Cinematic Portrait on a Wet Rooftop

Create a realistic cinematic 30-second vertical video set in a modern corporate business district, beginning with a hardworking street cleaner in an orange reflective uniform sweeping the pavement outside a tall glass office building while busy professionals walk past him. A stylish young Asian businesswoman in a white blazer and black skirt arrives from a luxury black car and notices the cleaner and the people around him. A group of young businessmen in formal suits begin laughing and mocking the cleaner while recording him on their phones, creating an uncomfortable atmosphere. The scene then transitions inside a luxurious glass-walled conference room where senior executives sit around a long meeting table, while the same cleaner quietly stands in the room holding his broom. Everyone looks surprised and confused as the cleaner becomes the unexpected center of attention. The businesswoman reacts emotionally, looking shocked and worried, trying to explain something to the people around her with expressive hand gestures. Use realistic facial expressions, natural body movements, cinematic camera angles, shallow depth of field, realistic lighting, detailed modern architecture, and smooth transitions. Make the entire video photorealistic, emotionally dramatic, consistent character appearance, natural skin texture, realistic clothing and environment, with high-quality cinematic storytelling and no distorted faces or extra fingers.

Created a 15-second cinematic photorealistic video of a peaceful alpine countryside during golden morning light, with majestic snow-covered mountains rising in the background and a cozy wooden farmhouse surrounded by rustic fences and open fields. A beautiful young woman with long dark hair, wearing a warm cream knitted sweater and natural winter outfit, slowly walks toward a calm brown horse standing near the wooden stable. She gently approaches the horse, smiles softly, and lovingly strokes its head while the horse remains peaceful and relaxed. The camera captures intimate close-up shots of her hand touching the horse and the horse’s expressive face, followed by smooth medium shots showing their connection. Warm sunlight shines through the mountain peaks, creating soft cinematic highlights and natural lens flare. Light morning mist drifts around the distant cabins and mountains while subtle smoke rises from a farmhouse chimney. The camera slowly tracks beside the woman as she walks back toward the cozy wooden house, keeping her appearance and outfit consistent throughout. Realistic horse movement, natural body motion, detailed skin and hair texture, authentic countryside atmosphere, soft depth of field, cinematic color grading, 35mm lens, gentle handheld movement, ultra-realistic 8K quality, peaceful emotional storytelling, no text, no logos, no watermark.

Create a 15-second fast-paced, smooth luxury fashion editorial video using the character sheet as the visual reference. Keep the same woman, face, brunette hair, fur jacket, red crop top, distressed denim, chains, makeup, and overall styling consistent throughout. Preserve her playful, confident, slightly mischievous personality. The video should feel like a fashion mugshot mixed with a playful “wanted for beauty” campaign. 0–3s — THE MUGSHOT Start with a sharp full-body mugshot composition in front of the height chart. She looks directly into camera. Quick smooth push-in → she gives a playful wink → instantly cuts to her holding the “WANTED FOR BEAUTY” board. She slowly tilts the board toward camera with attitude. 3–6s — BEAUTY DETAILS Move into a rapid but smooth beauty montage: wink → glossy lips → red nails → lollipop → fur jacket texture → chain details → sunglasses → playful smile. Use smooth snap zooms and seamless whip transitions. Each shot should be very short, around 0.3–0.6 seconds, but transitions should remain fluid and polished. 6–9s — PLAYFUL TROUBLEMAKER Return to the full mugshot. She casually leans toward one side while maintaining her confident stance. Quick smooth camera orbit around her → she puts on sunglasses → looks over the sunglasses directly at camera → gives a mischievous smile. Add subtle camera flash bursts like paparazzi photographers are surrounding her. 9–12s — FAST FASHION MONTAGE Rapidly showcase: fur coat → red top → belt and chains → distressed jeans → lollipop → lips → eyes → sunglasses. Use smooth continuous camera movement between each detail, with quick editorial cuts. Newspaper pages briefly fly across the frame, creating transitions. 12–15s — FINAL SHOT Return to the mugshot setup. She holds the board toward camera. The board now reads: “WANTED FOR BEING TOO BEAUTIFUL” She raises one eyebrow and smiles. Camera rapidly pushes toward her face. A newspaper suddenly sweeps across the lens. CUT TO BLACK. Motion & Editing FAST-PACED but SMOOTH. Use: 0.3–0.6 second cuts smooth snap zooms seamless whip transitions controlled camera orbit quick push-ins and pullbacks subtle speed ramps fashion-editorial motion blur paparazzi flash effects flowing newspaper transitions beat-synced editing Do NOT make it slow, dreamy, jittery, or chaotic. The camera should always feel controlled and fluid, while the editing stays fast. Overall feeling: playful troublemaker + luxury fashion campaign + cinematic mugshot + glossy magazine editorial.

