Prompt Video cho Documentary

Three.jsで1分のホラー映像を作って。題材は〇〇。見つかったビデオテープの最後の一分という設定で、手持ちカメラの主観、自動で進む。音も含めて、画像や音声ファイルは使わず全部コードで

A woman in her 40s sits alone at a small dining table at night, reading a letter. Over 20s, her rage slowly curdles into grief before breaking into physical destruction. A woman mid-40s, natural skin with visible pores, fine lines and under-eye texture, shoulder-length hair loosely tied back, worn cardigan over a plain top. 100% reference match. Small wooden table, warm lamp camera-left, letter in foreground, shelf with framed photos and glass vase behind her, shadowed corner of the room. Timed multishot: continuous 0–12s, then HARD CUTS only at 12s, 14.5s and 17s. 0–12s MCU→CU, FOV 29°, slow creeping handheld push-in with fine 1cm tremor. 12s WS FOV 63°, 14.5s MS FOV 47°, 17s CU FOV 18°. Final shot is low and static from floor level. 0–3s: eyes scan letter, jaw tightens, grip whitens knuckles. 3–6s: eyes glass over, one tear rolls down her right cheek, breathing shortens. 6–9s: hands tremble, paper shakes, shoulders draw inward. 9–10s: she reads the final line, sharply exhales and drops the letter. 10–12s: hands drag through her hair, head tilts back, first sob breaks. 12s: she violently shoves the table forward and rises. 14.5s: she sweeps framed photos and vase from the shelf; glass and frames scatter realistically. 17s: she sits collapsed in the corner, knees drawn up, crying, one hand over her mouth, the other loosely holding a kitchen knife resting against her knee, never raised. Keep the performance restrained and realistic: micro-tremors, wet reflective eyes, visible catch-light, natural flushing, no exaggerated crying. Realistic paper movement, table weight/friction, glass impact and body weight. Single 3200K practical lamp throughout; no new light sources. Diegetic audio only: paper, breathing, sobs, table scrape, lamp rattle, glass breaking, floor settling. No score. Photoreal, natural skin texture, fine film grain, handheld documentary intimacy, no stylized color grade, real-time, 24fps. Hair, cardigan, lighting and character remain consistent across every cut.

Create a 30-second realistic fixed security-camera recording inside an ordinary apartment building hallway at night. IMPORTANT: This is a SINGLE CONTINUOUS STATIC CAMERA SHOT. The camera never moves, never zooms, and never changes angle. The hallway has an elevator clearly visible at the CENTER of the frame. The elevator has two clearly visible sliding doors. The man must walk directly toward the elevator doors, stop in front of them, enter the elevator through the open doors, and then the elevator doors must close. Do NOT let the man walk into a wall. Do NOT change the hallway layout. Do NOT move the elevator. Do NOT introduce any other person. VISUAL STYLE: Authentic low-quality apartment CCTV footage. Fixed ceiling-mounted security camera. Wide 16:9 frame. Slight fisheye distortion. Low-light digital noise. Mild compression artifacts. Flat fluorescent hallway lighting. Small timestamp in the corner. No cinematic camera movement. No dramatic color grading. No horror-film effects. TIMELINE: 0–7 seconds: The hallway is completely ordinary and empty. A man enters the frame from the far end and casually walks toward the elevator in the CENTER of the hallway. His normal human shadow is clearly visible on the floor beneath and behind him. 7–12 seconds: The man reaches the elevator. The elevator doors OPEN clearly. The man walks through the OPEN elevator doorway and stands INSIDE the elevator. He is completely visible inside the elevator. The elevator doors then CLOSE completely. 12–16 seconds: The hallway is empty. The closed elevator remains completely normal. The man's shadow is no longer visible because he is inside the closed elevator. Nothing happens for a few seconds. 16–20 seconds: A subtle anomaly begins. A dark human-shaped shadow slowly appears from underneath the CLOSED elevator doors. There is NO PERSON outside creating this shadow. The shadow is clearly detached from any physical person. It slowly extends farther across the hallway floor. 20–25 seconds: The detached shadow continues moving away from the elevator across the floor. It moves like the shadow of a person walking, even though there is NO PERSON anywhere in the hallway. The elevator doors remain CLOSED the entire time. The physical environment remains completely unchanged. 25–30 seconds: The detached shadow reaches the middle of the hallway. It suddenly stops. Hold on the empty hallway with the impossible shadow visible on the floor. Then abruptly CUT TO BLACK. ABSOLUTE PRIORITIES: 1. The man MUST enter the elevator, not the wall. 2. The elevator must be clearly visible and centered. 3. The elevator doors must visibly OPEN before he enters. 4. The man must be completely INSIDE the elevator before the doors close. 5. The elevator doors must remain CLOSED while the detached shadow emerges. 6. The detached

