APIMODELSAPIMODELS

28+ Kling 4.0 手机实拍与社媒提示词

28+ 条。收录 270 多条可灵 Kling 4.0 视频提示词,大部分来自 X 上的创作者,其余来自 kling4.video 的提示词库、fal.ai 的 Kling 4 提示词指南和几条 Civitai 帖子。每条都逐字全文收录、署名作者、链回出处,出处带视频的,卡片里直接播放。这些提示词都不是我们写的。

← 回到全部 275 条 Kling 4.0 提示词

最近更新: · Kling 4.0 API 价格与在线体验 · API 文档

一句话摘要

可灵 Kling 4.0 是快手可灵系列最新的视频模型,每条视频都会连声音一起生成。在 apimodels.app 上,一个模型名 kling-4.0 就有三种用法:只写提示词的文生视频;带参考素材的参考生视频,最多 10 张图、5 段视频、7 个具名主体和 3 段音色,在提示词里用 @ 点名;还有关键帧生视频,一张首帧、最多 8 张中间帧和一张尾帧。时长 3 到 30 秒,720p 或 1080p,提示词最长 10000 字符;720p $0.157/秒、1080p $0.235/秒,失败不扣费。轻量档 Kling 4.0 Flash(kling-4.0-flash)用同一套请求,但只有 720p、3 到 20 秒,提示词最长 8000 字符,关键帧只收首帧,$0.094/秒。API 上没有负面提示词、没有多镜头分镜模式,也没有 4K。本页收录 270 多条别人公开发布的 Kling 4.0 提示词,大部分来自 X。

出处:Kling AI(模型方官方站点)

悬停预览 · 点击有声

男友夏日街头随手拍女友

Create a 30-second video in 1920×1080, with the realism of an ordinary recording a boyfriend made of his girlfriend during a summer walk in Japan. The small story: They have no particular destination. She has just opened a can of fruit juice, and he starts recording without announcing it. Over the next half-minute, she gradually notices the camera, gets a little embarrassed, and begins including him in the moment. Their familiarity should come through in how she waits for him, looks at him, and closes the distance between them. The woman: A beautiful Japanese woman around 24 years old, with an approachable, individual face rather than a polished model appearance. Dark brown eyes, natural skin texture, light everyday makeup, and straight black hair loosely gathered in a low ponytail. A few short strands around her temples have escaped in the summer humidity. She wears a faded navy cotton T-shirt, loose ivory trousers, comfortable gray sneakers, and a small rust-colored crossbody bag. A plain elastic hair tie is around one wrist. Keep her facial features, hair, outfit, bag, and proportions consistent in every shot. Her expressions are understated and responsive. Sometimes she looks distracted or squints in the light. Her smile develops because of something between them, rather than appearing on cue. The place: A residential neighborhood in a regional Japanese city, near a narrow canal. Late afternoon in midsummer, with warm sunlight still reaching the upper parts of the houses. Use ordinary local details: patched asphalt, short concrete bridges over the canal, metal railings warm from the sun, compact houses with tiled roofs, frosted entrance windows, outdoor air-conditioning units, bicycles tucked beside gates, and plants growing in mismatched pots. Utility wires cross the sky. A striped fabric awning hangs outside a closed shop farther down the street. Avoid famous landmarks and postcard compositions. The location should feel like somewhere the couple already knows. How the camera behaves: The boyfriend holds a small consumer camera at his own eye or chest level. Most of the time he is beside her or only a step behind. The viewpoint should feel like her companion's attention, with no third-person shots of the couple. Let framing respond to their interaction. When she suddenly looks at him, he may take a fraction of a second to correct the composition. When he laughs, the camera moves slightly. When they walk, his footsteps create a gentle, uneven rhythm. Use ordinary consumer-camera rendering: restrained detail, modest dynamic range, occasional clipped highlights, a little shadow noise, and slight focus hesitation when she moves closer. Preserve natural skin color. Keep the camera's attention on her face, gestures, and shared surroundings. Avoid fashion poses, body-scanning shots, elaborate camera moves, slow motion, or a glossy advertising finish. Suggested edit — six short memories from the same walk: 0:00–0:05 She stands beside a low canal railing with a small, unbranded can of mandarin orange juice. The recording begins just as she finishes lifting the pull tab. She takes a sip, lowers the can, and notices that he is filming. She looks directly at him with a briefly puzzled expression, then gives a small, amused smile, as though silently asking, “Really, you're filming this?” Begin with a slightly loose frame that includes her hands and the canal, then let the boyfriend adjust it toward her face. 0:05–0:10 A cut to the two of them walking side by side along the narrow lane. She carries the can naturally in one hand. She gets half a step ahead, realizes he has slowed down while filming, and turns back. Instead of posing, she waits with a mildly impatient but affectionate look. Once he catches up, she resumes walking beside him. Her gaze should meet the person holding the camera, not an imaginary audience. 0:10–0:15 They pause near a house entrance where a small glass wind chime hangs beneath the eaves. A breeze makes it ring. She glances upward and listens for a moment. Loose strands of hair move against her cheek; she brushes them aside with her free hand. She notices him still watching her through the camera and suppresses a smile before looking back toward the wind chime. Keep this moment quiet and unhurried. 0:15–0:20 On a short concrete bridge, she stops to look down at the canal. The camera first catches the rippling water, then tilts back toward her after she has already turned around. She leans slightly toward him to see what he is recording. Her face briefly comes too close for the camera's focus, making her laugh softly and step back. The laugh should feel like a shared interruption, with a small answering movement from the camera. 0:20–0:25 They rest beside the bridge in the shade. She is closer to the camera now, relaxed enough to stop paying attention to being filmed. She takes another small sip of juice, then holds the can toward him as an offer. Leave the can clearly visible in the foreground, separate from her face. He does not take it during the shot. She gives a tiny questioning lift of her eyebrows, then draws it back with an affectionate smile. 0:25–0:30 She steps toward him and gently touches his free forearm, inviting him to keep walking. Only a small, natural part of his arm may enter the lower edge of the frame. As they start moving together, he lowers the camera. The view drifts from her smiling face to her bag, their moving shadows, and the sunlit pavement. Let the last second settle on the overlapping shadows of two people walking close together. End with an ordinary recording stop, without a dramatic transition or title. Sound: Use the sound that would actually reach the camera microphone: cicadas, the juice can opening, a few footsteps, a wind chime, faint canal water, distant tires on the road, and their quiet laughter. A single softly spoken Japanese remark such as “撮ってたの?” may occur when she first notices the camera, but speech is optional. Do not add explanatory dialogue. No music, voiceover, or added dramatic sound effects. Continuity and physical detail: There is one juice can, carried by her throughout the sequence. Its color, size, opening, and remaining shape stay consistent. The crossbody strap stays on the same shoulder. Her ponytail remains tied. Hands should hold objects with believable contact and weight. Walking speed, foot placement, shadows, and reflections must agree with the surroundings. Keep the canal, bridge, houses, sunlight direction, and summer weather coherent between cuts. Emotional direction: The viewer should feel that she knows and trusts the person behind the camera. Build that feeling through small changes in distance and attention: noticing him, waiting for him, leaning in, offering a drink, and asking him to walk beside her. Let ordinary pauses remain. Her beauty is something her boyfriend happens to notice during their time together. The finished video should feel like a personal memory he would keep because of the way she looked at him that afternoon.

