Fisheye Skater in Graffiti Park

[CINEMATIC SETUP] Early-2000s consumer DV camcorder aesthetic / edited home video style with distinct hard cuts / slight handheld drift / candid framing. Faded colors, harsh realistic street light contrast. Late-night Manhattan street corner. Thick steam rising from a glowing stainless steel street food cart. Busy, gritty, unfiltered urban atmosphere with pedestrians walking by. Natural ambient sound only (spatial sound design, dense New York City street hustle, classic NYPD sirens echoing, yellow cab horns honking, subway vent rumbles, no narration). [IMAGE REFERENCES / LEGEND] @ image = [main character: 30-year-old man, 1.93m tall, movie-level realistic facial features, no deformation, stable throughout. Wearing a solid black t-shirt with JFB logo on the chest (back must be strictly plain black with absolutely no graphics, text, or logos), completely black cap. CRITICAL DETAIL: The back of the cap features a bare, empty semi-circular cutout revealing hair/skin, and a plain plastic snapback strap. Absolutely NO logos, text, patches, or embroidery on the back of the cap. Long full-length blue oversized jeans (pants completely cover the legs down to the sneakers, not shorts), White Nike Air Force 1] [TIMELINE SECOND BY SECOND] 0-3s: [Medium tracking shot] @ image stands casually next to a stainless steel hot dog cart on a bustling Manhattan street. A street vendor inside the cart hands him a hot dog. Anonymous pedestrians walk past in the background. [Slight handheld drift] + [Harsh overhead street light] + [Hissing steam, vendor talking, dense NY street hustle, and yellow cab honking SFX] 3-6s: [HARD CUT TO: Close-up side profile] @ image leans casually against the cart, taking a bite of the hot dog. He is viewed strictly from the side, looking completely away from the camera at the pedestrians walking by. He makes zero eye contact with the lens. [Gentle framing drift] + [Exposure adjusting smoothly] + [Muffled Manhattan crowd chatter, footsteps, and distant NYPD sirens SFX] 6-11s: [HARD CUT TO: Medium wide shot] @ image stands by the cart in the gritty street lighting, chewing his food. In the background, the street vendor is visibly wiping down the stainless steel counter. Several pedestrians walk across the frame, adding depth to the city environment. [Steady handheld] + [Harsh shadows] + [Deep subway vent rumble and overlapping city traffic SFX] 11-15s: [HARD CUT TO: Medium tracking shot] @ image tosses a paper wrapper into a metal street trash can. He turns sideways, maintaining a strict side profile to the camera, and begins walking down the illuminated stairs of a classic New York subway entrance. A rush of air from the underground tunnel slightly flutters his black t-shirt. The camera tilts down following his descent from the street level. [Handheld camera tracking down] + [Echoing footsteps fading into loud, iconic MTA subway train screeching SFX] [STYLE & QUALITY BOOSTERS] Ultra-realistic documentary realism, genuine candid behavior, accurate real-world physics of smoke and fabric flutter, natural body language, unscripted slice-of-life feeling. Strong environmental authenticity. Rich real-world details, active background extras, gritty early-2000s urban video texture. Exact character consistency with @ image. [NEGATIVES] looking at camera, eye contact with lens, breaking the 4th wall, disappearing objects, dropped objects, deformed faces, blurry motion, low resolution, text artifacts, inconsistent lighting, watermark, extra limbs, broken physics, illogical architecture, morphing environments, continuous morphing, empty streets, desolate, shorts, short pants, bermudas, bare legs, exposed legs, logos on cap, rear cap logo, back hat embroidery, logo on snapback strap, graphic on back of shirt, accidents, professional lighting, cinematic lighting, 8k, modern stabilized gimbals, clean pristine resolution, cinematic color grading, slow motion.

