everyday moments, still grounded in realism.
MAIN CHARACTERS
A group of ordinary young adults watching TV in a living room at night.
One young man stays central after the scare. He is calm, slightly confused, but kind.
A creepy ghost-like young woman emerges from the TV. She has very pale skin, long wet black hair, a tired eerie expression, and a simple white dress or nightgown.
She should look unsettling at first, but still human and realistic — not monstrous, not gory, not decomposed.
As the video continues, she should remain the exact same woman, only gradually becoming cleaner, fresher, and more presentable. Do not transform her into a completely different person.
IMPORTANT TONE
Start tense and scary.
Then switch into absurd kindness and sweet humor.
The humor should come from the contrast between:
a terrifying ghost entering the room, and a guy reacting like she just badly needs help and self-care.
TIMELINE
00:00–00:05 — LIVING ROOM / HORROR START
A few friends are sitting in a dim living room at night watching a scary movie on an old television.
The TV is flickering. The room is lit mostly by the TV glow and a weak lamp.
Suddenly, a creepy pale ghost girl with long wet black hair starts crawling out of the television into the room.
The people panic, scream, and run away in fear.
The handheld camera shakes hard and struggles to keep the action framed.
00:05–00:07 — THE GUY STAYS
One young man remains in the room, staring at her in shock.
Instead of screaming, he looks concerned and says something like:
“Oh no… we need to fix this.”
The ghost girl pauses and looks confused.
00:07–00:11 — QUICK CLEANUP / MAKEOVER START
Fast but readable montage.
He gestures for her to go clean up.
Show short realistic clips of:
her standing awkwardly in a bathroom doorway,
brushing her teeth,
washing up / shower steam / rinsing her hair, him handing her a towel,
her hair looking cleaner and less messy.
Keep this simple and visually clear.
Do not overcomplicate the bathroom actions.
00:11–00:15 — GETTING READY
More quick makeover moments:
he helps dry or gently brush her hair,
she now looks neater, still pale, still clearly the same ghost girl,
very light makeup or a tidier appearance, a cleaner casual outfit or a more presentable version of her original look.
She should still feel slightly strange and deadpan, just less terrifying.
00:15–00:23 — WHOLESOME MONTAGE
A sweet, funny montage of them spending time together in public:
walking through a park,
sitting on a bench,
eating ice cream or sharing a snack,
taking a selfie,
feeding birds or ducks,
laughing awkwardly,
one small cute moment where she tries to imitate normal human behavior.
Everything should feel warm, a little absurd, but still realistic and grounded.
00:23–00:27 — PROPOSAL
In a simple, heartfelt outdoor moment, the guy proposes to her.
He shows a ring.
She looks surprised, then gives a small quiet smile and accepts.
Keep it sweet, not exaggerated.
00:27–00:30 — WEDDING ENDING
Cut to a small wedding ceremony.
They are dressed for a wedding, standing together while guests smile and clap.
The ghost girl looks calm and beautiful but still slightly eerie in a charming way.
For the final beat, add one subtle joke:
a nearby TV screen flickers in the background, and she glances at it for a second.
End there.
VISUAL / PHYSICAL REALISM
Maintain believable anatomy and motion throughout.
No broken limbs, no extra fingers, no floating objects, no teleporting, no disappearing clothing, no sudden identity changes.
Keep the ghost girl’s face and identity consistent from scene to scene.
Keep the man’s identity consistent as well.
All interactions must look physically plausible.
CAMERA / EDITING
The clip should feel like a real recorded video with cuts, not a polished movie trailer.
The opening living room scare should feel especially raw and chaotic.
The montage can be faster, but each shot must still be readable and visually clean.
Do not use slow motion.
Do not use cinematic crane shots, drone shots, or glossy commercial lighting.
AUDIO
Use only natural scene audio:
TV flicker / room ambience,
frightened reactions,
bathroom ambience,
outdoor park ambience,
wedding ambience.
No dramatic score is necessary.
No narration.
IMPORTANT RESTRICTIONS
No gore
No blood
No body horror
No graphic violence
No grotesque facial distortions
No copyrighted logos or recognizable movie footage on-screen
Do not make the ghost girl cartoonish
Do not make the romance mocking or cruel
Keep the humor warm, weird, and unexpectedly wholesome
FINAL FEEL
The final result should feel like:
a terrifying found-footage setup that turns into a strangely sweet love story.
It should be funny, memorable, realistic-looking, and emotionally warm in the second half.

