Surreal Tokyo Dessert City

Style & Mood: AAA game cutscene aesthetic, live-action realism - real human actors, visible skin pores, real cloth and hair physics, soaked fabric clinging and dripping, no animation or cartoon stylization; hyper-saturated color grading, cinematic lighting, photorealistic 35mm film quality, sharp focus, fine film grain, bright noon light over a waterfall gorge, HDR bloom on spray and wet steel. Narrative Summary: In the churning basin beneath a great waterfall, a river-serpent tears through a flotilla of raider longboats mid-battle - a barefoot corsair woman fights her way across three boats, uses the serpent's own attack to destroy the raider flagship, and leaps clear as the beast drags it under. Dynamic Description - three layers of action in every beat: 0–3s: FOREGROUND - an oar snaps against the gunwale toward camera as two boats grind together, spray sheeting over the lens. MIDDLEGROUND - the heroine sprints down the centerline of a rocking longboat, hurdling rowers' benches, hooked axe in hand, and leaps the gap to a raider boat, kicking the first man overboard on landing. BACKGROUND - the serpent's vast green coil breaches out of the falls' basin, lifting a whole longboat clear of the water, crew and barrels raining off it. 3–6s: FOREGROUND - a raider swings a boarding hook past the lens; the heroine catches his wrist and throws him over her hip into the river. MIDDLEGROUND - she cuts a mast-line with the axe and rides the whipping rope across to the second raider boat, feet skimming the whitewater, ducking two spear thrusts as she lands. BACKGROUND - the lifted longboat crashes down across a rock spine and breaks in half; the serpent's head rises through the wreckage, jaws shedding timber, while archers on a far boat pour arrows into its neck. 6–9s: FOREGROUND - barrels and cargo wash past camera in the churn, a swimming raider clinging to one. MIDDLEGROUND - on the second boat the heroine fights three raiders at once in tight quarters - axe hooks a shield down, elbow, throw - soaked linen and braids whipping, the deck tilting as waves hammer the hull. BACKGROUND - the raider flagship turns broadside, its crew winching a huge harpoon ballista around toward the serpent as the falls thunder behind them. 9–12s: FOREGROUND - the harpoon fires past the lens with a whip-crack of rope. MIDDLEGROUND - the heroine cuts the harpoon line with a full-body axe strike as it goes taut, the recoil snapping the rope back across the flagship's deck and mowing down its crew line; she sprints for the bow. BACKGROUND - the enraged serpent, harpoon shed, turns on the source - the flagship - its body arcing over the basin, silhouetted against the waterfall's white wall. 12–15s: FOREGROUND - the heroine takes two steps up the bow and LEAPS, fully airborne over open water, axe trailing spray. MIDDLEGROUND - behind her the serpent slams down through the flagship, folding it in half in an explosion of timber, shields and water, dragging the wreck under in one coil. BACKGROUND - the surviving boats scatter toward the gorge walls through the raining debris; camera tracks the heroine crashing onto a drifting friendly boat's cargo, rolling up soaked and grinning as the basin behind her boils. Static Description: A wide waterfall gorge at noon - a thundering white falls feeding a churning basin, jagged rock spines, a battle-scattered flotilla of weathered painted longboats with iron-rimmed shields on the gunwales. Heroine: photorealistic young adult woman, warm brown skin, fierce dark eyes, long black braids with copper rings, barefoot; outfit 60% oxblood-red leather corsair vest over wrapped chest, 30% loose cream trousers and forearm wraps, 10% bronze buckles and a hooked boarding axe. Raiders: soaked leather and dull steel. The serpent: a colossal river-serpent with green-bronze scales. Original characters and creature design, not resembling any famous actress, no franchise resemblance, no logos, no readable text. Audio: The constant thunder of the waterfall, hulls grinding and cracking, oars snapping, bodies hitting water, rapid steel and axe exchanges, the serpent's subsonic bellow and hiss, arrows thudding into scale, the ballista's whip-crack and snapping rope, the catastrophic timber explosion of the flagship folding, spray hiss, crews shouting across the basin; surging orchestral score with a massive hit on the flagship kill. Quality: Photorealistic live-action, 35mm cinematic quality, HDR, sharp deep focus holding all three depth planes, natural motion blur on spray and debris, water physics with real weight, consistent character identity throughout, no morphing, no warped anatomy, no text or watermarks.