Use @[storyboard ref] as ordered key poses for Part One of a cinematic neon urban chase. Pass through P01–P09 in order with matched action, purposeful cuts, escalating momentum, and clear spatial continuity. Ignore storyboard borders, panel numbers, captions, speech bubbles, lettering, and line artifacts. Use @[character ref1] for Vex. Preserve her exact face, asymmetrical magenta-and-cyan hair, beanie, piercings, patched cropped bomber jacket, graphic tank top, loose paint-splattered cargo pants, chains, fingerless gloves, high-top sneakers, messenger bag, proportions, and colors. Keep Vex visually consistent in every shot. Police officers wear realistic matching dark uniforms and remain clearly separate individuals. Never duplicate Vex, alter her outfit, remove her bag, or change her scale. LOOK + WORLD: Cinematic stylized live action with graphic comic-book contrast, grounded realism, and subtle cel-shaded influence. A gritty rain-soaked city alley at night with graffiti-covered brick, torn posters, stickers, dumpsters, chain-link fencing, rusted fire escapes, pipes, vents, puddles, steam, litter, rooftop structures, and distant neon signs. Saturated magenta, cyan, violet, acid green, and orange accents. Harsh streetlamp shadows, neon rim lighting, atmospheric haze, realistic wet fabric, metal, paint mist, water spray, sparks, dust, and debris. CAMERA + ACTION: Begin quietly and escalate into a fast, readable chase. Use 24mm alley wides, 35mm handheld tracking, 50mm action coverage, 85mm facial close-ups, low-angle running shots, overhead rooftop views, and brief impact shake. Maintain screen direction and geography between cuts. Vex moves with elite parkour ability: fast acceleration, precise foot placement, wall runs, vaults, controlled landings, rail balances, climbing transitions, and efficient evasive movement. Her messenger bag and clothing react naturally to momentum. Police pursue urgently but realistically, communicating by radio and calling for backup. They never teleport ahead of her. SOUND: Spray-can rattle and hiss, paint mist, boots splashing through puddles, breathing, fabric snapping, chains jingling, bag straps shifting, police radios, shouted commands, sirens approaching, footsteps, fence rattles, metal impacts, steam hiss, distant traffic, neon buzz, rooftop gravel, and city ambience. MOTION PHRASES: P01: Wide establishing shot. Vex spray-paints a large neon piece on a brick wall beneath flickering signs. Wet pavement reflects the colors. P02: Tight profile and hand close-ups. She paints confidently, spray mist catching the neon while her marker-filled messenger bag rests against her hip. P03: Two officers enter the alley and catch her in the act. Flashlights snap toward her. Vex freezes briefly, looks over her shoulder, then calculates an escape route. P04: She suddenly bolts deeper into the alley. The officers chase, shouting for her to stop as shoes splash through puddles. P05: Low tracking shot. Vex clears a dumpster with a fast two-handed vault, lands smoothly, and maintains speed while the officers struggle around the obstacle. P06: Side tracking shot. She runs two steps along a brick wall, grabs a fire-escape rail, swings across a gap, and drops onto an elevated platform. P07: One officer stops to radio for backup. Sirens intensify as additional units enter surrounding streets and alley exits. P08: Wide rooftop chase. Vex climbs rapidly, balances across a narrow rail, jumps between rooftops, and rolls through the landing while multiple officers spread out behind and below her. P09: Cliff-hanger. Vex reaches the edge of a high rooftop and skids to a stop. Police lights surround the block below, officers emerge onto nearby roofs, and a helicopter spotlight sweeps toward her. She looks down at a dangerous gap, then back at the closing officers. End on an intense close-up of her determined expression as she prepares to jump. Cut to black before takeoff.

cinematic 15 second sequence shot on 35mm film with Fujicolor emulation. A wide shot establishes an atmospheric alleyway at dusk outside a traditional Izakaya. Two male models stand leaning against a wooden slatted wall wearing luxury floral techwear blending tactical urban fabrics with subtle embroidered botanical patterns. The camera is handheld adding a raw documentary realism. The lighting is moody with warm amber light spilling from the Izakaya paper lanterns contrasting with the cool blue twilight of the alley. One model adjusts a tactical strap while the other looks down the alley. The camera pans right following their gaze to a mysterious courier dressed in a sleek black leather jacket. The courier approaches rapidly and tosses a heavy tactical duffel bag onto a wet manhole cover. The camera cuts to a high angle close up as the courier quickly unzips the bag revealing premium leather garments inside. The camera then cuts to a top down shot of the manhole cover. Ensure there are no text overlays anywhere in the video to maintain an uninterrupted view. Audio track features a 120 to 128 BPM electronic trap rhythm with heavy sub bass and a distinct metallic shimmer sound effect during the final shot.