A cinematic drawing film follows one artist drawing, completely by hand and from a blank sheet, the fingerprint ridges portrait of the subject in the referenced image. The referenced image is the finished artwork. It defines the final drawing's likeness, composition, position on the sheet, mark style, line directions, density and tonal values; the video must end on exactly this drawing. Ignore anything outside the sheet. Begin directly on a blank sheet of bright white smooth hot-press drawing paper, lying flat on a plain wooden desk, framed horizontal 16:9. Use a stable straight-down overhead view that keeps the whole sheet readable, with brief close inserts that follow the pen tip. There is one artist throughout, represented by the same consistent right hand drawing and left hand resting at the edge of the sheet. Show no more than two hands at once. The sheet never moves, rotates or changes size. From 0 to 14 seconds, the entire drawing unfolds as a clearly accelerated timelapse with rapid, purposeful hand movement and restrained motion blur. Short jump cuts compress repetitive strokes only after each pass has visibly completed. Every cut inherits the exact marks left by the previous pass. 0-1 seconds: One hand enters already holding a pencil and lays down the first stage: a faint pencil outline of the head, features and hair, with small spiral marks showing where the whorl centres will be, no ink. 1-4 seconds: The artist switches to a black fineliner pen and places the first tight concentric loops at the whorl centres on the cheek, the forehead and around one eye. The sheet now shows tight concentric ink loops radiating from the whorl centres on one cheek, the forehead and around one eye, covering about a third of the face; the rest is pencil. 4-8 seconds: The artist extends the ridges outward from the cores until they flow across the whole face, hair and neck, then erases the pencil. The sheet now shows ridge lines extended across the whole face, hair and neck at one uniform thin weight, so the portrait reads flat and low-contrast, with the pencil erased. 8-11 seconds: The artist goes back over the ridges in the shadows with a brush pen, thickening them until the face emerges from the pattern. The sheet now shows ridges thickened with the brush pen in the shadows (eye sockets, under the nose, the far side of the jaw, the hair) so the face emerges from the pattern. 11-14 seconds: The artist merges the darkest ridges, adds breaks and forks in the highlights and final accents to the eyes and lips, until the drawing closely matches the referenced image. 14-15 seconds: The artist lifts the pen away and withdraws both hands. The timelapse returns to normal speed as the camera makes a restrained push toward the finished drawing and holds on a clean final view. Every mark appears only where the pen tip is touching the paper and moving; nothing draws, fills, darkens or fades in by itself. Tools do not need to be visible before use, but any new tool enters the frame already held by the artist's hand. Marks only accumulate; the one exception is the pencil guide, which is erased once the ridge field is complete. Audio: smooth gliding of a felt-tip pen on paper synchronised to the strokes, no music. Maintain one artist, one sheet, one continuous drawing, one working position and one coherent fingerprint ridges technique. No assistants, extra hands, second drawing, photograph, printed image, projection, tracing, colour change, magical morphing, software interface, cursor, annotations, text overlays or signature.