来自 @agi_aibusi看原帖
悬停预览 · 点击有声

咖啡店手机前置镜头独白

Raw vertical 9:16 video recorded on a phone's front camera (not on screen), one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft image, a little digital noise and compression, deep focus with the café behind her as sharp as her face, auto-exposure brightening and dimming as people pass the sunny window behind her, no colour grading. The phone sits close and a little low, looking slightly up at her; around 9 s a waiter's arm bumps the table, the phone slides and tilts, and she catches it and props it back up. Sadie, a young woman in her early 20s with long honey-blonde waves, light natural makeup, small gold hoop earrings and a cropped cream knit top, sits at a small marble table by a big sunny window in a busy brunch café. Her sister sits across from her, out of frame except for her arm and a hand with a silver ring. Behind Sadie: other tables of people talking, a counter with a glass pastry case, plants, staff moving about, chalkboards too far away to read. Around 9 s a waiter sets down two plates, a lemon ricotta pancakes and a plate of eggs on toast, plus two iced coffees in plain glasses. Around 15 s she turns the pancake plate toward the camera; around 20 s she cuts a bite, eats it and closes her eyes; around 26 s her sister's hand reaches in with a fork and steals a piece of pancake, and Sadie swats at it, laughing. [Sadie: casual young woman's voice, warm and a bit giggly, recorded on the phone's built-in mic from half a metre away]: "Okay, I promised I'd show you what we get. ... We get the same thing every single time. ... Oh, here it is! Thank you! ... Look at this. Lemon ricotta pancakes. ... Mmm. Okay. Yeah. Still the best. ... Hey! Get your own!" [Her sister, off camera: a young woman's voice, dry and amused]: "That's the tax." [Sadie, laughing]: "She does this every time." Audio as captured by the phone mic in a busy café, not studio clean: her voice close but with the room around it, the hiss and grind of an espresso machine, cups and cutlery clinking, overlapping chatter and laughter from other tables, a door bell jingling, a chair scraping, faint music in the background. No subtitles, no captions, no on-screen text.

来自 @minchoi看原帖
悬停预览 · 点击有声

猫咪披萨外卖手机短片

Create an exactly 20 second ultra-realistic live-action comedy video, filmed as if recorded on an iPhone 18 Pro Max. Vertical 16:9, 1080×1920, 24fps. Everything must look like authentic real-world smartphone footage. The cat is a completely realistic domestic cat with natural fur, realistic anatomy, whiskers, eyes, paws, and believable movement. The cat wears only a realistic pizza-company delivery jacket. No pants, shoes, hat, or other clothing. SCENE 1 — 0.0–3.5s — PACKING THE PIZZA Inside a realistic neighborhood pizza shop. The cat stands at the preparation counter. The cat physically takes the freshly prepared pizza and places it into a pizza box. The cat then manually closes the pizza box. The cat picks up the closed pizza box and physically places it inside the pizza delivery bag. The cat then manually closes the delivery bag. Every action must be clearly visible and happen in the correct order. Pizza → pizza box → close pizza box → place box inside delivery bag → close delivery bag. Nothing appears or disappears automatically. SCENE 2 — 3.5–6.5s — SCOOTER AND DRIVING The cat carries the closed delivery bag to its scooter. The cat physically places the delivery bag onto the scooter, secures it, climbs onto the scooter, starts the engine, and rides away. Cut to a front-facing camera angle showing the cat and scooter driving toward the camera. The cat drives naturally through a realistic neighborhood road. The cat's face is normal and clean during the driving sequence. Natural iPhone handheld movement, autofocus, exposure changes, motion blur, realistic road movement, scooter engine and street ambience. SCENE 3 — 6.5–10.0s — ARRIVING AND TAKING THE PIZZA The cat arrives outside the customer's house. It stops the scooter and physically puts the scooter on its stand. The cat gets off and walks to the delivery bag. The cat manually opens the delivery bag, reaches inside, grips the closed pizza box, and physically pulls the pizza box completely out of the bag. The cat then closes the delivery bag. The cat carries the pizza box toward the customer's front door. As the cat approaches the house, clearly show a small amount of pizza sauce and a little melted cheese around its mouth, making it obvious that the cat has secretly eaten some pizza. The pizza residue must not appear suddenly. It must remain naturally visible. SCENE 4 — 10.0–12.0s — DOORBELL FIRST The cat reaches the front door while holding the pizza box. The door is completely closed. The cat raises its paw and physically presses the doorbell. Only after the doorbell is pressed, the customer opens the door. Strict order: Cat reaches closed door → cat presses doorbell → door opens. The door must NOT open before the bell is pressed. SCENE 5 — 12.0–14.5s — HANDING OVER THE PIZZA After the door opens, use a camera angle facing the cat from the customer's side. Do not show the customer's face or body. Only the cat should be visible. The cat holds the pizza box forward and says: Cat: "Here is your pizza." The cat physically hands the pizza box to the customer. The pizza box must visibly transfer from the cat's paws into the customer's hands. The pizza residue around the cat's mouth remains clearly visible. SCENE 6 — 14.5–17.0s — THE MISSING PIZZA Cut to the customer's POV. The customer is physically holding the same pizza box. The customer manually opens the pizza box. The cat is standing behind the open box and becomes naturally hidden by the box. Inside the box: - One pizza slice is missing. - Another remaining slice has a small realistic bite taken from it. - All other pizza remains physically present. - Nothing disappears during the shot. The customer looks shocked and says: Customer: "Where is my missing slice?" The customer manually closes the pizza box and looks toward the road. SCENE 7 — 17.0–20.0s — ESCAPE AND FLYING KISS The cat is already moving toward its scooter. The cat climbs onto the scooter, starts it, and physically rides away down the road. The customer runs out toward the road and shouts: Customer: "Stop! Stop!" The cat keeps riding farther away. While actively riding away, the cat turns its head back toward the customer and raises one paw, clearly making a flying-kiss gesture directly toward the customer. The flying kiss must happen mid-ride while the cat is moving away, not before riding away and not after stopping. The cat then faces forward and continues riding farther down the road. End naturally with the cat getting smaller in the distance while the customer remains on the road. STRICT PHYSICAL CONTINUITY Every action must happen physically and in logical order. Nothing may appear automatically. Nothing may disappear automatically. No teleportation, morphing, object replacement, duplicated objects, or sudden prop changes. The exact pizza box used in the shop must remain the same pizza box throughout the entire video. The exact delivery bag must remain the same bag throughout. The pizza must be physically packed into the box, the box physically placed into the bag, the bag physically placed on the scooter, the box physically removed from the bag, and the box physically handed to the customer. The door must remain closed until the cat physically presses the doorbell. The customer must physically open and close the pizza box. The missing pizza slice must simply not be present when the box is opened; it must not disappear during the shot. REALISM REQUIREMENTS - Exactly 20.0 seconds - 9:16 vertical - 1080×1920 - 24fps - iPhone 18 Pro Max realistic footage - Realistic domestic cat - Realistic fur and anatomy - Only pizza-company delivery jacket - Realistic scooter physics - Natural handheld camera movement - Natural autofocus and exposure - Realistic shadows and reflections - Realistic road and house - Natural environmental audio - No CGI appearance - No cartoon style - No animation - No exaggerated expressions - No extra limbs - No distorted paws - No morphing - No teleportation - No automatic object movement - No subtitles - No text overlays - No watermark