Use @[storyboard ref] as the authoritative 15-second cinematic shot blueprint. Do not render the storyboard sheet itself: exclude all panel borders, headers, text, labels, project card, director strip, style swatches and page layout from the final video. Treat each panel as an individual sequential shot guide, not as one combined image or split-screen. Follow panel order exactly, preserving staging, camera angle, framing, character placement, action beats, motion direction, timing rhythm and shot-to-shot continuity. Use the panel drawings for choreography, geography and composition only; translate the rough sketches into the intended final video style. Interpolate natural motion between panels, preserving start/end poses, prop states, effect states and screen direction. Allocate the 15 seconds across panels according to visual emphasis and rhythm. Do not invent new major actions, locations, characters, props, effects or alternate camera setups. Connect the panels into one smooth continuous action sequence with readable silhouettes, controlled motion blur and clear cinematic momentum.

2D anime action animation, feature-film quality, bold clean ink outlines with varying line weight, flat cel shading with hard-edged shadow shapes, rich saturated comic-book colouring, dramatic rim lighting, hand-drawn impact frames, radial and directional speed lines on fast movement, animated on twos with snappy pose-to-pose keyframes and brief holds. Non-photorealistic, never 3D render, never CGI. 16:9. 15 seconds, SIX shots, hard cuts only. Motion starts on frame 1, no opening freeze. Nobody speaks, no dialogue, no lip-sync. THE LOCATION: a dense neon-lit city at night in heavy rain. Narrow alleys between tall buildings, wet asphalt mirroring signage in cyan and magenta, fire escapes overhead, steam pouring from vents, puddles everywhere. Deep blue-black shadow between the lights. THE BOY: fifteen or sixteen, wiry and small, messy black hair plastered flat by rain, a determined set to his jaw, a fresh cut on one cheek. A too-large faded red hooded jacket soaked dark at the shoulders, dark shorts, worn trainers. He carries a battered brown paper parcel tied with string, held tight against his chest with both arms. Identical in every shot. THE PURSUERS: three older figures in dark rain-slick coats, faces always in shadow under hoods or hat brims, never clearly seen. They appear only as silhouettes and blurred shapes at the frame edges. THE GIRL: a small girl of about seven in a hospital bed, pale, dark hair, a thin blanket. She appears only in the final shot. PALETTE LOCK: cold blue-black night, cyan and magenta neon, wet black asphalt. ONE single accent — the faded red of his jacket, and one warm amber that appears only in the final shot. Nothing else carries either. CAMERA: hard lateral tracking that moves with the boy, low angles looking up, and locked-off frames on impacts. The camera never floats free. It either tracks hard with him or holds absolutely still. SHOT 1 — 0.0s to 2.5s — CAMERA: wide, low, tracking hard laterally with him. INTENT: open mid-chase with the parcel already established as the thing that matters. The boy sprints through a narrow neon alley in pouring rain, both arms locked around the brown parcel, water exploding up from every footfall, directional speed lines raking the frame. Three dark silhouettes round the corner behind him. HARD CUT. SHOT 2 — 2.5s to 5.0s — CAMERA: close on the parcel and his hands, tracking with him. INTENT: make the audience care about the object before they know what it is. His taped fingers gripping the wet paper parcel against his chest, the string cutting into his knuckles, rain running off the paper, one corner already torn and soaked. His breath fogging past the lens. HARD CUT. SHOT 3 — 5.0s to 7.5s — CAMERA: low angle, locked off, shaking on impact. INTENT: the cost — he takes the hit rather than drop the parcel. A hand catches his hood and yanks him backward. He hits the wet asphalt hard, a burst of radial white speed lines and spray on impact — but he twists mid-fall so his body takes it and the parcel stays clear of the ground, still clamped to his chest. HARD CUT. SHOT 4 — 7.5s to 10.0s — CAMERA: close on his face, low angle, locked off. INTENT: the turn — this is where the story stops being a chase. On his back in the puddle, rain hammering his face, a silhouette closing over him. He looks up. His expression is not fear. It is refusal. He kicks out hard and the frame holds on a single drawn impact frame. HARD