[ENGINE DIRECTIVE] Generate a continuous 15-second hyper-realistic live-action sci-fi action sequence using Seedance 2.0 at 120fps ultra-realism in 2.39:1 anamorphic widescreen. Maintain 100% strict visual consistency with the provided asset sheets: "Agile Dagger Specialist" (Lean female in charcoal Kevlar and dual daggers) and "Armored Juggernaut" (7-foot exo-plated fighter with hydraulic war hammer). Setting strictly follows "Descending Industrial Freight Elevator Shaft". Tone: Claustrophobic, high-impact, kinetic, structural destruction. [TIMELINE BREAKDOWN] 00:00 - 00:05 | CLAUSTROPHOBIC ENCLOSURE & HYDRAULIC IMPACT • VISUAL: Low-angle claustrophobic tracking shot inside the descending freight elevator, framed between rusted iron grates and passing concrete walls. • ACTION: The Juggernaut raises his massive hydraulic war hammer and slams it downward. The Agile Specialist executes a high-speed lateral slide between his legs as the hammer shatters the iron floor grate, sending sparks, rusted fragments, and hydraulic fluid bursting into the air. • CAMERA & SOUND: Violently shaking low-angle camera, rapid whip-pan following the slide. Booming hydraulic pressure release, heavy metal crunch, and screeching elevator cables. 00:05 - 00:10 | CLIMBING COMBAT & WEAPON BLEND • VISUAL: Tight tracking shot moving vertically up the interior corner of the elevator cage. • ACTION: Using the Juggernaut's extended arm momentum, the Specialist steps onto his knee armor, vaults onto his shoulder plate, and strikes twin curved daggers directly into his neck-guard seams. Sparks shower across her focused face. The Juggernaut thrashes, slamming his back against the shaft wall to crush her, but she releases and drops cleanly into a backflip. • CAMERA & SOUND: Tight handheld tracking shot following her upward climb. High-pitched metal scraping, sparks crackling, heavy armor thuds against concrete, and aggressive combat grunts. 00:10 - 00:15 | WALL REBOUND & EXTREME NEAR-MISS • VISUAL: Fast-moving wide angle capturing the full vertical scale of the narrow elevator enclosure. • ACTION: The Juggernaut swings his war hammer in a wide horizontal arc, shearing through two corner steel suspension pillars. The Specialist wall-runs two steps along the passing shaft wall, diving over the swinging hammer head—the weapon missing her boots by centimeters—and plants a double dagger strike into his visor casing. • CAMERA & SOUND: Fast whip-pan catching her mid-air dive, brief shutter speed ramp at the near-miss moment. Shearing steel sound, high-velocity air whiplash, visor glass cracking, and flickering amber strobe pulses.

subject_definitions: <Subject 1> is the young Japanese office worker in <Picture 1>, a man in his mid-twenties with short, slightly messy black hair, a white dress shirt with rolled sleeves, a navy necktie, and an ID badge on a dark lanyard. <Subject 2> is the female senior colleague in <Picture 1>, a woman in her late twenties with brown hair tied up in a loose low bun, a charcoal-gray blazer over a white top, and a friendly, confident expression. <Subject 3> is the visual style of <Picture 1>: Japanese anime-style illustration with clean cel shading, crisp dark line art, soft office lighting, and a bright, clear color palette. summary: [reference generation] A 15-second vertical Japanese anime short. <Subject 1> is overwhelmed by a pile of meeting minutes late at night, then <Subject 2> shows him a single AI instruction on her laptop, and he finishes the work in ten minutes. The story follows the six panels of <Picture 1> in order, rendered as one continuous full-screen video. retention_analysis: <Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3], [Shot 4], [Shot 5], [Shot 6]): fully_preserved - face, black hairstyle, white shirt, navy necktie, and ID badge stay identical in every shot. <Subject 2> (appears in [Shot 3], [Shot 4], [Shot 6]): fully_preserved - face, brown bun hairstyle, gray blazer, and white top stay identical in every shot. <Subject 3> (appears in all shots): partially_preserved - the anime line art, cel shading, and color palette are kept; the storyboard layout itself is not. detailed_description: [Shot 1] Japanese anime style, vertical 9:16 full-screen frame. A chest-up shot of <Subject 1> (S1) at a desk in a dark office late at night, city lights glowing through the window behind him and a round