悬停预览 · 点击有声

千禧DV风首尔周日街头

30-second, 4K, 16:9 ultra-realistic home video shot on an early-2000s consumer DV camcorder. A quiet Sunday afternoon in an older Seoul residential neighborhood, filmed by a friend walking with her. One continuous handheld recording with a couple of casual cuts, like someone pausing and restarting the tape. SUBJECT A young Korean woman in her early 20s, the same person in every moment. Natural beauty, realistic skin texture, minimal makeup. Black wavy hair in a messy side ponytail over her left shoulder. Fitted pastel-blue short top, loose cream pajama-style pants, black sneakers, a thin simple silver necklace. A small beige reusable canvas shopping bag hangs on her left shoulder the whole time. Her face, hair, outfit and bag stay exactly the same throughout. LITTLE GIRL A Korean girl about 6 years old, short bob haircut, yellow T-shirt, denim shorts, small pink sneakers. THE BALLOON One round red helium balloon on a thin white string. Only one balloon exists in the video. LOCATION A quiet older Seoul neighborhood on a mild sunny afternoon. Narrow residential lane, low brick houses, red-brick villas, potted plants by doorways, parked bicycles, utility poles with tangled overhead wires, mature trees. A small neighborhood playground with a slide, a sandbox and a few low trees along a short fence. CAMERA & LOOK Raw handheld early-2000s DV camcorder footage. Shaky hand movement, autofocus hunting, faded slightly cyan-tinted colors, exposure shifts between sun and shade, motion blur, soft standard-definition-like detail upscaled, occasional accidental zoom in and out, imperfect off-center framing. The camera operator sometimes reacts a little late. AUDIO Natural neighborhood ambience recorded by the camcorder mic: birds, children playing in the distance, a bicycle bell, footsteps on asphalt, wind hitting the mic, faint distant traffic. No music. 0–5s — LANE She walks slowly through the quiet residential lane, shopping bag on her left shoulder, the camera following slightly behind and to her right. Sunlight flickers through the trees. 5–10s — PLAYGROUND She passes the small playground. The camera accidentally zooms in and finds the little girl on tiptoe under a low tree branch, reaching up but unable to touch the red balloon caught in the leaves, its string dangling just out of reach. The woman slows down, notices her, and walks over. 10–17s — FREEING THE BALLOON Medium shot from the side. Standing beside the girl, the woman reaches up with her right hand, gently untangles the string from the branch and pulls the red balloon down. The string is clearly held in her right hand. She bends down and places the string into the girl's small hand; the girl grips it tightly and the balloon floats above her head. 17–21s — THANK YOU Same continuous shot, both of them in frame, the camera slightly off to the side. The girl looks up holding the balloon and says brightly in English: "Thank you!" The woman smiles and replies softly: "You're welcome." The girl runs back toward the slide, still holding the balloon string, the red balloon bobbing above her. 21–26s — WALKING HOME Cut. The woman walks on down the lane, smiling to herself, bag on her left shoulder. A light breeze moves loose strands of her hair, and a few dry leaves skitter across the street. The camera follows a few steps behind, wobbling. 26–30s — HOME She reaches the gate of a small red-brick house, opens it, then turns back toward the camera and says with a relaxed smile: "Pretty nice Sunday." She steps inside and the gate swings shut. Hard cut to black.