CUT. SHOT 5 — 10.0s to 12.5s — CAMERA: wide, low, tracking hard with him. INTENT: sustained motion so the ending arrives at full speed. He is up and running again, out of the alley into a wide rain-lashed street, the three silhouettes falling behind him in the neon glare, the parcel still locked against his chest, red jacket streaming. HARD CUT. SHOT 6 — 12.5s to 15.0s — CAMERA: medium, locked off, very slow push-in. INTENT: the reveal — everything before this exists to make this land. A quiet hospital room. Warm amber lamp light, rain running down the window. He stands soaked and bleeding beside a bed, holding out the battered parcel. The small girl in the bed takes it and pulls the wet paper open. Inside is a single worn stuffed rabbit, dry. She holds it against herself. He drops his head, breathing hard, and smiles. The frame holds. Six hard cuts at 2.5s, 5.0s, 7.5s, 10.0s and 12.5s. No dissolves, no fades, no slow motion, no speed ramps. Continuous motion within every shot except the deliberate held impact frame. AUDIO: hammering rain throughout, fast wet footfalls, ragged breathing, cloth snapping, one hard body impact on wet asphalt, water spraying, distant city noise. In the final shot the rain drops to a soft patter on glass, a heart monitor beeping slowly, and quiet. No music, no voice, no dialogue. NEGATIVE: photorealism, live-action, 3D render, CGI, smooth interpolation, motion blur instead of speed lines, soft gradients, airbrush shading, digital painting look, blood spray, gore, weapons, guns, knives, visible faces on the pursuers, readable text, letters, words, numerals, readable signage, hospital charts, on-screen text, captions, subtitles, logos, watermark, changing costume, changing hair, changing face, extra children, cloned characters, dissolves, fades, slow motion, speed ramping, opening freeze, static first frame, anyone talking, lip-sync.

FORMAT: cinematic continuous shot / motivated camera movement / 15s SCENE A crowded medieval market street inside a stone city at dusk. Narrow cobblestone road, wooden stalls, hanging banners, livestock moving through the crowd. Warm torchlight reflects on damp stones while light mist drifts between buildings. CAMERA CONCEPT A continuous motivated camera move where each new moving subject entering the frame redirects the camera’s attention. Every motion naturally hands the focus to the next subject. SEQUENCE 0:00–0:03 Close street-level view of a market stall. CAMERA FOCUS: a woman bargaining with a merchant while selecting fruit from a wooden basket. She hands coins to the merchant. 0:03–0:05 A horse pulling a heavy wooden cart suddenly crosses the foreground from the opposite direction, briefly blocking the frame. CAMERA SHIFT: the camera catches the cart and begins tracking it as it moves through the market. 0:05–0:07 The cart squeezes between stalls and brushes past a hanging banner. The banner swings violently across the frame. CAMERA SHIFT: as the banner clears the view it reveals chickens scattering across the cobblestone street. 0:07–0:09 A street boy runs after the escaping chickens, chasing them through the crowd. CAMERA SHIFT: the camera begins following the boy as he runs between villagers. 0:09–0:12 The boy rushes past a tavern entrance and disappears into the crowd. CAMERA SHIFT: the tavern door suddenly swings open as someone exits. 0:12–0:15 The camera glides through the open doorway into the dim tavern interior. Lantern light flickers across wooden tables and drifting smoke. CAMERA FINAL FOCUS: a lone armored knight sitting quietly at a corner table, a massive sword leaning beside the bench as the knight slowly lifts his gaze. STYLE Layered medieval street life, natural crowd choreography, continuous motivated camera movement. LIGHTING Warm torchlight outside, dim lantern glow inside the tavern, smoke and dust catching the light. QUALITY photorealistic, cinematic lighting, grounded camera motion, rich medieval atmosphere, highly detailed