A cinematic, realistic short film of a young East Asian woman with wavy shoulder-length dark hair living a rushed, exhausted daily routine. Muted earthy tones, slightly grainy 35mm film look, natural lighting, handheld camera, quiet and melancholic mood. Shot sequence: - Close-up of a smartphone on a worn marble nightstand showing 06:50 AM. - She jolts awake in floral pajamas, eyes wide, messy hair. - Washing her face in a dim tiled bathroom, wet hair hanging down. - Dark empty hallway. - Reaching into a crowded fridge for a container. - Eating a sandwich while throwing on a green-and-maroon hoodie. - Putting on grey New Balance sneakers. - Walking down a dim hallway with a beige backpack. - Crowded subway car with motion blur. - Standing on the train looking at her phone. - Walking up a long tiled subway staircase. - Changing into a white button-down shirt. - Working late on a laptop in an open office. - Close-up of a messy spreadsheet. - Wide shot of a dark empty office at night; she is the last one there. - Walking out through the empty office. - Collapsing onto her bed exhausted. - Sleeping on her side in pajamas. - Final black screen: “18:50 AM”. Style: grounded documentary feel, natural skin texture, lived-in spaces, 16:9, 24fps, cinematic color grade.

Create a 30-second vertical 9:16 ultra-photorealistic miniature mountain village construction time-lapse. CORE: An empty miniature spring valley is physically transformed into a beautiful countryside village by hundreds of tiny realistic workers. Real handcrafted miniature diorama look — macro photography, natural sunlight, realistic soil, stones, wood, grass, flowers, water, shadows and shallow depth of field. NOT CGI, NOT cartoon, NOT fantasy. No magical appearances, teleportation, instant buildings, instant trees or instant landscape changes. Every object must be physically constructed. 0–2s — FOUNDATION Empty mountain valley. Hundreds of workers arrive, mark plots with ropes/stakes, dig foundations and mark the future road. 2–4s — ROAD Workers excavate, level and compact a winding dirt-and-stone road while others carry gravel and place roadside stones. 4–6s — HOUSE FOUNDATIONS Multiple teams simultaneously build house foundations using individual stones and wooden beams. 6–8s — WALLS Stone masons stack walls layer-by-layer while carpenters assemble wooden frames. Houses visibly rise but remain unfinished. 8–10s — ROOFS Workers lift roof beams, install roof tiles, doors and windows using ladders and tiny carts. Several cottages become recognizable. 10–12s — MORE HOUSES Additional cottages are rapidly constructed farther up the slope: foundations → walls → frames. Workers also build retaining walls. 12–14s — GRASS & GARDENS Workers spread soil, add grass, plant flowers and assemble small wooden garden fences around the cottages. 14–16s — CHERRY BLOSSOMS Workers transport young trees, dig holes, plant and secure them. Through natural time-lapse, branches develop and pink-white blossoms appear. No popping into existence. 16–18s — FARMING Farmers prepare soil, plant vegetable rows and construct tiny irrigation channels. Baskets and farming tools remain visible. 18–20s — STREAM & BRIDGE Workers excavate a streambed, place stones and create flowing clear water. Carpenters build a wooden bridge: supports → beams → planks. 20–22s — PATHS & FENCES Stone paths connect the houses. Workers install fences, stone walls and pathways throughout the village. 22–24s — VILLAGE LIFE Tiny realistic goats, sheep, cows and chickens appear through natural movement into the completed environment. Farmers carry baskets, push carts and work between homes and fields. 24–26s — FLOWERS & DETAILS Hundreds of gardeners finish the landscape with spring flowers, grass, flower pots, benches and farming tools. 26–28s — COMPLETE LIFE Busy village: farmers working, villagers walking, animals moving, water flowing, cherry blossoms swaying and subtle chimney smoke. All construction remains physically consistent. 28–30s — GRAND REVEAL Start close to one tiny farmer. Rapidly pull backward and upward to reveal the entire handcrafted village: dozens of stone-and-wood cottages, winding roads, green hills

Pheasants and red foxes chase the black SUVs along the winding Cotswolds country road. Birds flutter and fly low, foxes sprint and leap kicking up dust, vehicles drive slowly forward through the rolling green hills with stone walls under cloudy sky....

Italian amphitheater, tour guide to camera: “Qui cantavano migliaia di persone.”

做一支史诗编年片《中华文明史》，可以写代码逐帧渲染，再用 ffmpeg 合成。 配乐当时钟：五声调式，乐器从骨笛、编钟一路演进到管弦，BPM 随年代加快，所有切点踩拍。 宣纸白描和玄底泥金两种画风交替；每卷一个主色和一套随时代演变的纹样（彩陶纹 → 饕餮纹 → 云气纹 → 卷草纹 → 缠枝纹 → 回纹）。 每镜一个按词组出现的书法大字关键词，配一幅线稿。全片 HUD：左上朱印卷号，右侧竖排朝代名，底部卷轴时间尺和年份计数。 卷交界用朱印盖下、鼓钟重击的冲击转场。先定拍点网格和分镜表，再渲染。 地图只画示意，不画近现代真实人物，年代核对后再交付。