来自 @AI__TSUBAKI看原帖
悬停预览 · 点击有声

金发女子浴室时尚摆拍

Create a 13-second vertical 9:16 ultra-realistic cinematic fashion/lifestyle video featuring a beautiful young blonde woman with long, voluminous wavy hair, wearing a fitted beige short-sleeve top and high-waisted light-blue denim shorts, standing in a luxurious modern bathroom with dark marble walls and warm ambient lighting. She looks confidently into the camera and performs a smooth, playful sequence of fashion-model poses and subtle dance movements. Start with a medium full-body shot as she raises one hand and poses naturally. She then turns slightly to the side, showing her outfit from a flattering side angle, before facing the camera again and moving her arms rhythmically. Continue with elegant hair-touching and confident poses, maintaining natural body movement and realistic facial expressions. Near the end, smoothly push the camera closer as she leans slightly toward the camera, gives a playful kiss gesture, and finishes with a charming smile. Use realistic skin texture, detailed blonde hair strands, natural body proportions, accurate hand anatomy, smooth motion, subtle handheld camera movement, shallow depth of field, warm bathroom lighting, realistic reflections, cinematic color grading, high detail, photorealistic quality, consistent character identity throughout, no flickering, no distorted hands, no sudden scene changes. Style: luxury fashion reel, photorealistic, cinematic, elegant, playful, visually polished. Duration: 13 seconds Aspect ratio: 9:16

悬停预览 · 点击有声

手持跟拍猫咪夜间狂奔

An amateur handheld recording in third-person perspective rapidly follows a cat through its nighttime adventures in one continuous shot, with no cuts, no stopping, and no pause. The sequence feels like fast-forward, showing all the curious, mischievous, funny, and chaotic things a cat might do while the town, house, garden, and everything around it are half-asleep.The cat races through about ten places a cat would naturally explore at night, each one flowing into the next in a single continuous chase: a quiet bedroom, a dim hallway, a kitchen, a living room, a backyard garden, a fence line, a garage, a rooftop, a tree, and a sleepy street or alley. The camera is shaky and imperfect, always struggling to keep up as it follows close behind. In every location, the cat causes some kind of chaos. In the bedroom it jumps onto furniture and disturbs sleeping people. In the hallway it sends small objects tumbling. In the kitchen it knocks items off the counter. In the living room it darts across shelves and tips things over. In the garden it rustles plants, digs, or startles birds. On the fence it rattles loose objects. In the garage it weaves through clutter and causes things to wobble or fall. On the rooftop it skids across surfaces and startles other animals. In the tree it shakes branches and scatters leaves. In the street or alley it darts through trash, shadows, and unexpected little messes, always leaving brief chaos behind before sprinting onward.The cat behaves like a restless, curious nighttime troublemaker: climbing, leaping, squeezing through gaps, chasing tiny movements, staring at strange things, swatting objects, knocking things over, and vanishing into corners before bursting back out again. Every scene should contain a clear action, reaction, or small disaster caused by the cat.Everything around it feels sleepy and still, with dark houses, quiet gardens, dim windows, silent rooftops, and empty streets. The contrast is that the cat is pure energy, moving so quickly that the world smears into motion blur and fragmented glimpses of chaos. The overall feeling is funny, lively, mischievous, and cinematic, like a whole secret night of cat behavior compressed into one rapid continuous chase.

来自 @umesh_ai看原帖
悬停预览 · 点击有声

相扑选手花样滑冰

Realistic vertical handheld phone footage at a bright indoor ice rink during a figure skating competition, no text or captions on screen. Two enormous sumo wrestlers in mawashi loincloths and top-knots perform a graceful pairs figure skating routine: gliding hand in hand, one lifts the other high overhead with one arm and spins, then throws him into a triple jump that lands with a thud that cracks the ice, both finishing with elegant arms out while judges in the background hold up score cards. Single continuous shot, no cuts, romantic orchestral music, scraping blades and a heavy thud.

悬停预览 · 点击有声

厨房咖啡洒落小插曲

Create a realistic 17-second vertical 9:16 cinematic lifestyle video set in a warm, modern home kitchen. A young woman with long curly brown hair, wearing a cozy beige sweater and dark pants, stands at a kitchen counter preparing coffee. A young man in a dark navy sweater works casually in the background. The woman places a red ceramic coffee mug on the countertop and prepares to make coffee. She uses a spoon to scoop instant coffee powder from a jar and accidentally spills a small pile of coffee powder onto the countertop. She looks down at the mess with a surprised but amused expression. She then continues preparing the coffee while the man notices what happened and reacts with a confused, slightly amused expression. Near the end, milk suddenly spills dramatically across the countertop, spreading outward in an exaggerated but physically realistic splash. Both characters react naturally with surprise. Use photorealistic human characters, natural facial expressions, realistic hand movements, accurate coffee and milk physics, detailed kitchen environment, warm indoor lighting, shallow depth of field, subtle handheld camera movement, realistic textures, natural body motion, and cinematic composition. Keep the coffee jar, mug, characters, clothing, kitchen layout, and object positions visually consistent throughout the entire video. No text, subtitles, logos, or unnatural morphing. Style: ultra-realistic cinematic lifestyle commercial, authentic home-video feel, high detail, natural lighting, smooth motion, 4K quality. Aspect ratio: 9:16 Duration: 17 seconds

悬停预览 · 点击有声

偶像电玩城休息日

A Day Off in the Arcade"The entire video is a single continuous self-filmed vlog captured by the idol using a compact consumer camcorder. It should feel like genuine footage she'd upload privately rather than professionally produced content. The camera is always in her hand and never visible. Natural arm movement, occasional imperfect framing, realistic autofocus hunting, minor exposure adjustments, subtle walking bounce, and authentic human timing. No jump cuts or scene transitions.Visual StyleAuthentic early-2000s MiniDV look with soft detail, light digital noise, gentle highlight bloom, muted contrast, accurate skin tones, and slight tape compression. No cinematic color grading, no artificial blur, no beauty filters.CharacterJIWON, a Korean female idol in her twenties. Long black hair in a loose ponytail, natural makeup, warm smile, expressive eyes, casual oversized denim jacket over a simple T-shirt, relaxed jeans, sneakers, and a baseball cap.LocationA lively retro arcade during the evening. Bright neon signs, claw machines, racing games, basketball arcade, photo booth, capsule toy machines, colorful carpets, cheerful background music, and other visitors naturally enjoying themselves.The recording begins just outside the arcade entrance.JIWON pushes open the glass door and laughs."I promised myself I wouldn't work today... so this is today's mission."She slowly walks through the arcade, turning the camera to show rows of flashing machines and excited players.She stops in front of a claw machine filled with plush toys."I've never actually won one of these."Holding the camera in one hand, she awkwardly controls the joystick with the other.The claw almost grabs a plush before dropping it.She bursts into genuine laughter."That was definitely rigged."She wanders toward a rhythm game and watches someone playing.She quietly cheers for them before giving a thumbs-up when they http://finish.Next, she enters a small photo booth.The camera captures her fixing her hair while laughing at herself.A countdown begins.She quickly strikes a funny pose just as the photo is taken.Leaving the booth, she opens a capsule toy machine and twists the handle.The capsule drops.She excitedly opens it."Actually... this one's kind of cute."As she continues exploring, she notices the arcade staff announcing that the venue will close in a few minutes.She looks around dramatically."I think I lost track of time."Walking toward the exit, she turns the camera back to herself.The neon lights reflect softly behind her."Days like this help me recharge. Next time I'll definitely win that plush."She smiles, waves goodbye, pushes open the arcade door, and steps outside into the cool evening air. The city sounds replace the arcade music as the recording ends naturally, as though she simply stopped filming after leaving.Generation RequirementsOne continuous uninterrupted take. Stable identity throughout. Realistic hand-held camera movement, believable eye contact with the lens, accurate hand interactions, physically correct walking and body mechanics, consistent lighting, authentic ambient sound, no duplicated people, no warped anatomy, no AI artifacts, no scene changes, and no unnatural motion. The final result should be indistinguishable from a real idol's casual day-off vlog filmed on a personal camcordesr.