NO MUSIC ONLY NATURAL SOUND — Epic 2D cel-shaded animation, hand-drawn anime-style linework and flat hard-edged cel shading, layered with graphic-novel comic techniques: halftone Ben-Day dot texture, subtle chromatic offset on the edges, cross-hatch shadows, hand-drawn motion lines and action smears, punchy stepped "on-twos" framerate, saturated high-contrast urban palette. Sun-drenched New York City — glass towers, water towers, fire escapes, yellow cabs far below — dynamic energetic camera cutting between POV, handheld and tracking. Hard opening hook — NICO (@Image1) in freefall straight down a canyon of skyscrapers, tumbling, the street and cabs rushing up fast below, wind screaming — POV snapping between the ground rushing closer and his own outstretched arm. At the last instant a web line fires from his wrist with a sharp THWIP, snaps taut on a crane arm and yanks him into a fast upward swing, the camera whip-panning up with him. NICO (@Image1): a 16-year-old with dark skin, short black textured hair, gold hoop earrings, oversized taupe tee, baggy tan cargo pants and chunky orange-and-white high-tops, grinning wide as he arcs skyward. Handheld tracking chasing behind him as he releases at the top of the swing, tumbles through open air over the avenue, then fires a second web from his other wrist and swings again, building momentum between the towers, tee and cargos snapping in the wind. Cut to a low street-level angle looking straight up — tiny against the buildings, cabs and pedestrians blurring beneath. Cut to POV down the web line as the next glass tower rushes toward him; he plants both feet, wall-runs across the windows, kicks off and dives. Dynamic finish: soaring wide aerial as NICO (@Image1) swings up in a huge arc over a busy intersection, arms spread, his silhouette crossing the sun, the Manhattan grid sprawling below. He releases, flips, and lands hard in a crouch on a rooftop water tower, one hand down, then rises to his feet — city wind pulling at his clothes, a confident grin, skyline glittering behind him. Face, hairstyle, earrings, wardrobe and proportions locked to the reference sheet (@Image1). SFX: screaming wind, the sharp thwip and twang of web lines firing and going taut, fabric snapping, sneakers scuffing glass and gravel, distant traffic and horns, the impact thud of each landing, Nico's breath and a quick laugh. Clean 2D cel-shaded animation, bold hand-drawn linework, flat cel shading, halftone texture, chromatic-offset edges, smooth fluid motion, subtle film grain, no music, no text, no watermark, no photoreal, no 3D CGI, no live-action.

Use the provided image as the only character design reference for the Fox Spirit Wanderer. Maintain exact consistency in her adult face, long dark-purple hair, bamboo hat, white fox-mask ornament, purple-black floral outfit, single large fox tail, sword, scabbard, and body proportions. Create a 15-second, 16:9 cinematic Eastern dark-fantasy CG boss battle. The action begins immediately on the first frame. No introduction, staring, walking, or slow sword-drawing setup. 0–3s | Thunder Stomp and Running Draw A colossal Thunder Qilin slams its raised front hoof into the mountaintop shrine plaza, sending blue-white lightning through the cracked stone floor. The Fox Spirit swordswoman is already sprinting toward it. She jumps from a tilted stone lantern and performs a complete running iaijutsu draw, cutting through a small incoming lightning arc. She clears the ground shockwave, lands on the shrine roof, and continues running as horn lightning destroys the tiles behind her. 3–6s | Rooftop Chase and Close Combat She sprints across the roof, wall-runs three steps along a stone pillar, then leaps toward the Qilin’s front leg. She narrowly avoids a hoof strike, lands briefly on a golden armor plate, and slashes through a gap in its scales. Blue electricity, metal sparks, and rainwater burst from the impact. The Qilin shakes her off. She backflips onto a fallen pillar, slides beneath its body, avoids the crushing hooves, and deflects debris with her scabbard. 6–9s | Lightning Breath and Collapsing Bridge She exits beneath the Qilin and runs across a damaged cliff bridge. The Qilin gathers energy in its throat and releases a continuous lightning breath from its mouth, destroying the bridge behind her. She jumps across a broken section, climbs onto the opposite stone bridge, and keeps moving. The Qilin charges forward and leaps across the abyss, creating a dramatic scale contrast. As it passes, she steps onto its side armor, runs two steps along its body, and performs a fast sheathed iaijutsu slash across a glowing scale seam before jumping onto a torii beam. 9–12s | Tail Sweep and Lightning Core The Qilin sweeps its massive tail through the torii gate, shattering wood, stone lanterns, and shrine structures. The swordswoman jumps from the collapsing beam onto a stone pillar, then toward the Qilin’s shoulder. She cuts through a thin horn-generated lightning branch and lands near its neck, gripping a golden armor ring for support. The glowing cyan lightning core on the right side of its neck becomes visible. She delivers a short horizontal slash against the core’s outer seal. The core cracks briefly but is not destroyed. The Qilin roars and releases a circular surge of lightning across the cliff platform. 