Main Subject: Young Korean woman, 24, natural everyday appearance, shoulder-length black hair worn loose with soft bangs, realistic skin texture with a natural evening glow, minimal makeup, relaxed and cheerful personality. Wearing an oversized charcoal bomber jacket over a white hoodie, loose light-wash jeans, black Converse-style sneakers, a small black nylon shoulder bag, and silver hoop earrings. Maintain consistent identity, clothing, hairstyle, and appearance throughout. Location: A quiet neighborhood street food alley in Seoul on a cool autumn evening. Small pojangmacha tents, folding tables, warm hanging lights, light steam rising from food stalls, parked scooters, narrow side streets, and a handful of local residents enjoying dinner. Cozy, authentic, and non-touristy. Visual Style: Ultra-realistic documentary realism. Natural, candid behavior with genuine interactions and subtle body language. Warm tungsten lighting mixed with cool streetlights creates an authentic early-2000s nighttime atmosphere. Camera Style: Early-2000s Sony MiniDV camcorder. Heavy handheld shake, autofocus hunting in low light, soft focus, faded colours, visible digital noise, slight motion blur, exposure pumping around bright stall lights, MiniDV compression artifacts, no stabilization. 00:00–00:03 She walks through the food alley, smiling as steam drifts from a tteokbokki stall. The camera briefly focuses on the steaming food before correcting. 00:03–00:06 She orders a cup of tteokbokki, thanks the vendor with a polite smile, and carefully accepts the warm cup with both hands. 00:06–00:09 Standing beside the stall, she takes a small bite, laughs quietly because it's still too hot, and gently fans the food with her hand. Autofocus shifts between the steam and her face. 00:09–00:12 She continues walking, sipping a canned drink while watching people pass under the glowing streetlights. The operator accidentally tilts the camera upward toward the lights before reframing. 00:12–00:15 She notices the camcorder, raises the snack cup with a playful smile as if inviting the viewer to join, then walks farther into the softly lit alley. The recording abruptly cuts to black. Audio: Natural ambience only–quiet conversations, sizzling food, chopsticks clinking, distant traffic, scooter engines, footsteps, light wind, and soft laughter. No music. No narration. Goal: A forgotten MiniDV home video from autumn 2004, capturing a cozy evening enjoying street food in a quiet Seoul neighborhood–warm, imperfect, nostalgic, and completely believable.

Animate the provided first frame into a 15-second vertical 9:16 K-pop TikTok fancam-style dance video. Keep the same five original idols, same faces, same hairstyles, same outfits, same pastel music show stage, same lighting, same formation logic, and same focus on Nari throughout the entire video. STYLE: Vertical K-pop idol fancam, cute dreamy girl group concept, pastel pink and pale blue stage lighting, glossy floor, music show performance feel, clean synchronized choreography, bright expressions, high replay TikTok edit. The video should feel like a viral focus fancam, with Nari as the main subject and the other four members supporting the formation. CHARACTER LOCK: Exactly five fictional female idols appear for the entire video. Nari is always center focus: long dark brown wavy hair, white cropped cardigan, pale pink top, navy pleated skirt, white knee socks, white platform sneakers. Yuna is left-front: short black bob, pale blue cropped blouse, white pleated skirt. Mina is right-front: long ash-blonde hair, white crop top, pastel pink varsity jacket, light gray shorts. Sera is left-back: high ponytail, baby blue zip jacket, white tennis skirt. Lia is right-back: light brown wavy hair, white sleeveless top, soft pink skirt. Do not swap faces, outfits, hairstyles, positions, or identities. No extra dancers. TIMELINE: 0–1.5s: Start exactly from the first frame. Nari holds the center-front pose, one hand near her cheek, hips angled. The other four members hold matching poses behind her. Pastel LED lights glowing. 1.5–3s: The beat drops. All five snap their heads up, smile, and flick one wrist outward from the cheek. Camera stays vertically centered on Nari. 3–4.5s: They step right with a small hip bounce, then step left. Arms cross at the chest and open outward in clean synchronization. 4.5–6s: Nari points softly toward the camera and silently lip-syncs one beat. The other members mirror smaller hand accents behind her. 6–7.5s: Full-body choreography. All five perform quick step-cross footwork, heel pivot, and knee bounce. Feet stay grounded and readable; Nari remains the focus. 7.5–9s: Signature cute hand sequence: palm up, palm down, cheek tap, tiny heart gesture, wrist flick. Hands must be clean and synchronized. 9–10.5s: Formation shifts from V shape into a soft horizontal line while keeping Nari slightly forward. The group performs a gentle shoulder roll and short body wave. 10.5–12s: All five do a small synchronized jump-step, landing softly with arms angled near their faces. Skirts, hair, jackets, and socks move naturally. 12–13.5s: Subtle fancam push-in on Nari. She smiles brightly, keeps eye contact with the camera, and hits a cheek-level hand pose. 13.5–15s: The group returns to the original V formation. They repeat the wrist-flick and cheek-pose signature move. End on a loopable final frame with Nari smiling center-front and the pastel LED stage glowing behind her. CAMERA: Vertical 9:16 focus fancam. Stable handheld or monopod feeling, slight smartphone zoom compression, minimal camera movement, Nari centered most of the time. Keep full-body visibility during footwork and medium-full framing during close moments. MOTION: Cute K-pop challenge choreography: head snap, wrist flick, cheek gesture, side step, hip bounce, arm cross, camera point, step-cross footwork, tiny heart gesture, shoulder roll, body wave, jump-step, final pose. LIGHTING: Soft pink and pale blue LEDs, white face light, glossy stage reflections, gentle haze, bright pastel idol-stage atmosphere. ENDING: End with exactly five idols in a clean V formation, Nari center-front smiling with one hand near her cheek, ready to loop back to the opening.