悬停预览 · 点击有声

高定礼服派对视频

20 seconds | 9:16 | authentic UGC vlog | premium creator content | MULTISHOT CORE CONCEPT A young female creator casually documents her experience attending an exclusive evening party while wearing a spectacular artistic couture gown. The video should feel like a real spontaneous social-media vlog filmed by a friend rather than a polished commercial. The creator speaks naturally to the person filming, shows her beautiful dress, walks through the party and shares genuine moments from the evening. The performance should feel unscripted, confident and relatable while maintaining premium visual quality. Use the FIRST uploaded reference image as the EXACT visual identity reference for the creator. Preserve her facial identity, facial proportions, hairstyle, hair color, body proportions and recognizable appearance throughout the entire video. Use the SECOND uploaded reference image as the EXACT dress design reference. Preserve the sculptural yellow bodice, multicolored textured fabric, blue-green and yellow transitions, dramatic navy floral train and exceptionally long flowing sleeve drapes. CREATOR Young adult female creator based directly on the uploaded reference image. Long, naturally straight auburn-brown hair. Elegant artistic couture gown based on the second reference image. Natural hair. Realistic skin texture. Elegant evening makeup. Natural facial expressions. No exaggerated influencer behavior. She should feel like a genuine lifestyle creator enjoying an exclusive evening party while a friend records her. SETTING Luxurious dressing area and exclusive evening party venue during the late evening. Natural warm lighting. Elegant full-length mirror and sophisticated interior details. Beautifully illuminated party environment with chandeliers, fashionable guests and refined architectural surroundings. The environment should feel real and slightly imperfect rather than like a professional advertising studio. STORY — VLOG MULTISHOT VISUALS 00:00–03.0 — HOOK A second person films the creator using a handheld camera. The creator stands naturally in her elegant couture gown. She looks toward the camera and says: “Okay, you have to see this dress.” Natural conversational delivery. Small smile. Natural blinking. Subtle handheld camera movement. No smartphone or filming device visible in the frame. 03.0–06.0 — DRESS REVEAL Camera angle changes naturally to a slightly wider handheld shot. The creator steps back and turns slightly to reveal her gown. The camera captures the colorful sculptural bodice and flowing silhouette. She says: “Look at the details on this thing.” The long sleeve drapes move naturally as she adjusts her arms. The dramatic navy train remains visible. 06.0–09.0 — DETAIL SHOT Close handheld camera framing. The creator gently raises one arm to show the long, flowing sleeve drapes. The camera moves closer to capture the intricate yellow, turquoise, green and blue fabric textures. She lightly touches the gown and smiles naturally. The fabric responds realistically to her movements. 09.0–13.0 — REAL-LIFE MOMENT Cut to a casual wider handheld angle. The creator walks through the dressing area while the camera operator follows her. The dramatic train trails naturally across the floor. She gestures toward the gown and says: “I wanted something completely different for tonight.” Her movements remain spontaneous and imperfect. 13.0–16.5 — PARTY CUTAWAY Handheld footage. The creator enters an elegant evening party venue. Guests mingle naturally in the background. The camera operator follows her, capturing the colorful gown, flowing sleeve drapes and navy floral train. She briefly turns toward the camera with a natural smile. No cinematic transition. 16.5–20.0 — FINAL MOMENT The camera operator frames the creator in a natural medium-wide shot. She stands inside the beautifully illuminated party venue. She looks toward the camera and says: “Be honest, would you wear this?” She gently turns to reveal the gown’s silhouette. End on a casual handheld frame rather than a polished commercial hero shot. CAMERA Authentic handheld camera operated by another person. The camera operator remains off-screen. Natural handheld movement. Minor framing imperfections. Occasional autofocus adjustment. Realistic exposure adaptation. Natural perspective and realistic lens characteristics. No selfie shots. No visible smartphones or filming devices. No cinema-camera movement. No gimbal perfection. No artificial depth-of-field. No commercial beauty retouching. VISUAL / COLOR SETTING Premium UGC vlog × contemporary luxury fashion content. Natural camera image quality with excellent exposure and realistic detail. Soft natural skin texture. Real fabric texture. Accurate colors and construction of the reference gown. The multicolored couture dress remains the visual centerpiece. The footage should feel like genuinely captured social content filmed by a friend during an extraordinary evening. COLOR GRADE Natural warm indoor lighting. Neutral skin tones. Realistic whites and blacks. Accurate yellow, turquoise, green and navy fabric colors. Subtle contrast. No cinematic teal-orange grade. No excessive HDR. LIGHTING Primarily warm practical lighting. Soft ambient illumination across the creator’s face. Realistic highlights on the gown’s textured fabric. Natural shadows beneath the train. Realistic exposure changes when moving between rooms. No studio beauty lighting. No artificial rim lights. AUDIO Authentic creator voice recorded naturally in the environment. Natural room tone. Clothing movement. Footsteps. Distant conversations. Natural party ambience and background music. Voice remains clear but not artificially processed. REALISM The video must feel genuinely user-generated and filmed by another person. Natural skin pores. Natural facial asymmetry. Real blinking. Natural pauses. Small speech imperfections. Authentic hand gestures. Realistic handheld camera movement. Physically accurate fabric weight and movement. No perfect influencer posing. No plastic skin. No beauty filter. No AI-generated facial expressions. No unnatural eye contact. No floating fabric. No melting embroidery. No smartphone or filming device visible anywhere. BRAND CONTROL The gown must remain accurate to the second uploaded reference image. Preserve the original sculptural bodice, multicolored fabric, long sleeve drapes and dramatic navy floral train. Do not redesign the dress. Do not invent additional branding. No watermarks. No random text. No fake UI overlays. EDITING Fast but natural vlog pacing. Cut whenever the creator changes activity. Use jump cuts sparingly. No cinematic transitions. No dramatic zooms. No glitch effects. No excessive B-roll. Maintain the feeling that a friend personally recorded the entire evening. CONTINUITY The creator remains identical throughout all shots. Same face. Same hairstyle. Same hair color. Same clothing. Same accessories. Same makeup. The same gown remains physically consistent throughout the video. Maintain the same party environment and time of day. The camera operator remains off-screen throughout. No smartphones or filming equipment appear in reflections or backgrounds. FINAL QUALITY TARGET The final result should look like a genuinely captured fashion vlog filmed by a friend accompanying a creator wearing an extraordinary couture gown to an exclusive evening party. It should NOT look like a traditional fashion commercial or a selfie video. The viewer should feel: “Her friend is filming her night out, and that dress is incredible.” Authenticity first. Dress visibility second. Commercial polish third.