Create a 30-second ultra-realistic personal home-video of a young Korean woman enjoying a relaxed Friday morning in an older Seoul residential neighborhood. Use the attached image as the character reference and keep her face, long black messy side ponytail, pastel-blue fitted top, loose cream pants, black sneakers, silver necklace, and overall appearance perfectly consistent. She leaves her home with a cheerful expression and walks casually through the quiet neighborhood. She reaches a nearby outdoor water tap, stops, splashes cool water onto her face, wipes her face with her hands, smiles and continues walking. She arrives at a small neighborhood shop and buys a colorful lollipop for herself. She unwraps it immediately, puts it in her mouth and happily enjoys the candy while walking through the street. A few school children pass by carrying their backpacks. She smiles, waves at them and casually says, “Hello!” as they walk past. She then approaches a newspaper seller outside a small local shop, picks up a newspaper, gives him a few coins, thanks him with a smile and starts walking back toward home while holding the newspaper. Near the end, she looks toward the camera with a happy smile, still enjoying her lollipop, and says, “Happy Friday!” before continuing home. Use raw early-2000s consumer DV-camera footage: handheld shake, imperfect framing, autofocus hunting, exposure shifts, soft digital detail, mild digital noise, natural motion blur, occasional awkward zooms and authentic home-video imperfections. Natural Seoul neighborhood ambience only — footsteps, running water, children talking, shop sounds, newspaper rustling, distant traffic, birds and summer insects. No music, no narration, no subtitles, no polished commercial cinematography, no beauty-filter skin, no CGI look, no duplicate characters, no changing face, hairstyle or outfit.

Create a 30-second, 1080p, ultra-photorealistic historical documentary video using Seedance 2.5. Depict an ordinary day in London during the construction of the medieval stone London Bridge, approximately between 1176 and 1209. The video must look like genuine, naturally observed footage from that era, as if an invisible documentary camera had somehow captured real medieval life. No AI-generated appearance, no fantasy aesthetics, no overly polished cinematic visuals. VIDEO SPECIFICATIONS: Duration: 30 seconds Resolution: 1080p Aspect ratio: 16:9 Style: Authentic historical realism, observational documentary Camera: Unseen camera operator, natural handheld movement Lighting: Realistic daylight, natural atmospheric haze Audio: Period-appropriate environmental sounds, workers, carts, river, birds, and human conversations. No modern music. HISTORICAL SETTING: London, England, during the construction of the first major medieval stone London Bridge. The year should feel approximately 1190–1200, with the bridge still incomplete. Use historically plausible medieval architecture, clothing, tools, construction methods, and daily life. The bridge is being constructed across the River Thames under the broader project begun in 1176 and completed in 1209. The construction should visibly involve stone masonry, wooden scaffolding, timber piles, workers moving stone, ropes, hand tools, and boats transporting building materials. Avoid modern construction equipment, industrial machinery, cranes, safety helmets, modern roads, or anachronistic architecture. SCENE 1 — 0–6 SECONDS: THE BRIDGE CONSTRUCTION Open with a natural medium-wide view from the riverside, showing the partially constructed stone London Bridge dominating the frame. Several workers are actively building the bridge: masons shaping stone blocks with hand tools, laborers hauling materials, and workers standing on rough timber scaffolding. Wooden piles and temporary construction platforms extend into the Thames. The camera moves slowly and naturally, as though an unseen person is observing the work from the riverbank. Workers behave independently, concentrating on their tasks. Some pause to communicate, while others carry stones and operate ropes. The unfinished arches and rough stone surfaces must look physically authentic and historically plausible. The Thames flows naturally beneath the construction area, with wooden boats passing nearby. Dust, small stone fragments, wood textures, and subtle movement create a sense of a real working construction site. SCENE 2 — 6–12 SECONDS: PEOPLE GOING ABOUT THEIR DAY The camera shifts naturally toward the nearby riverside surroundings without a hard cut. Medieval London residents go about their everyday routines. A woman carries a woven basket, a man transports goods on a wooden handcart,

15-second ultra-realistic early-2000s Sony MiniDV home video, filmed casually by another person. Subject: Young Korean woman, 24, naturally attractive, realistic skin, minimal makeup, long dark hair loosely down. Wearing an oversized maroon hoodie and loose casual pants, carrying a simple canvas shoulder bag. Preserve her exact identity, facial features, hairstyle, body proportions and appearance throughout. Location & Atmosphere: Small old Seoul apartment during a dark, rainy early morning. Cozy bedroom connected to a small entryway, simple bed with a soft blanket, wooden furniture, rain-covered window showing blurred neighboring apartment buildings outside. Dark blue-hour ambience, cold blue-gray light from the rainy window mixed with weak warm interior light. Muted colors, soft shadows, slightly underexposed surroundings, steady rain outside. Quiet, sleepy, intimate atmosphere. Style: Authentic early-2000s Sony MiniDV home-video footage filmed by another person. Completely candid and unstaged. Natural handheld movement, subtle camera shake, imperfect framing, gentle reframing, occasional autofocus hunting, mild exposure shifts, faded colors, soft contrast, authentic DV compression, subtle low-light digital noise and microphone noise. Smooth continuous real-time motion throughout. No stuttering, judder, frame skipping, duplicated frames, stop-motion appearance, excessive motion blur, speed changes, low-frame-rate look, stabilization or modern cinematic movement. 00:00–00:04: She sits on the edge of her bed, sleepy and slightly groggy, wearing her oversized maroon hoodie and loose casual pants. She stands up and reaches for her canvas shoulder bag near the doorway. 00:04–00:07: She takes a couple of steps toward the door, but the blanket is still loosely wrapped around one ankle. It gently pulls her foot backward. She stops and looks down, confused. 00:07–00:10: She tries taking another step. The blanket catches again and slides across the floor with her. She looks down at it with a mildly annoyed expression, then bends down and tries to free her ankle. 00:10–00:12: She finally gets her foot free, stands up and takes one confident step toward the door. 00:12–00:15: The blanket catches around her ankle one more time and gently pulls back. She freezes, looks down at it, then slowly looks directly at the camcorder with a completely deadpan expression. After a tiny sigh, she gives the blanket a small kick away and walks off. Audio: Natural sound only—steady rain against the window, quiet apartment ambience, soft footsteps, fabric and blanket rustling across the floor, subtle clothing movement and her natural sigh. No music, narration or added sound effects. Goal: Cute, relatable and lightly funny, like an ordinary sleepy morning accidentally captured on an old family MiniDV camera. The humor should come entirely from the blanket repeatedly refusing to let her leave. Her final deadpan look toward the camera is the comed