wall clock beside it. He clutches his head with one hand, sweat drop on his temple, in front of tall stacks of paper; the top sheet shows the printed title "議事録". A white speech bubble pops in above him, its tail pointing to his mouth, reading "うわ…" on the first line and "議事録が終わらない" on the second line. The camera pushes in with small amplitude at slow speed as S1, a young man with a slightly hoarse, tired voice, groans: <d>[Japanese] うわ…議事録が終わらない</d> [Shot 2] At 00:02.500, the camera cuts to a medium shot of the same night desk. S1 slumps forward and drops his face onto the paper stacks, a pile of sheets slides slightly, and a small black scribble swirl appears above his head. A white speech bubble pops in at the upper left, its tail pointing down toward his head, reading "まだこんなに" on the first line and "残ってる…" on the second line. Static shot. S1 mumbles into the papers: <d>[Japanese] まだこんなに残ってる…</d> [Shot 3] At 00:04.500, <scenetrans> the scene changes to the same office the next morning, bright daylight and a green plant in the background. <Subject 2> (S2) leans in from the right holding a white mug. A white speech bubble pops in above her, its tail pointing to her mouth, reading "もう終わったよ", and S2, a young woman with a calm, bright voice, says: <d>[Japanese] もう終わったよ</d> S1 turns toward her with wide eyes, and a jagged white shout bubble containing only "えっ!?" pops in beside him as S1 says: <d>[Japanese] えっ!?</d> [Shot 4] At 00:07.000, the camera cuts to an over-the-shoulder shot from behind S1 toward a laptop screen. S2 points at the screen with a smile, and a white speech bubble above her, its tail pointing to her mouth, reads "この一文でOKだよ". On the screen, a navy header bar reads "AIへの指示文", and below it three soft, blurred lines of text appear one line at a time. The camera pushes in with small amplitude at slow speed as S2 says: <d>[Japanese] この一文でOKだよ</d> [Shot 5] At 00:09.500, the camera cuts to a bright shot of S1 raising a fist in joy, light rays bursting behind him. A white speech bubble pops in above him, its tail pointing to his mouth, reading "10分で終わった！". Below the bubble and beside him, a simple bar chart assembles in two stages: first a tall gray bar labeled "3時間" rises on the left, then a short blue bar labeled "10分" appears on the right, and a curved blue arrow travels from the top of the tall gray bar down to the top of the short blue bar, pointing only at the short blue bar. S1 shouts happily: <d>[Japanese] 10分で終わった！</d> [Shot 6] At 00:12.500, the camera cuts to a medium shot of S1 and S2 standing side by side in the bright office, both smiling, S2 holding a closed laptop against her chest. A white speech bubble pops in above S1, its tail pointing to his mouth, reading "今日から," on the first line and "仕事をもっとラクに" on the second line. The camera pulls out with small amplitude at slow speed. The lower third of the frame stays clean and uncluttered. S1 says cheerfully: <d>[Japanese] 今日から、仕事をもっとラクに</d> Hold this final shot until exactly 15.000 seconds. All speech bubbles share one manga style: white fill, thick black outline, rounded shape, bold black Japanese lettering written horizontally, placed in the upper-middle part of the frame and never in the top 15% or the bottom third. Each bubble pops in as its line begins and disappears when the shot cuts. The spoken words match the bubble text exactly. Only these seven lines of dialogue are spoken; no other speech, narration, or ad-libbing. The only on-screen text is the bubble text above plus "議事録", "AIへの指示文", "3時間", and "10分". No subtitles, no captions, no storyboard panel borders, no panel numbers or header labels, no split-screen, no watermark. overall_soundscape: A quiet office hum and a ticking wall clock fill the night shots, with paper rustling as the stacks shift and a soft thud when he slumps onto them. In the morning, gentle office ambience plays, a mug clinks lightly, and a comic pop accompanies the surprise. Soft keyboard taps and a small UI chime sound as the text lines appear, followed by a bright whoosh and a rising chime as the chart assembles. non_diegetic_music: A light piano and pizzicato-string cue at a moderate tempo, sparse and low in the two night shots. It turns brighter with plucked synth and soft hand percussion from 00:04.500, builds slightly during the chart at 00:09.500, and resolves on a clean major chord at 15.000 seconds.