12–15s | Into the Thundercloud Realm The lightning blast throws her backward. While airborne, she accurately sheathes her sword, rotates, and lands on the final broken sky bridge. She immediately resumes sprinting as the Qilin charges behind her, causing the bridge to collapse beneath its weight. The Qilin leaps into the storm clouds using temporary lightning paths. The swordswoman jumps from the final stone platform, draws her sword again, and lands on its right shoulder armor. Together they charge into a vast thundercloud realm filled with floating torii ruins, ancient stone rings, lightning bridges, and distant airborne temples. The camera rapidly pulls back. The Qilin continues running through the storm while the swordswoman moves across its shoulder, searching for another attack opening. In the final moment, the Qilin releases a close-range lightning arc. She jumps toward a floating stone ring as sword light and lightning cross again. End while the battle is still continuing. No winner, death, victory pose, or static freeze frame. Visual Style Eastern dark fantasy, AAA action-game cinematic, mythological boss battle, realistic physics, heavy rain, volumetric thunderclouds, large-scale environmental destruction, wet stone and wood, detailed metal, silk, hair, fur, scales, and lightning effects. Emphasize the extreme size difference between the agile swordswoman and the enormous Qilin. Keep the action fast but spatially clear. Fixed Character and Boss Rules The swordswoman always has one large fox tail, no visible fox ears, one sword, and one attached scabbard. Her bamboo hat must remain securely on her head. The Thunder Qilin always has one head, two branching horns, four legs, one tail, a silver-white mane, dark blue-black scales, and a fixed cyan lightning core on the right side of its neck. Lightning must originate only from the Qilin’s hooves, mouth, horns, scales, mane, or core. No blood, gore, extra limbs, extra tails, extra weapons, transformation, teleportation, unsupported flying, giant sword beams, random explosions, HUD, subtitles, text, logos, or watermarks.

REFERENCE DEFINITIONS @[Image 1](image_1): START FRAME - the video's very first frame is this exact image, 1:1: a colossal translucent amber caramel golem, glassy candy body swirled with molten toffee, caught mid-punch as his huge fist smashes through a striped candy-cane tower in a frozen explosion of rainbow sprinkles and sugar dust, thick syrup strands dripping from his other fist; below him a cavalry of gingerbread knights in white piped-icing armor and silver tin helmets charges on glossy black licorice horses across powdered-sugar snow cut by a milk creek, candy-cane lances raised; cream-and-gumdrop candy mountains frame the valley under a brilliant blue sky with cotton-candy clouds, hard high-key midday sun. Exact composition, camera position, framing, lighting, palette and scale lock for frame one — the charge described in the prompt ignites out of this very still. Reference. TECHNICAL BLOCK Image-to-video from @[Image 1](image_1) - @[Image 1](image_1) is the literal first frame of the shot, matched 1:1 in composition, camera position and framing, with the camera already accelerating from frame one - no static opening hold. Photoreal cinematic. 16:9. 15s. SFX only, no music. Single continuous shot, one take no cuts, aggressive racing-FPV oner - whip-banks, barrel rolls, slaloms, snap-reframes and speed ramps all inside the one take, 4K ultra-detailed, photorealistic macro detail, anamorphic film look, epic cinematic scale, playful whimsical tone, hard high-key saturated candy daylight, cinematic lighting, professional color grading, sharp focus, hyper-detailed texture, film grain, depth of field mastery. NON-IP, no on-screen text. PROMPT The shot opens on @[Image 1](image_1) as its exact first frame - the colossal amber caramel golem mid-punch through the candy-cane tower, sprinkles erupting, the gingerbread cavalry thundering toward him - and the camera is already accelerating from the very first frame: no static hold, it rips forward out of the exact camera position of @[Image 1](image_1) like a racing FPV drone punched off the line, dropping hard to hoof height in one breath, and never returns to a wide frame