【生成模式】 Seedance 2.0 Fast 【视频规格】严格生成 15 秒视频，16:9 横屏，三个连贯清楚的镜头。【整体风格】电影级写实质感，以庄严仙侠悬疑开场，逐渐反转为温暖克制的冷面喜剧；采用宏大的古典中国山水构图、克制表演、清冷风暴蓝阴影、暖色火光、浓郁空间纵深、具有真实重量的灵兽动作，以及清楚明确的三段式叙事递进。【角色锁定】角色 A｜剑仙师姐使用 @图片 1 严格锁定角色 A 剑仙师姐：25–30 岁的东亚女性，椭圆脸，白皙自然肤色，深色杏眼，黑色长发半挽，以白玉簪固定，身材高挑纤细，穿白色刺绣真丝汉服、半透明分层宽袖、银色腰封、玉佩和白色布靴。角色 B｜小师妹使用 @图片 2 严格锁定角色 B 小师妹：20–25 岁的东亚女性，圆润灵动的脸，黑发编辫，身形娇小，穿青绿色汉服、深色腰带、木簪和黑色布鞋。【核心生物设定】始终只出现一条成年巨龙和三只幼龙。【镜头 1｜0-5s｜低角度全景跟拍】黄昏时分，一座巨大的悬崖龙窟入口，黑色岩壁、盘曲古松、残破石柱、古老爪痕、缓慢漂浮的山雾、散落兽骨，以及洞穴深处暗淡的橙色火光，共同建立清楚稳定的空间关系。同一个穿白色刺绣真丝汉服的剑仙师姐手持唯一一柄银色长剑向前行进，同一个穿青绿色汉服的小师妹跟在她身后三步。震耳欲聋的龙吟让崖壁尘土落下；两人同时停步，黑暗中缓缓睁开一双巨大的金色眼睛。【镜头 2｜5-10s｜中景牛仔景】同一条覆盖黑色鳞片的古老巨龙从洞穴中缓慢低下龙首，鼻孔逸出弯曲烟气。同一个剑仙师姐拔出同一柄银色长剑，挡在同一个小师妹身前，厉声说道：“孽龙，离开此地！” 巨龙只展开一侧翅膀，强烈气流将山雾、两人的头发和衣袖同时向后推开；英雄式鼓声与剑气嗡鸣逐渐升起。【镜头 3｜10-15s｜近景揭晓转极近特写】摄影机沿巨龙翅膀后方平稳环绕半圈，清楚揭示翅膀内侧保护着一个温暖巢穴，巢中正睡着三只幼小龙崽。同一条古老巨龙小心收拢翅膀护住它们，压低声音说道：“小声些，刚哄睡。” 其中一只小龙崽轻轻打了个喷嚏，喷出一颗微小火星，随后向巨龙怀中靠近。同一个小师妹缓慢压下剑仙师姐的剑锋，轻声说道：“师姐，它只是护崽。” 极近特写：同一个剑仙师姐默默将长剑归鞘，原本凌厉的表情逐渐柔和；巨龙发出一声充满疲惫感的老父亲式叹息。【技术要求】严格总时长 15 秒，16:9 横屏，三个连贯清楚的镜头，原生同步普通话对白，人物面部与服装稳定，始终只出现一条成年巨龙和三只幼龙，翅膀、鳞片、山雾、火星、真丝衣料、头发和尘土物理效果真实，不生成字幕，不出现任何现代元素。【负面词】 blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props

Cinematic 15-second continuous single-take video, ultra-photorealistic, rainy Tokyo night, extreme low ground-level racing POV camera that never rises above knee height, invisible viewpoint racing unseen through the city, seamless fluid transitions, no cuts, no jump cuts. Wet reflective asphalt and tiles glowing with neon reflections, cinematic volumetric neon lighting, shallow depth of field with selective focus, subtle film grain, high dynamic range, 1280x720 or 1920x1080, 24fps, dynamic motion blur on fast movement. 0-3s: Extreme low-angle racing forward across wet reflective tiled Shibuya scramble crossing at night. Silhouetted legs and umbrellas of pedestrians stride past, giant glowing LED billboards and neon signs (Japanese text, H&M style) illuminate the rain-slicked ground with blue-red-orange reflections. Camera skims just above the pavement. 3-6s: Seamless plunge into darkness under a large black wet umbrella, extreme close-up of the fabric underside and metal ribs filling the frame, rain droplets, dark moody atmosphere. 6-10s: Camera emerges into a warm ramen shop interior, slow push-in on a steaming red-rimmed ceramic bowl of tonkotsu ramen on wooden counter (chashu pork slices, green onions, nori, soft-boiled egg), rising steam, soft warm lighting, chopsticks and small dishes nearby. 10-13s: Hard cut-free transition into close-up of a row of aluminum drink cans inside a dark blue-lit vending machine, pull-tabs and condensation details, cool cyan lighting. 13-15s: Final explosive acceleration into blurred high-speed neon Tokyo street canyon, wet reflections stretching, colorful light streaks, camera racing forward into pure motion blur and neon bokeh. Overall style: photorealistic cinematic Tokyo night photography, continuous low-angle tracking / racing camera, rainy wet surfaces, rich neon color grading (cyan, magenta, orange, deep blues), atmospheric depth, no text overlays, no watermarks, highly detailed textures, natural physics of rain and reflections, immersive “unseen night” feeling.