Why it works

  • The joke is spelled out in the prompt: a terrifying ghost met by a guy who treats her as someone who needs help and self-care.
  • She has to stay 'the exact same woman', only cleaner, so the makeover does not swap her for someone else.
  • Each montage stretch lists several simple actions and asks for them to stay readable.
Made with Seedance 2.530s

by @Diplomeme2.9K viewsOriginal post

Hikers and a figure that keeps getting closer in the forest

Ten three-second blocks on a forest trail: a figure between the trees appears, vanishes and reappears closer each time, until it steps out behind one friend and the camera drops.

  • Text to video
  • Found footage
  • Slow reveal
  • No effects
Video prompt5,714 chars
30 seconds | 16:9 | live-action supernatural horror
Fictional dramatization. Realistic handheld documentary footage.
SETTING
A dense remote forest hiking trail surrounded by tall trees, thick undergrowth, fallen branches and uneven natural terrain.
Time: late afternoon.
Weather: overcast with occasional sunlight filtering through the canopy.
Natural forest ambience, light wind, distant birds and insects.
CHARACTERS
Two young adult friends on a casual forest hike.
Maintain consistent faces, clothing, hairstyles, backpacks and physical appearance throughout the sequence.
Human behaviour remains casual, natural and restrained.
STORY
00–03s — ENTERING THE FOREST
Handheld camera follows the two friends entering the forest trail.
One friend walks ahead while the camera operator follows casually behind.
Natural conversation, footsteps and branches brushing against the camera.
03–06s — SOMETHING IN THE BACKGROUND
The camera briefly turns toward the forest behind them.
For less than a second, a distant human figure is visible between the trees.
The camera immediately turns forward again.
Neither friend notices.
06–09s — KEEP WALKING
The friends continue along the trail.
Autofocus hunts between nearby leaves and the friend ahead.
Natural sunlight flickers through the trees.
The camera operator briefly looks back again.
Nothing is there.
09–12s — FIRST REALIZATION
The friend ahead suddenly stops.
He looks toward the trees.
The camera quickly moves closer.
He asks quietly:
“Did you hear that?”
Both listen.
12–15s — THE FIGURE
The camera slowly turns toward the sound.
A human figure stands far between the trees.
Completely motionless.
The camera struggles to focus.
The operator zooms slightly.
The figure disappears behind a tree.
15–18s — IT MOVED
The camera swings toward the friend.
He looks confused.
The camera immediately turns back.
The figure is now standing significantly closer.
The operator takes an involuntary step backward.
18–21s — RUNNING
The friend says:
“Let's go.”
Both begin walking quickly back down the trail.
The camera shakes heavily while following.
Branches hit the lens.
Footsteps become faster.
21–24s — THE EMPTY TRAIL
The camera suddenly stops.
The trail behind them appears completely empty.
No figure.
No movement.
The forest is unnaturally quiet.
24–27s — BEHIND HIM
The camera turns toward the friend.
He is standing several meters ahead.
Behind him, deep between the trees, the figure is visible again.
This time it is much closer.
The friend still doesn't notice.
27–30s — FINAL REVEAL
The camera operator freezes.
The figure slowly steps out from behind the trees.
The friend turns toward the camera after hearing the operator's breathing.
The camera abruptly drops toward the ground.
Cut to black.
CAMERA
Authentic consumer handheld camera footage.
Heavy but believable handheld movement.
Natural walking shake.
Imperfect framing.
Frequent small reframing adjustments.
Autofocus hunting.
Natural lens breathing.
Exposure changes beneath the forest canopy.
Occasional motion blur.
Realistic operator reaction.
No stabilization.
No cinematic gimbal movement.
No impossible camera transitions.
LIGHTING
Lighting originates entirely from the natural environment.
Soft overcast daylight with occasional sunlight passing through the trees.
Natural exposure changes when moving between open areas and dense forest.
Realistic shadows.
Natural highlight roll-off.
No artificial rim lighting.
No supernatural glow.
PHYSICS
Real gravity and momentum.
Feet maintain contact with uneven ground.
Branches move when touched.
Leaves respond naturally to footsteps and wind.
Clothing and backpacks move with the characters.
The figure remains physically grounded between the trees.
No floating objects.
No impossible movement.
No supernatural visual effects.
HUMAN PERFORMANCE
Natural behaviour between close friends.
Casual conversation.
Normal walking pace.
Natural breathing.
Small posture adjustments.
Genuine confusion.
Fear gradually replaces curiosity.
No exaggerated screaming.
No theatrical horror acting.
No heroic posing.
AUDIO
Natural handheld-recording audio.
Footsteps on dirt and leaves.
Branches brushing against clothing.
Breathing.
Backpack movement.
Casual conversation.
Birds.
Insects.
Wind through trees.
Environmental sound gradually becomes quieter whenever the figure appears.
No music.
No jump-scare sound effects.
No cinematic bass drops.
No oversized Hollywood trailer effects.
VISUAL CHARACTER
Photorealistic live-action footage.
Authentic amateur documentary character.
Natural skin texture.
Realistic clothing and materials.
Natural forest textures.
Subtle sensor noise.
Realistic motion blur.
Imperfect handheld framing.
Consumer-camera image quality.
The supernatural element must remain visually ambiguous and physically realistic.
Avoid:
video-game aesthetics
CGI appearance
excessive HDR
oversaturated colors
glowing eyes
transparent ghosts
floating figures
monster faces
supernatural visual effects
impossible physics
perfect camera movement
perfectly choreographed reactions
beauty-filter faces
graphic gore
cheap jump scares
CONTINUITY
Maintain character identity, clothing, backpacks, forest geography and lighting throughout the sequence.
Preserve the same trail and recognizable trees between shots.
Maintain consistent screen direction and spatial orientation.
The figure must be the same person in every appearance.
The figure progressively appears closer throughout the sequence.
Every action begins from the physical state established by the previous shot.
The final moment should feel like genuine accidental footage rather than a staged horror scene.