悬停预览 · 点击有声

电梯隐藏摄像头恶作剧

Create a 15-second vertical 9:16 video. Real raw smartphone footage of a real prank inside an elevator of an ordinary residential apartment building. No text, no captions, no logos, no watermark, no music. It must look exactly like a real viral prank clip filmed on a phone by friends, not a movie, not CGI, not acted. CAMERA A smartphone is hidden high in the top front corner of the elevator cabin, just under the ceiling, right above the door on the LEFT side. It looks steeply down and diagonally into the cabin toward the back wall. The lens is a wide phone lens with mild natural barrel distortion at the edges, no fisheye. The camera is completely fixed for the whole video: no shake, no zoom, no pan. From this angle the viewer sees almost the whole floor, the mirror back wall with the handrail, the right side wall, and on the left edge of the frame the open doorway and a strip of the landing outside. The man is visible head to sneakers at every moment, seen from above and slightly from behind when he faces the back wall. IMAGE QUALITY Real smartphone video indoors: cool white light from a flat LED ceiling panel mixed with warmer light from the stairwell outside, slightly saturated colours like a typical viral clip, light digital noise in the darker corners, natural sharpness, the landing outside a bit overexposed. Natural motion blur only on the fastest moments (the jump, the release, the fall) and the image is sharp again right after. Skin, fabric and steel look real. No smearing, no ghosting, no melting shapes, no plastic skin, no beauty filter, no cinematic lighting, no slow motion. LOCATION A narrow passenger elevator in a modern residential apartment block, about 1.1 metres wide and 1.4 metres deep. The back wall is one large mirror in a brushed steel frame. A round horizontal brushed steel handrail runs across the full width of the back wall, at about hip height, mounted on two short brackets, about 5 cm away from the mirror. The left and right side walls are brushed stainless steel panels with fine vertical grain, a few scratches and fingerprints. On the right wall near the door: a steel button panel with round buttons and a small white paper notice taped above it. The ceiling is a flat white LED light panel. The steel sliding doors are fully open the whole video and never move. Outside the doors: an apartment building landing with light grey stone tiles, a white wall, a brown apartment door with a metal number plate, a radiator under a window with daylight. Nobody is on the landing. THE MIRROR The back mirror reflects the man and the cabin naturally, from the opposite side, matching his movements exactly. The camera itself is never visible in the mirror. The reflection never moves on its own. THE FLOOR ILLUSION (make it terrifying) The entire cabin floor, wall to wall and right up to the door threshold, is covered by one flat printed vinyl floor sticker. The print is a sharp, ultra realistic photo shot straight down an endless elevator shaft in an old high-rise: two rusty steel guide rails and thick greasy black cables running down into pitch black darkness, stained wet concrete walls with cracks, rows of closed shaft doors of lower floors getting smaller and smaller, metal crossbeams, a few dim orange work lights far below fading into darkness, dust in the air. The perspective goes down at least 30 floors, so the drop looks bottomless and deadly. At the very bottom, tiny and far away, a broken elevator car roof is barely visible in the dark. From the camera angle and from the doorway it looks completely real: as if there is no floor and stepping in means falling to death. The print is matte, does not glow, has no reflections. The sticker lies perfectly flat and its edges are tight against the walls and invisible until 00:11. Under it is a plain dark grey rubber elevator floor with small square anti-slip texture. CHARACTER One man, about 28 years old, clearly an adult. Stocky, strong, a little heavy, broad shoulders, thick arms. Short curly dark brown hair, a short dark beard, light olive skin, a small tattoo on his left forearm. He wears an oversized washed black cotton t-shirt, grey cotton jogger shorts above the knee, white socks and white sneakers, a small dark grey crossbody bag on a strap across his chest. Wireless earbuds in his ears. He looks like an ordinary guy coming home, not a model, not an actor. Real skin texture, real pores, natural face. His face, hair, beard, tattoo, clothes and bag stay exactly the same in every frame. ACTING STYLE He does not act. Every reaction is real, instant, instinctive and clumsy, like a real person who truly believes there is a deadly drop under his feet. Nothing is choreographed, graceful or symmetrical. Real body weight: he jumps heavily, hangs with straining arms, falls hard. Loud real breathing. He never looks at the camera until the last two seconds. TIMELINE (the most important part, follow every detail) 00:00.0 - 00:00.3 The video starts with him already entering the elevator: he is mid-stride in the open doorway, his body already leaning into the cabin, walking fast and relaxed, looking at the mirror ahead, bag bouncing on his chest. His LEFT foot stands on the steel door threshold strip. His RIGHT foot is already swinging forward inside the cabin, in the air about 20 cm above the floor sticker, about to step down, but it never touches the floor. 00:00.3 - 00:00.5 Right before his RIGHT foot lands, his eyes drop and he sees the bottomless shaft directly under it. His whole body jolts in shock, both arms fly up, eyes huge, and he lets out a loud piercing scream: "AAAH!" His RIGHT foot stays in the air and jerks up toward his chest instead of stepping down. 00:00.5 - 00:00.9 Still screaming, with the RIGHT knee already pulled up, he pushes off hard from the threshold with his LEFT foot and throws himself forward into the cabin toward the back wall, both arms reaching for the handrail, both feet in the air, knees pulling up, flying over the "shaft". His sneakers never touch the elevator floor at any moment. 