Use @[char1] only for C1 identity, hair, luminous floral kimono, and exactly one unsheathed slender katana: black grip, small floral tsuba, narrow polished blade; ignore all other weapons and scabbards. Use @[char2] only for C2 identity, wide kasa, black robes, and exactly one unsheathed heavy katana: white grip, round iron tsuba, broad polished blade; ignore all other weapons and scabbards. LOOK + WORLD: 2D cel anime; moonlit rainy bamboo grove, wet earth, pools, cut stalks, canopy gap; fixed low blue-green fill, 24-35mm depth. ACTION + CAMERA: Pro Anime Combat: burst, hit-stop, recoil, re-entry; cut on action at each P##. Keep C1 left, C2 right; low tracks rise to crane, then locked P14. VFX SYSTEM: Pair N is one cold white-blue bolt from storm sky through canopy at blade contact N, causing an instant full-screen flash; each clears to fixed moonlight within its beat, with no lingering glow or buildup; C1 energy crimson-white, C2 silver-blue. SOUND: No music; rain hiss, thunder impacts, sandal pivots, cloth snaps, blade hum. BEATS: P01: Ground profile WS, track: C1's crimson-white low sweep meets C2's heavy katana at contact 1, triggering pair 1; recoil feeds C2's descending answer. P02: Low front 3/4 FS, lead: C2's silver-blue down-cut meets C1's slender katana at contact 2, triggering pair 2; C1 redirects into a rising diagonal. P03: Waist front 3/4 MLS, pan: C1's rising crimson-white diagonal meets C2's cross-parry at contact 3, triggering pair 3; her overshoot carries toward one bamboo. P04: Profile FS, track: C1's slender katana cuts that bamboo; no blade clash, no pair; its severed top falls and stays down as C2 closes low. P05: High OTS MLS, chase: C2's low silver-blue counter meets C1's dropped guard at contact 4, triggering pair 4; C1 snaps into a reverse cut. P06: High profile MS, pan: C1's crimson-white reverse cut meets C2's high parry at contact 5, triggering pair 5; C2 compresses for a close thrust. P07: Profile FS, push: C2's silver-blue thrust meets C1's crossed deflection at contact 6, triggering pair 6; his overshoot carries toward the bamboo. P08: Low rear 3/4 WS, lead: C2's heavy katana shears several trunks; no blade clash, no pair; split sections fall and stay down as C1 re-enters. P09: Low rear 3/4 MLS, track: C1's broad crimson-white wave meets C2's blade wall at contact 7, triggering pair 7; C2 redirects into a descending diagonal. P10: Ground profile FS, chase: C2's descending silver-blue diagonal meets C1's kneeling catch at contact 8, triggering pair 8; C1 pivots upward among fallen bamboo. P11: Low front 3/4 MLS, pan: C1's crimson-white waist cut meets C2's low counter at contact 9, triggering pair 9; both recoil into crossing diagonals. P12: Top-down WS, crane: their diagonals cross at contact 10, triggering pair 10; recoil lifts both blades into the heavy exchange. P13: High front 3/4 FS, push: C2's heavy silver-blue sheet meets C1's upward split at contact 11, triggering pair 11; both load the overhead collision. P14: Low frontal WS, locked: their two assigned katanas collide overhead at contact 12, triggering the largest pair 12; the flash clears to fixed moonlight as both blast apart, remain visible, and reorient.