Create a 30-second ultra-photorealistic live-action cinematic adventure video following a young woman crossing a shallow mountain river by carefully jumping between natural stepping stones. Show a realistic outdoor landscape with clear flowing water, moss-covered rocks, wet gravel, grass, trees, distant mountains and small rapids. Keep the woman, clothing, backpack, terrain and environment perfectly consistent throughout. Capture authentic human biomechanics, jumping, balance, momentum, center-of-mass shifts, foot placement, friction and realistic landing mechanics. Include a naturally unstable rock that shifts slightly under her weight, forcing her to regain balance before continuing across the river. Make the river behave with realistic fluid dynamics: continuous downstream flow, water splitting around rocks, small eddies, ripples, turbulence, droplets and proportional splashes. Wet rocks should have reduced traction, while gravel and soil respond naturally to her footsteps. Use modern cinematic adventure-documentary cinematography with handheld trail-following shots, rock-level close-ups, side tracking during jumps, realistic autofocus, natural camera inertia and a wide landscape reveal at the end. Natural diegetic audio only: rushing water, footsteps, gravel, shoe impacts, small splashes, wind, birds, breathing and clothing movement. No music, narration or subtitles. End with the woman safely reaching the opposite bank, adjusting her backpack and looking back across the river as the camera slowly pulls away to reveal the surrounding mountain landscape. No CGI, animation, fantasy, superhero movement, impossible jumps, floating objects, teleportation, exaggerated splashes, distorted anatomy, changing clothing, impossible camera movement, logos or watermark.

30-second comedy short. Two realistic house cats, a chunky orange tabby and a skinny black-and-white tuxedo cat, sit side by side on a dim rooftop ledge at night, city lights blurred behind them. Warm orange and blue lighting, shallow depth of field, mockumentary-style camera with sudden zoom-ins for comedic timing. Their mouths move naturally in sync with speech. They whisper in sad-angry, deadpan voices like two exhausted men at a bar, constantly glancing over their shoulders. Burned-in white English subtitles at the bottom of the frame, synced to speech. 0-5s: Close-up of the orange tabby, ears flat, one eye twitching. Long, dramatic sigh. Orange (whisper): "Bro... she woke me up at 5 AM to ask if I love her." Subtitle: "Bro... she woke me up at 5 AM to ask if I love her." Quick zoom-in on his tired face. 5-10s: Cut to the tuxedo, tail whipping, slowly shaking his head. Tuxedo (whisper): "Mine knocked a glass off the table, looked me dead in the eye, and blamed ME." Subtitle: "Mine knocked a glass off the table, looked me dead in the eye, and blamed ME." Orange nods slowly in the background: "Classic." 10-15s: Two-shot, both staring at the skyline. Orange (whisper): "She says I never listen. I'm a CAT. I don't listen to anybody." Subtitle: "She says I never listen. I'm a CAT. I don't listen to anybody." Tuxedo: "Respect the honesty, though." 15-20s: Tuxedo leans in, furious whisper, eyes wide. Tuxedo (whisper): "She told me to 'use my words.' I HAVE four words. All of them are 'feed me' and 'move.'" Subtitle: "She told me to 'use my words.' I HAVE four words. All of them are 'feed me' and 'move.'" Orange wheezes silently, shoulders shaking. 20-25s: Both go quiet. Orange stares into the distance. Orange (whisper): "That's it. I'm done. Tonight I'm leaving her." Tuxedo (whisper): "Finally. Say it again, louder." Orange (louder, voice cracking): "I'M LEAVING HER!" Subtitles match. 25-30s: A distant female voice calls out: "Dinnerrr!" Both cats' ears shoot up instantly. They stare at each other, then bolt off the ledge in opposite directions. Cut to the empty ledge, one tiny "meow" echoes. Orange (off-screen): "Coming, babe!" Tuxedo (off-screen): "Coming, honey!" Subtitles match. Fade to black. Style: photorealistic, 4K, cinematic color grade, subtle city ambience, light comedic sound effects (record scratch on the "four words" line, tiny whoosh on the bolt-off). No cartoon look, no text other than subtitles.