Why it works

  • The scare is distance alone: the figure is written far, then closer, then 'much closer', with no effects.
  • Forest sound quietens whenever the figure appears, and music and jump-scare sounds are ruled out.
  • The avoid list removes the usual ghost tells: glowing eyes, transparent figures, floating, monster faces.
Made with Seedance 2.010s

by @jerrod_lew2.7K viewsOriginal post

Found footage in a dark forest, in one line

The shortest prompt here: a format, a viewpoint and a place, and the model decides the rest.

  • Text to video
  • One line
  • Found footage
  • Seedance 2.0
Video prompt49 chars
Found footage, POV shot, exploring a dark forest.

Why it works

  • It is a baseline to compare against: with no camera faults, story or sound written, everything on screen is the model's choice.
  • Run it before writing a longer found-footage prompt to see what the model adds on its own.
How to prompt

How to make your own version of a viral prompt

Most clips in this library are built the same way: the inputs the creator brought, a fixed run of shots and time blocks, a sound line, and one lead you can swap. Here is that structure as a template you can copy, block by block.

Template802 chars
[INPUTS] @Image 1: the lead, as a photo or a character sheet. @Video 1: the reference clip, when the trend copies one.
[LEAD] A woman in her 30s with short black hair, a denim jacket and white sneakers, with the same face and outfit in every shot.
[0–5s] Wide shot: the lead walks into a crowded mall food court, handheld camera.
[5–10s] Everyone around her freezes mid-motion; she keeps walking as the camera circles her.
[10–15s] Close-up: she snaps her fingers and the room moves again.
[SWAP] Replace the person on the left in @Video 1 with the person in @Image 1. Keep the motion, the camera, the timing and the cuts of @Video 1.
[SOUND] Food-court chatter, then sudden silence and a single heartbeat before the noise rushes back. No BGM.
[LEAVE OUT] No subtitles. No on-screen text. No logos.
  1. 01

    Check what the card needs

    Every card says what its creator brought: text only, a photo for the first frame, a character sheet, or a reference clip. Bring the same kind of input. A trend built on a reference clip will not come out the same from text alone.