ルール： - 予算5億円規模の劇場アニメ品質・日本の2Dセル画アニメ - マルチショット - 高速カット割り・作画枚数多め。スピード感と迫力を最優先 - VFXプロ品質 - カメラワークは毎ショット変える（同じ動きの繰り返し禁止） - BGMなし・環境音/効果音あり - 字幕なし - セリフは全て日本語 - 緩急：静かな芝居は間をとってじっくり見せ、見せ場で一気に動かす - 雨は細く速い線で描き、しぶき・水たまりの反射・濡れて束になった髪まで作画で描きこむ - 夜の雨。街灯やネオンの色を濡れた路面に映し、淡い水色と白のキャラが背景に溶けて灰色に沈まない配色にする - 主役は常に画面の大きな位置に置き、顔と衣装がはっきり見える。遠景で豆粒のように小さくしない - 画面に文字・ロゴ・数字を一切出さない @ Image1：被写体・シルヴィ @ Image2：被写体・アイル [一貫性] シルヴィ（白銀のボブで右目が前髪に隠れ、見えている目は黄緑。水色の悪魔の角とフリルのヘッドドレス、水色の小さなコウモリの翼と細いしっぽ、真っ白なフリルドレスに水色のリボン、水色と白のしましまの靴下）とアイル（青いぱっつんボブ、頭の両側に白い丸い羽根飾り、肩に青と白の小さな翼、白地に青い丸模様のドレス、後ろに長く垂れる青い羽根のすそ）の外見を全編で一貫させ、2人を混ぜない。アイルは雲の上の天使、シルヴィは地上の悪魔 [生成ゴール] 雨の日だけ、会える（6ステージ・1本の連続動画。雨の糸を伝って会いに来て、雨がやむまで） [S1] 雲の上。@ Image2のアイルが、雲のすき間から落ちていく雨の糸をつかみ、すべり台のように地上へすべり降りていく（うれしくて待ちきれない顔）。カメラ：雨の糸に沿って、すべり降りるアイルを真上から追う高速ドリー [S2] 雨の夜の町の屋根の上。@ Image1のシルヴィが、傘もささずに空を見上げて待っている。雨粒が水色の角とコウモリの翼を伝って落ちる。カメラ：屋根の低いアングルから、空を見上げるシルヴィの横顔 [S3] アイルがふわりと屋根に着地し、2人が向かい合う。シルヴィは照れてそっぽを向くが、しっぽの先がうれしそうに揺れている。カメラ：2人を横から並べるツーショット。しっぽの揺れが見える引きアイル（はずむ声で）は言う：{きたよ} [S4] 2人は屋根の上に並んで座り、シルヴィがコウモリの翼を傘がわりに広げてアイルの上にかざす。雨の町の灯りが2人の下でにじんで光る。カメラ：2人の背中越しに、雨の町の夜景を見下ろす [S5] 雨が弱まり、雲の切れ目から月の光が差す。アイルの体が光に引かれてふわりと浮き上がり、空へ戻っていく。手を伸ばすシルヴィ。2人の指先が離れる。カメラ：離れていく指先の寄りから、見上げるシルヴィの顔へアイル（泣き笑いで）は言う：{つぎの雨の日に} [S6] 雨の上がった屋根の上に、シルヴィがひとり。手のひらに、アイルの青い羽根が一枚残っている。シルヴィは羽根をぎゅっと握り、空を見上げて小さく笑う。カメラ：羽根の寄りから、夜空を見上げるシルヴィの引きへ

Create a 31-second ultra-realistic cinematic action sequence set in a destroyed modern city, featuring a powerful female superhero in a sleek black futuristic armored suit with glowing red details, running through a city street as a massive explosion erupts behind her, then transition into a dark urban passage where she moves at high speed with dramatic camera tracking, powerful mechanical wings emerging from her armor as she prepares for battle, followed by intense close-up shots of her fighting a huge monstrous creature with glowing red eyes and massive strength, show fast aerial movements, powerful punches, kicks, sparks, smoke, flying debris and realistic impact effects, then capture her flying through the city with large black mechanical wings glowing with red energy before landing dramatically on the damaged road, maintain consistent character appearance, armor design, face, wings and creature throughout the entire video, with dark blue cinematic lighting, red highlights, realistic destruction, dynamic camera movement, dramatic slow motion moments, detailed textures, volumetric smoke, cinematic depth of field and high-end Hollywood-style visual effects, ending with the superhero standing firmly in the destroyed city as smoke and debris move around her.