给我制作一个介绍纽约的视频，视频素材和背景音全部找商用免费的平台, 配音请使用edge-tts，选择男音，字幕和声音都是中文，时长５分钟左右，这个视频将用来吸引新加坡人去纽约旅游

Create a 30-second, 1080p ultra-realistic personal home-video showing a young Korean woman visiting a traditional countryside farm and learning how to milk a cow. No reference image. MAIN SUBJECT: A naturally pretty Korean woman in her early 20s, realistic skin texture, minimal makeup, long black hair tied in a loose low ponytail. She wears a simple pale-yellow cotton T-shirt, relaxed blue jeans, white sneakers, and a small beige crossbody bag. Keep the same face, hairstyle, clothing, body proportions, and appearance throughout. SETTING: A small traditional Korean countryside farm on a warm summer afternoon. Old wooden barn, cows, hay, buckets, farming tools, green fields and simple farmhouse surroundings. Everything should feel authentic, ordinary and lived-in. CAMERA: The camera is ALWAYS IN FRONT OF THE GIRL, never behind her. A friend films her while standing or walking backward in front of her using a handheld smartphone/early-2000s consumer DV camcorder. Her face stays clearly visible. Strong natural handheld shake, imperfect framing, autofocus hunting, exposure changes, slight motion blur, soft digital detail and mild image noise. NO cinematic camera movements, NO gimbal, NO drone, NO rear shots, NO polished commercial look, NO beauty filter. TIMELINE 0:00–0:05 — Entering the Barn The girl walks toward the camera with her friend filming from directly in front. She looks excitedly at the cows and smiles. She says in Korean: “오늘 소 젖 짜는 걸 배워볼 거예요.” 0:05–0:10 — Getting Ready Hard cut. She sits beside a calm cow while an experienced farmer demonstrates how to position the bucket and gently milk. The camera stays in front, showing her face and hands naturally. 0:10–0:15 — Awkward First Attempt She tries milking herself. At first she squeezes incorrectly and barely gets any milk into the bucket. She looks confused, then laughs at herself. “어? 왜 안 나오지?” 0:15–0:21 — Trying Again Hard cut. The farmer calmly shows her the correct hand movement. She tries again while the friend films from the front. This time milk starts flowing into the bucket. Her face immediately lights up. 0:21–0:26 — She Gets It She continues successfully, looking proudly toward the camera. She laughs and says: “됐다! 진짜 된다!” 0:26–0:30 — Happy Ending She stands beside the cow holding the small bucket and smiles at the camera. She gently pats the cow and gives a little wave. “오늘 진짜 재밌었어요. 안녕!” Natural cut to black. AUDIO: Only real farm sounds: cow mooing, bucket sounds, footsteps, birds, insects, distant farm activity and natural conversation. No music and no narration. OVERALL STYLE: It must look like a genuine friend-recorded memory from an ordinary farm visit—not a movie or advertisement. Front-facing coverage throughout, with natural reactions and imperfect handheld footage.

ultra-realistic viral smartphone video inside a packed Japanese 4DX cinema, filmed as ONE continuous handheld iPhone 17 Pro shot with realistic HDR, low-light grain, micro-shake, motion blur, and exposure shifts. The cinema has illuminated burgundy walls, red seats on both sides, ceiling spotlights, and a real metal train track running down the center aisle toward the screen. Use the uploaded references for scene and character consistency, and match the reaction reference composition only for the final shot. 0–3s: Camera begins already panning right-to-left before naturally settling. The auditorium is warmly lit, screen blank, audience holding phones, tracks visible. 3–7s: Camera locks completely still. The screen transforms into a dark subway tunnel perfectly aligned with the real theatre walls and tracks, creating a seamless illusion with no visible screen boundary. A tiny white light appears deep inside. 7–10s: The light splits into two headlights as a metro train rapidly approaches. Headlights illuminate the real tracks, mist forms, audience tenses, and the train grows until it fills the screen. 10–12s: A violent wind blast hits as the train arrives. Camera involuntarily jerks left and downward in one chaotic arc with heavy motion blur. The train suddenly appears racing beside the left seating section, its blue-white windows streaking past. 12–15s: Camera settles into the exact reaction-reference angle. Foreground: young South Asian man, 22–24, lean, short textured dark hair, light stubble, gold chain, navy blazer over open-collar white shirt. His hair is windblown as he laughs in shock, initially covering his mouth. White cup with straw in cupholder, blue bottle in his other hand. Young woman behind him laughs with disheveled hair. The train continues thundering past at full speed beside them, rapidly strobing blue-white window light across his face and gold chain. He lowers his hand, looks wide-eyed into the camera for a split second, overwhelmed and exhilarated. End abruptly while the train is still racing past at maximum speed. Physics must remain realistic: one unified wind direction, natural hair/clothing movement, consistent train speed, realistic camera inertia, and synchronized window-light flashes. The tunnel must connect seamlessly with the physical theatre geometry. Audio: cinema ambience, building distant rumble, train horn, then deafening steel wheels, rushing wind, screams, and laughter with realistic Doppler shift. No music, captions, logos, UI, or visible filming phone.