  2. 02

    Keep the creator's structure, change the lead

    Leave the shots, the time blocks, the camera directions and the sound line as the creator wrote them, and change only who is in the clip, how they look and where they are. The structure is what made the original work.

  3. 03

    One action and one camera move per time block

    Split the clip into time blocks and give each block one action and one camera move. The longer prompts in this library are built this way, and Seedance 2.5 renders up to 30 seconds in one take.

  4. 04

    For a swap trend, say who replaces whom

    Trends like Hotel Lobby AI Rap Duo and STORM Clone Dance keep a real performance and put a new face in it. Add your photo and the reference clip, then write which person in the clip your photo replaces, the way the swap cards in this library do.

  5. 05

    Write the sound, then what to leave out

    Seedance 2.5 renders sound with the picture, so name what we hear in each block, and write "No BGM" if you will add your own music. End on what must not appear.

  6. 06

    Test at 480p first

    On Seedance 2.5 a second at 480p is 25 credits against 55 at 720p, so check the motion and the timing in a 480p take first, and spend the higher resolution on the take you keep.

Why Pick Our Viral AI Video Prompts Library

@auqibhabib

Full Creator Prompts Free to Copy

Browse 38 viral AI video prompts word for word in their original language with no paywall, no DM, and no comments needed. Our viral AI video prompts open in full text so you can click the Copy button and paste them anywhere.

Copy Free Prompts
@YourAlphaMom
@AIwithkhanOriginal post

Original Clips with Creator Credits

See the exact clip beside each prompt before you spend a single credit. Our viral AI video prompts credit every creator with their @handle, link directly to the original post or reply, and show the real result.

See Trending Clips
Seedance 2.015sImage input

Exact Recipe Tags on Every Card

Check the model, clip length, and needed assets before you start creating. Every card in our viral AI video prompts tags whether you need text only, a photo, a character sheet, or a reference clip, along with a note on why it works.

Check Recipe Tags

Use Viral AI Video Prompts for Every Type of Creator

Short-Form Creators

Catch Every Trend on Day One

Jump on new trends the day you see them on TikTok, Reels, or Shorts. Bring your idea or a few photos to get a 9:16 clip of up to 30 seconds with sound, or explore our GTA video prompts to spark your next post.

Create Your Clip Free
@RishuaVR
@Just_sharon7
Friends & Pet Owners

Star Your People and Pets

Upload photos of your favorite people or pets to drop them straight into viral memes. Turn your cats into a rap duo with Hotel Lobby AI or feature the family dog inside a Zombie Match Cut.

Star Your Pet Free
Faceless Channels

Run Consistent Weekly Formats

Bring a fresh theme each week to publish clips in formats that already pulled millions of views. Build your audience with recurring clips using Found Footage, Ordinary-Day Home Video, FPV Fly-Through, or Zombie Outbreak.

Start Your Series Free
@Sheldon056
@umesh_ai
Brands & Marketers

Put Products Inside Viral Formats

Upload a product photo to produce eye-catching vertical or horizontal clips with sound. Reveal your item inside a dramatic Time Freeze or give shoppers a virtual tour through an FPV Fly-Through.

Promote Your Product Free

How to Make Videos with Our Viral AI Video Prompts

Step 1

Pick a Trend

Pick a card by its trend or its clip to see what it needs, from text only to a photo or a reference clip.

Step 2

Customize the Prompt

Press Use to load the prompt into our AI video maker, swap in your own lead, add your photo, character sheet, or reference clip, and choose 9:16 or 16:9, your resolution, and a length up to 30 seconds.

Step 3

Generate Your Video

Generate on Seedance 2.5 and download your MP4 with sound.

Try Seedance 2.5 for Free
FAQ

Questions about viral AI video prompts

What are viral AI video prompts?

They are the exact text a creator gives an AI video model to make a clip that takes off. Our viral AI video prompts sit right next to the creator's clip, their @handle, a link to the original post, the model they used, and the inputs you bring. A strong video prompt names the subject, the action, the scene, the visual style, the camera move, and the sound. You can structure longer scenes in time blocks using our Seedance 2.5 prompt guide.

What are the trending AI video prompts right now?