00:00.9 - 00:01.3 Mid-air he slams into the back wall: both hands grab the handrail from above, fingers wrapped tight around it, his chest almost touching the mirror. His knees yank up to his chest and both sneakers slam against the bottom of the mirror and the lower steel frame below the rail. 00:01.3 - 00:02.0 He settles into a tight terrified crouch hanging on the rail: facing the mirror, back to the camera, elbows bent and raised high, both hands gripping the rail, head pulled into his shoulders, knees tucked against his chest, sneakers pressed flat against the bottom of the mirror and the left and right corners. His butt sticks out over the "shaft". His whole weight is held by his arms on the rail and his feet on the wall. In the mirror we see his face: pale, eyes wide, mouth open. 00:02.0 - 00:04.0 He hangs there like a frozen scared spider, shaking. His arms tremble, his legs tremble. He keeps looking down over his shoulder at the shaft, then back at the mirror, breathing in short loud gasps. He whimpers in a high cracking voice: "No no no no no!" His sneakers squeak as they slip a little on the mirror and he pushes them back. 00:04.0 - 00:04.8 A quick cut: the camera shows a straight-down view of the sticker filling the whole frame, the endless dark shaft looking totally real, as if we are falling with him. Then cut back to the same corner camera. 00:04.8 - 00:06.5 Back in the corner view, he is still hanging, now in full panic: he shifts his grip one hand at a time, his body twisting left and right, his t-shirt riding up, the bag swinging under him. He looks toward the open door to his left, realizing it is the only way out. 00:06.5 - 00:08.0 He tries to climb toward the door: he slides his hands along the rail to the LEFT end, then pushes his LEFT sneaker against the left side wall next to the door frame and his RIGHT sneaker against the back corner, legs spreading wide into a shaky split over the shaft, crotch low, knees bent, thighs shaking. He is almost there, his left foot touching the door frame. He grunts loudly with effort. 00:08.0 - 00:08.5 He lets go of the rail with his RIGHT hand and swings it back and to the left, reaching for the door frame behind him. His body twists open, his weight shifts backward. His left hand is now the only grip. 00:08.5 - 00:08.9 His left hand slips off the rail. Both feet slip off the walls. He tips backward into the air with arms flailing and legs kicking up. 00:08.9 - 00:09.4 He falls backward and lands hard on his back on the floor with a heavy thud, head toward the door and the camera, legs flying straight up in the air, sneakers pointing at the ceiling. He screams loudly the whole fall: "AAAAAH!" His mouth is wide open, eyes squeezed shut. 00:09.4 - 00:10.2 His legs bounce and drop. He lies on his back, arms spread, still screaming, then the scream breaks into gasping. His body is curled, knees up, hands clawing at the floor, waiting to fall into the shaft. 00:10.2 - 00:11.0 Nothing happens. He freezes. He opens his eyes, rolls a little onto his side, and slaps the floor with his right palm twice, feeling it. It is solid. His face turns from terror to confusion. 00:11.0 - 00:11.6 He spots the edge of the sticker where it meets the right wall. His fingers scratch at it, catch the corner and lift it. The vinyl peels up with a sharp ripping sound, showing the plain grey rubber floor underneath. 00:11.6 - 00:12.6 He scrambles onto his knees and rips the sticker up with both hands in one big angry pull, tearing it off the floor, the whole shaft image bending and crumpling in his hands, the normal grey elevator floor appearing under him. 00:12.6 - 00:13.2 He jumps to his feet and throws the crumpled sticker out of the open doors onto the landing tiles. 00:13.2 - 00:15.0 He spins around, looks straight up into the camera in the corner, and screams at it in pure rage, red faced, arms spread wide, stomping one foot: "WHO DID THIS?! ARE YOU SERIOUS?!" Loud laughter from his friends echoes from the stairwell off screen. The video ends on him still yelling at the camera. BODY AND PHYSICS RULES He always has exactly two arms, two legs and two hands with five fingers each. Joints bend only in natural directions. At every moment his weight is supported by something real: his feet on the floor, his hands on the rail together with his feet on the walls, or his body on the floor. He never floats, never hovers, never sticks to a wall without holding the rail, never turns upside down, never walks on walls or the ceiling. His hands never merge with the rail, the bag or the sticker. The rail never bends or moves. The floor is always solid and flat, he never sinks or falls through it. Clothes and bag move with gravity and inertia. Before the fall at 00:08.9 his feet never touch the cabin floor: he jumps directly from the door threshold onto the handrail. OBJECTS One handrail on the back wall. One mirror. One button panel. One paper notice. One crossbody bag, always on its strap across his chest. One floor sticker: flat until 00:11, then peeled and ripped up in one piece and thrown out at 00:13. No other objects appear, nothing duplicates, vanishes or changes shape. No readable text anywhere. CONTINUITY Same man, same face, same clothes, same lighting, same corner camera angle for the whole video, except the single short straight-down shaft shot at 00:04.0 - 00:04.8. The doors stay open. Only one person is ever visible. The camera is never seen in the mirror. The background never warps. SOUND Real phone audio, slightly echoey in the small steel cabin: fast footsteps, the elevator ventilation hum, his sharp gasp, his loud scream as he jumps from the doorway, sneaker squeaks on the mirror, his whimpering "no no no", grunts and heavy breathing, a long loud scream during the fall, a heavy thud on landing, the loud ripping of vinyl, his furious yelling at the camera, and friends bursting out laughing in the stairwell. FINAL FEEL A raw, real, hilarious prank clip that looks completely genuine: a real guy, real panic, real fall, real anger. Not acted, not polished, not CGI.