Created a video, featuring beautiful Korean words and a warm, nostalgic Korean countryside lifestyle style. Show a loving Korean mother and three young children enjoying a peaceful summer evening in a traditional rural Korean village, with warm golden-hour sunlight, old tiled-roof houses, stone walls, green rice fields, and a calm countryside atmosphere. Begin with the mother gently opening an old wooden door, then transition into a cinematic tracking shot of her riding a bicycle with the children happily cycling alongside her through the village road. Capture natural childhood laughter, soft bicycle movements, flowing clothes, and authentic family interactions with realistic Korean facial features and expressions. Move into wider cinematic shots of the family cycling beside lush rice fields and a small stream, creating a peaceful sense of freedom and nostalgia. Show the mother sitting beside the children near the field, sharing fresh fruit and snacks while they smile and talk together. Use beautiful natural lighting, soft sunset glow, subtle film grain, shallow depth of field, realistic textures, and smooth cinematic camera movement. End with a wide emotional shot of the family sitting together beside the quiet stream under a large tree as the sun sets behind the rice fields. Photorealistic live-action, cinematic Korean family film aesthetic, warm colors, gentle pacing, authentic emotions, 16:9 composition, highly detailed, natural physics, no artificial-looking faces.

【基础设定】 15秒，9:16竖屏，一镜到底，真人纪录片摄影风格。记录一名男子从直升机向火山口投下巨型肥皂，随后大量肥皂泡从火山升起，涌入机舱的完整过程。 以现场观察的方式呈现超现实事件，画面自然、克制，人物反应真实。没有喜剧表演、搞笑动作或刻意设计的包袱。全程同一个连续镜头，无切镜、转场、隐藏剪辑、慢动作或时间跳跃。 【人物与道具】 一名成年男子，穿深色户外夹克、长裤和靴子，腰部安全带连接机舱固定点，位于打开的直升机侧舱门内。 男子双臂抱着一块约80厘米长、45厘米宽、25厘米厚的巨型粉色肥皂。圆角长方体，表面光滑，带少量湿润反光，具有明确厚度与重量。无包装、无文字。 直升机内部可见真实座椅、固定把手、金属门框和地板。驾驶员留在前方驾驶舱。人物、机舱与道具外观全程保持一致。 【环境与摄影】 白天，自然光。深色粗糙岩壁围绕宽阔火山口，底部是缓慢翻滚的橙红色岩浆，空气中存在轻微热扰动。 直升机悬停在火山口边缘上方，侧舱门朝向火山内部。构图清楚交代男子、舱门与下方落点的位置，投掷高度与实际下落时间保持一致。 摄影机始终由机舱内的拍摄者持握，在侧舱门附近完成前倾俯拍、抬镜和回转。广角纪实构图，带有旋翼引起的细小振动与自然手持调整，动作清楚可辨。自然曝光与色彩，保留皮肤、衣料、岩壁和金属的真实纹理。 【连续时间序列｜0:00—0:04｜投下巨型肥皂】 从机舱内部朝侧舱门拍摄。男子和巨型肥皂占据近景，下方火山口透过舱门清楚可见。 男子把肥皂下缘抵在门槛，双手扶稳两侧，屈膝前移重心，将肥皂向外推出。双臂随推送伸直，肥皂重心越过门槛后，双手明确松开。 肥皂脱离机舱，带着轻微翻转向下加速坠落。男子收回双手，扶稳机舱把手，留在原位观察。 摄影机自然前倾，沿肥皂轨迹连续向下俯拍，画面边缘保留一小段门框。肥皂在纵深中逐渐变小，接触下方岩浆，激起局部翻动。完整呈现下落过程，不跳过距离，不突然切成落点特写。 【连续时间序列｜0:04—0:09｜泡沫生成，大量泡泡升起】 摄影机保持俯拍。肥皂接触岩浆的位置出现细密白色气泡，随后泡沫持续膨胀，逐渐覆盖周围岩浆。 透明肥皂泡从翻涌的白色泡沫中连续鼓起、脱离，随上升气流向直升机方向飘来。最初只有零星几颗，随后数量迅速增加，形成具有前后纵深的密集泡泡群。 泡泡大小从葡萄到篮球不等，表面具有薄膜反光和微弱虹彩，可以透过泡膜看见火山与岩壁。上升速度略有差别，彼此碰撞、变形，部分自然破裂，后方继续生成。 摄影机持续观察泡泡接近的过程。远处火山仍能从间隙中看见，近处泡泡逐渐占据更多画面。岩浆上的白色泡沫与升空的透明泡泡具有明确区别。 【连续时间序列｜0:09—0:13.5｜泡泡抵达，涌入机舱】 泡泡升至舱门高度，被旋翼与机身周围的气流卷动。一部分破裂或偏离，更多从侧舱门外持续涌入。 摄影机跟随泡泡连续抬起，平滑转回机舱，重新拍到男子。拍摄位置和门框方向连续一致，不能突然变成机外视角。 男子一只手握住固定把手，身体略微向舱内退让，另一只手自然抬到面前挡开贴近眼睛的泡泡。他主要观察眼前变化，不看镜头、不夸张表演、不说话。 泡泡从舱门向机舱内部不断扩散，挤在座椅、男子肩膀和摄影机之间。近处泡膜轻轻变形、滑动，破裂后在头发、衣袖和座椅上留下湿润白沫。更多泡泡紧接着填入，机舱结构和人物轮廓逐渐被遮挡。 【连续时间序列｜0:13.5—0:15｜机舱被泡泡填满，直接结束】 摄影机保持当前机舱视角，仅随拍摄者身体轻微后退。 大量泡泡已经挤满舱门和人物周围，男子上半身几乎被遮住。泡泡仍在自然拥挤、漂动和破裂，透过局部间隙能看见深色衣袖与握住把手的手。 到此直接结束，不再增加拨开泡泡露脸、看向镜头或再次被覆盖的动作，不定格、不拉远。 【声音】 真实旋翼声与舱门风声贯穿全程，配合肥皂摩擦门槛的声音、远处低沉的火山翻动声，以及泡泡进入机舱后细微密集的破裂声。声音距离随摄影机朝向自然变化。无背景音乐、旁白、台词或喜剧音效。 【关键约束】 9:16竖屏，15秒，全程一镜到底，真人纪录片摄影质感。摄影机始终位于同一架直升机内部的侧舱门附近。 只有一块巨型粉色肥皂，投出后不再出现在男子手中。泡泡先从火山生成，再连续升至舱门并涌入机舱，不能凭空出现。没有单个包裹直升机的巨型泡泡。 透明泡泡具有清楚的薄膜结构，白色泡沫由细密气泡组成，避免表现成烟雾、棉花或雪。末尾仅用1.5秒呈现男子上半身几乎被泡泡遮住的状态，随即结束。无字幕、标志和水印。