The newest trends come first in our library of 38 prompts across 9 trends. You can run Zombie Match Cut (4), Zombie Outbreak (6), Time Freeze (4), STORM Clone Dance (4), Hotel Lobby AI Rap Duo (3), Ordinary-Day Home Video (5), Found Footage (4), Magic Pen (3), and FPV Fly-Through (5). Pick any card to copy the prompt and make your clip.

Are these viral AI video prompts free?

Yes. Every prompt is free to view and copy, with no comment, direct message, or payment required. Sign up free with no credit card to get 40 starter credits and run prompts in our AI video maker. You can check Seadanse pricing for paid plans that export your videos with no watermark.

Which AI made these viral videos?

Seedance 2.5 is the model creators name most often on this wave. Our Seedance 2.5 AI video maker runs this ByteDance Seed team model, launched on July 31, 2026, as announced in ByteDance's Seedance 2.5 launch post. It generates clips up to 30 seconds with sound and supports up to 50 references. You can also try our earlier Seedance 2.0 AI video maker.

How do I make those viral AI videos?

Find the trend card you want, see what it needs, and press Use to load the prompt into our AI video maker on this page. Swap in your lead, add your photo or reference clip, and pick 9:16 or 16:9. Set your length up to 30 seconds, then generate and download your MP4 with sound.

Can I put myself, my friends or my pet in a viral AI video?

Yes. Add a photo as your first frame through our image to video AI, or edit a clip you upload using our video to video AI. The AI video maker on this page takes up to 30 images, 10 videos, and 10 audio clips per generation, tagged by number like @Image 1. Swap-yourself-in trends pair your photo with the reference clip, and each card shows you what to upload.

How do I find the prompt behind a viral AI video I saw?

Look up the trend name in our library to open the matching card. Creators often post the prompt in the first reply under their clip, so each card links directly to that reply and the original post. The card also lists the exact model and all the input files you need.

How do I get a result that looks like the original?

Use the model named on the card and bring the required inputs, such as a photo, a character sheet, or a reference clip. Keep the creator's shots and time blocks from our Seedance 2.5 prompt guide, changing only the lead. Test your prompt at 480p first before you generate your final version at 720p or 1080p.

Do these prompts work in Kling, Veo3 or other AI video tools?

Yes. Every prompt is plain text, so you can paste it into any AI video tool. For the closest match, run it on the model named on the card. Our AI Video Creator runs Kling O3, Veo3, Hailuo 3, and Seedance 2.5 in one composer on the same credits.

How long can a viral AI video be, and what size works for TikTok, Reels and Shorts?

You get up to 30 seconds per generation on Seedance 2.5. The 9:16 aspect ratio fits TikTok, Reels, and Shorts, and 16:9, 21:9, 4:3, 1:1, and 3:4 are also ready to use. Generate your clip at 480p, 720p, or 1080p and download your MP4 with the sound in it.

How much does it cost to make one?

Seedance 2.5 on Seadanse costs 25 credits per second at 480p, 55 at 720p, and 98 at 1080p, so a 5-second 480p clip costs 125 credits. Plan your projects with our Seedance 2.5 cost calculator. The exact cost shows beside Generate before you submit, a failed render returns its credits automatically, and signing up adds 40 starter credits.

Who made these prompts, and are the creators credited?

Every prompt was published by its own creator on a public platform, and every card carries their @handle and a link to the original post. The prompt uses the creator's exact words, while Seadanse writes the card title and the note on why it works. A creator who wants a card corrected or taken down opens an issue on GitHub or emails support@seadanse.com.

Is there a GitHub version of these prompts?

Yes. The entire library is published as a public repository at awesome-ai-video-effect-prompts on GitHub. You can browse all 38 prompts across our 9 viral trends and study the text structure for your own workflow.

SourceCollected from public posts on X and mirrored to our own storage so the clips keep playing. Every card credits its creator and links the original post, and the reply that holds the prompt when it was posted there. Counts are as of collection. Prompts are the creators' own words; titles and notes are ours. A creator who wants a card corrected or taken down can open an issue on the GitHub copy of this library or email support@seadanse.com. awesome-ai-video-effect-prompts on GitHub

Make Videos with Our Viral AI Video Prompts Free

Make your own version of a viral clip with sound that is ready to post, or discover more Seedance 2.5 prompts for your channel.

Make Viral Videos Free