again. The golem's punch completes its swing, the tower's upper half bursting apart in a fountain of rainbow sprinkles and sugar dust, and he wheels on the riders with a glassy groan, syrup strands whipping from his fists, while the knights close the distance and swarm his ankles, jabbing striped candy-cane lances into his caramel shins, stretching sticky amber threads with every strike. The whole valley blazes in saturated candy color - translucent amber, piped white icing, candy-stripe red, licorice black, pastel gumdrop pink and teal under hard high-key midday sun, no trace of golden hour. The camera slaloms THROUGH the charging cavalry line, knife-edge banking left and right between licorice horses at full gallop, extreme 4K macro brushing past crumbly gingerbread thighs, glossy icing armor and snapping helmet plumes close enough to touch, powdered sugar spraying off hooves across the lens - then snap-rolls into a full barrel roll as it punches straight through the billowing cloud of falling sprinkles from the shattered tower, glittering candy confetti corkscrewing past the glass in hyper-detailed macro. Around the three-second mark it whip-pans down without slowing and grazes one small detail at full speed - a licorice hoof punching through the crusted sugar snow, the crust cracking in a starburst, icing splashing and candy gravel pattering off the lens - then yanks back up in a hard climbing bank. Surging on, the camera surfs the golem's translucent flank inches off the amber, speed-ramping faster as his backhand tears through a second candy-cane spire - the camera flinching aside and threading the gap UNDER the swinging fist as poles and gumdrop boulders tumble past the lens - then dives through his stomping legs once, whips around a rearing horse, and slings itself at the seven-second mark straight into the path of one gingerbread knight leaping from the saddle at full gallop, lance leveled at the giant's chest - and time freezes into the single extreme-macro orbit of the shot: the camera corkscrews slowly around the suspended knight - crumb-grain gingerbread skin and piped icing armor razor-crisp, helmet plume frozen mid-flutter, the striped lance tip aimed at the swirling molten heart inside the golem's glassy amber chest, a halo of suspended sprinkles, sugar motes and syrup droplets hanging in the hard light, the giant's grimacing translucent face warped huge behind him. Time snaps back to full speed - the lance strikes home in a burst of amber chips and stretching syrup threads - and the camera slingshots out of the orbit like a released catapult, diving low beneath the staggering golem, barrel-rolling once as it threads the arch between his massive caramel legs where dripping syrup strands whip past the lens, grazing the glassy inner amber where sunlight refracts through, and bursts out the far side into the open valley in a rising speed ramp just as a second candy-cane tower collapses there in a slow-blooming fountain of sprinkles, cream and tumbling gumdrops - the camera skimming the milk creek so low its wake kicks spray, slaloming two last riders, and charging straight into the drifting sugar cloud without ever pulling up or back. SFX only: thundering toffee hooves on sugar crust, rushing wind whipping past the lens, crunching shattering candy, deep glassy caramel groans, whooshing fist swings, sprinkle showers pattering like hail, sticky stretching syrup creaks, tiny muffled gingerbread battle cries, rumbling tower collapse.

A single continuous ultra-realistic cinematic shot in vertical 16:9. Bright sunny daytime. A beautiful young woman walks out of her house carrying a fluffy silver tabby cat, smiles, walks to her scooter parked outside, sits on the scooter, gently places a tiny spinning propeller fan helmet on the cat's head, starts the scooter, and rides away. As the propeller spins faster, the cat smoothly lifts into the air and flies beside the moving scooter at the same speed, staying close to the woman. The cat's legs dangle naturally, its fur flows in the wind, and it looks around curiously while flying. The woman occasionally glances at the cat and laughs. The camera follows them in one seamless handheld smartphone tracking shot with natural motion blur, realistic lighting, detailed fur, realistic physics, no cuts, viral social media aesthetic, photorealistic. Negative Prompt: Cartoon, CGI, anime, low quality, extra limbs, deformed cat, floating without propeller, flickering, teleporting, cuts, text, watermark, unrealistic physics.