Make a cinematic 2-minute video on the history of artificial intelligence, titled "{argument name="video_title" default="Building the Brain: A History of Artificial Intelligence"}". Make it epic and emotional, like a documentary trailer. Go all out. Compose the soundtrack yourself, fully synthesized in code. No samples, no audio files, no VST instruments. Cut the visuals to the music. 1920x1080, 60fps. No downloaded images or footage, every frame generated in code. Render the final video with the music mixed in as an MP4 and save it [here] (your preferred file location).

Create a 10-second ultra-photorealistic live-action university vlog of a European brunette in her mid-20s waking to a rainy morning, making coffee, getting ready, grabbing her umbrella, walking through a wet city, and arriving at university just before class. Modern smartphone vlog style: natural handheld movement, subtle shake, autofocus changes, realistic rain, puddles, reflections, traffic, umbrellas, wet clothing, natural expressions and believable human movement. Dialogue: “Good morning… apparently, it’s raining.” → “First coffee. Then lecture.” → “Two minutes early. I’m impressed.” → “Okay, made it.” Natural diegetic audio only. No music, narration, CGI look, artificial rain, face changes, distorted anatomy, impossible physics, subtitles, logos or watermark. 16:9 • 24fps • 4K • Ultra-photorealistic live action

Montage, multi-shot candid observational footage, don't use one camera angle or single cut. Handheld documentary feel of something caught by accident, slightly imperfect framing, small drifts and reframing corrections, gentle exposure breathing, autofocus arriving half a beat late. Real skin texture, natural overcast daylight, film grain, shallow depth of field. Relaxed breathing, small blinks, restrained natural performance — 11 shots, 30 seconds. A woman <<<image_1>>> 1 in a loose linen shirt and rolled-up jeans carries a plastic laundry basket up to a quiet apartment rooftop to hang wet clothes. Wind keeps tugging at the sheets, she fumbles with clothespins, gets a call from her mom mid-task, and finally sits on a low ledge to enjoy the breeze before heading back down. Shot 1 (0-3s): She pushes open the heavy rooftop door with her hip, basket balanced against her waist, wind immediately catching loose strands of hair. She mutters in Korean, lips visible: "바람 되게 세네." Handheld camera trails behind her entrance, slightly unsteady. Shot 2 (3-5s): She sets the basket down near the clothesline and pulls out a large wet sheet, shaking it out once. Camera circles slightly to catch the fabric snapping in the wind. Shot 3 (5-7s): She pins one corner, but the wind flips the sheet before she reaches the other side, momentarily covering her face. She laughs, muffled. Camera reacts with a small startled jolt, then corrects. Shot 4 (7-9s): She finally gets both corners pinned and steps back, hands on hips. She says quietly: "됐다." Camera pushes in slightly on her satisfied expression. Shot 5 (9-11s): She bends to pull more clothes from the basket — a few t-shirts and socks — and starts pinning them in a row. Camera stays low, following her hands moving item to item. Shot 6 (11-13s): Her phone buzzes in her back pocket. She wipes one hand on her jeans and answers. Camera swings up from her hands to her face as she picks up. Shot 7 (13-16s): She talks to her mom, half-laughing. She says in Korean, lips visible: "엄마, 나 지금 빨래 너는 중이야." Wind noise makes her raise her voice slightly, hand cupped near her ear. Camera holds a loose over-the-shoulder angle, wind visibly moving her hair. Shot 8 (16-18s): She tucks the phone against her shoulder and keeps pinning one-handed, slightly clumsy. Camera catches the awkward multitasking from a low three-quarter angle. Shot 9 (18-21s): She laughs again at something her mom says, nearly dropping a clothespin, catching it against her chest. She says: "아 진짜, 알겠어." Camera reacts with a quick handheld dip toward the falling pin and back up. Shot 10 (21-25s): She ends the call, exhales, and walks to the ledge, leaning her forearms on it to look out over the city skyline, wind still moving the hanging laundry behind her. Camera drifts wide to include both her and the skyline, slow push in. Shot 11 (25-30s): She closes her eyes briefly, enjoying the breeze, then glances back at the empty

Main Subject: Young Korean woman, 24, naturally attractive, realistic skin, minimal makeup, long dark hair loosely down. Wearing simple casual home clothes while getting ready. Preserve her exact identity, facial features, hairstyle, body proportions and appearance throughout. Location: Small old Seoul apartment bedroom during a dark, rainy early morning. Open wardrobe, simple wooden dresser, small mirror, folded clothes, canvas shoulder bag and a rain-covered window overlooking blurred neighboring apartment buildings. Steady rain outside. Lighting & Atmosphere: Cozy, dim blue-hour ambience. Cold blue-gray light from the rainy window mixed with a weak warm bedside lamp. Dark cloudy sky, muted colors, soft shadows and slightly underexposed interior. Quiet, sleepy and intimate rainy-morning atmosphere. Style: Ultra-realistic early-2000s Sony MiniDV home video filmed by another person holding the camcorder. Completely candid and unstaged, like a family member casually filming her while she gets ready. Natural handheld movement, subtle human shake, imperfect framing, gentle reframing, occasional autofocus hunting, mild exposure shifts, faded colors, soft contrast, authentic DV compression, subtle low-light digital noise and microphone noise. Smooth continuous real-time motion throughout. No stuttering, judder, frame skipping, duplicated frames, stop-motion appearance, excessive motion blur, speed changes or low-frame-rate look. No stabilization or modern cinematic movement. 00:00–00:04: She stands sleepily in front of her open wardrobe, looking through several hanging clothes. She pulls out a simple oversized shirt and holds it against herself while looking toward the mirror. 00:04–00:08: She changes her mind, puts it back and pulls out another shirt. She holds the second one against herself and studies her reflection with a thoughtful expression. 00:08–00:12: She looks between the two shirts, holding one in each hand. She glances toward the camcorder with a tiny uncertain smile, as if silently asking which one she should wear. 00:12–00:16: She places both shirts on the bed and sits down for a moment. She looks toward the rainy window, listening to the rain, then looks back at the clothes. 00:16–00:20: She picks up one shirt again and stands in front of the mirror, briefly holding it against herself from different angles. She turns slightly left and right, checking how it looks. 00:20–00:24: She finally seems satisfied. She hangs the other shirt back in the wardrobe and neatly folds the chosen shirt over her arm. 00:24–00:27: She notices a lightweight jacket hanging nearby, pauses and considers it too. She touches the jacket, then looks outside at the rain through the window. 00:27–00:30: She decides to take the jacket as well. She looks toward the camcorder with a small satisfied smile, picks up her canvas shoulder bag and walks out of the bedroom. The camera casually follows her for a couple of steps before cutting.

Art Style: Photorealistic wildlife photography, shallow depth of field, natural cold lighting, subtle film grain, honest and unposed, no glamorization. Content: A Siberian tiger walking

A continuous 12-second, photorealistic handheld smartphone video shot in raw 4K vertical format along a paved countryside road. Seconds 0–4: Wide tracking shot following a large white domestic goose sprinting along the asphalt. An orange tabby cat is mounted firmly on the goose's back, front paws wrapped tight around the goose's neck, whiskers flattened by the wind. Just behind them, an excited golden retriever bounds into frame at full sprint, barking furiously in loud, breathless bursts as it tries to catch up. Seconds 5–8: The goose suddenly swerves off the tarmac onto a soft grass verge and brakes hard. The ginger cat instantly dismounts, hits the turf, and drops into an absurd, ultra-fast, cartoonish four-legged turbo-gallop—legs moving in a manic, high-speed blur. Instead of fleeing into the bushes, the cat loops in a blazing circle and dashes straight past the confused dog's legs. Seconds 9–12 (The Twist): The golden retriever skid-stops, head swiveling in complete bewilderment as the hyper-speed cat springs straight back onto the waiting goose’s back like a pro stunt rider. The goose immediately accelerates away down the road, leaving the dog sitting on its haunches, panting dumbfoundedly as the duo escapes. Visual & Style Specs: Raw viral phone footage, realistic physics for the animals with high-frame-rate comical leg-blur for the cat's hyper-sprint, sunny natural lighting, true fur/feather textures, dynamic handheld camera sway. Audio Track: Rapid rhythmic footfalls on tarmac, energetic dog barks and panting, abrupt goose honks, set against an off-camera woman wheezing into uncontrollable, hysterical snorting laughter that completely loses it when the cat engages ultra-fast sprint mode.