悬停预览 · 点击有声

巨指戳相扑小人

Use the input image as the first frame. A tiny photorealistic sumo wrestler figurine with an oversized head, oiled topknot, navy mawashi with hanging sagari strings, standing barefoot on a miniature clay sumo ring on a wooden tabletop, inside a softly blurred sumo arena with a hanging shrine roof, purple tassels and warm bokeh lights. A giant human index finger is already touching his bare belly from the right. 0–1.5s: The giant finger pulls back slightly, then pokes his round belly. The soft belly dimples inward and jiggles, his cheeks puff up comically and his eyes bulge in surprise. His body bounces back to its normal shape. 1.5–3s: The finger pokes his belly a second time, a little harder. His cheeks inflate even more, his eyes go wide and his lips purse into a funny shocked "oh!". He wobbles on his bare feet, the sagari strings swinging. 3–4.5s: The giant hand rises and gently presses down on top of his head with a fingertip, squishing his soft rubbery body down into a flat round pancake shape on the clay ring, playful cartoon squash. Hold the flattened shape for a brief beat. 4.5–5.5s: He springs back up into his original shape like a soft rubber toy, with a quick elastic overshoot and settle. 5.5–7s: He shakes himself off, tugs his mawashi back into place with both hands, then turns his face toward the giant finger and gives an exaggerated angry glare, puffed cheeks and deeply furrowed eyebrows. Hold on his funny angry expression until the end. Style: photorealistic miniature figurine, soft rubber skin texture, physically believable belly deformation combined with playful cartoon squash-and-stretch, expressive face, macro tabletop photography, shallow depth of field, warm arena lighting with soft bokeh, natural human hand movement, cute, funny and harmless mood. Camera: locked-off shot matching the input image framing, vertical 2:3, eye-level macro view, no camera movement, the sumo wrestler stays fully visible on the miniature ring throughout. Keep the sumo wrestler's face, topknot, body shape and navy mawashi consistent throughout the entire video.

来自 @AI__TSUBAKI看原帖
悬停预览 · 点击有声

雨中停车场自拍倾诉

Raw vertical 9:16 video recorded on a phone's front camera, one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft image, digital noise and compression, flat grey daylight through rain-streaked windows, no colour grading. the picture wobbles whenever she moves and jolts when she pulls the door shut. Sadie, a young woman in her early 20s with long honey-blonde hair in a messy bun with loose strands, no makeup, small gold hoop earrings and an oversized sage-green hoodie with no print, sits in the driver's seat of a small, slightly messy hatchback parked in a supermarket car park on a rainy afternoon. Rain streaks every window; through the rear window, blurred parked cars and a trolley bay. At the start she drops into the seat and pulls the door shut. She tells the story with her hands. Around 18 s she covers her face with both hands. Around 26 s a trolley rattles past her window and she slides down low in the seat as if hiding, laughing. [Sadie: casual young woman's voice, a bit breathless and mortified, recorded on the phone's built-in mic in a small car]: "Okay. I need to tell someone this immediately. ... So I'm in the cereal aisle, and this guy waves at me. Like, a big wave. So I wave back. Big wave. ... He was waving at his wife. Who was right behind me. ... And then I saw them again at the checkout. ... I'm never going back in there. I live in the car now. This is my home." Audio as captured by the phone mic inside a closed car, not studio clean: rain drumming on the roof and windscreen, her voice close and boxy in the small space, the door's heavy thud, fabric rustling against the seat, and outside, muffled through the glass, a trolley rattling past, a car door slamming and an engine starting. No subtitles, no captions, no on-screen text.

悬停预览 · 点击有声

沙滩日落自拍手机视频

Raw vertical 9:16 selfie video shot on a phone's front camera held at arm's length while she sits on the sand, one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft, a little digital noise and compression, deep focus so the sea behind her is as sharp as her face, a low golden sun to one side with lens flare drifting across the frame, auto-exposure dimming and brightening as her head blocks and reveals the sun, no colour grading. Handheld and relaxed: the phone sways gently and tilts as she leans back on her other hand. Sadie, a young woman in her early 20s with long honey-blonde waves blowing across her face in the wind, light natural makeup, small gold hoop earrings and an oversized white knit sweater with no print, sits on cool sand on a wide, quiet beach at sunset, knees pulled up. Behind her: low waves rolling in and spreading foam up the sand, the sun sinking toward the horizon, a couple walking a dog along the waterline in the distance, a few gulls. Around 8 s she tucks her hair behind her ear and the wind blows it straight back. Around 16 s a bigger wave runs further up the sand behind her and she glances back at it. Around 24 s she turns her face into the sun for a moment, eyes closed, then smiles back at the camera. [Sadie: casual young woman's voice, soft and unhurried, a little buffeted by the wind, recorded on the phone's built-in mic]: "Okay. Long week. ... I just needed to sit somewhere and not look at my phone. ... Which, yeah, I'm on my phone. I know. ... But look at this. ... I'm staying till it's gone. ... Okay. Bye." Audio as captured by the phone mic outdoors on an open beach, not studio clean: waves rolling in and breaking softly, the long hiss of foam sliding up the sand and pulling back, wind buffeting the mic in gusts, gulls calling overhead, a dog barking far down the beach, faint voices from people far away, the sand crunching as she shifts. Her voice close on top with no echo at all, sometimes partly swallowed by a gust. No subtitles, no captions, no on-screen text.

悬停预览 · 点击有声

机场登机口清晨自拍

Raw vertical 9:16 selfie video shot on a phone's front camera held low at chest height while she sits, one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft image, a little digital noise and compression, deep focus so the gate area behind her is as sharp as her face, flat white overhead light mixed with pale daylight from the big windows, no colour grading. The phone rests in her hand near her knee, so the angle is a little low and drifts as she shifts in her seat. Sadie, a young woman in her early 20s with long honey-blonde waves in a loose low ponytail, no makeup, small gold hoop earrings, an oversized grey crew-neck sweatshirt with no print and black leggings, sits in a row of connected grey seats at an airport gate early in the morning, a small carry-on suitcase beside her knee and a paper coffee cup with no logo in her free hand. Behind her: a wall of tall windows with a parked plane on the tarmac in pale morning light, other travellers sitting with bags, a man on a phone call pacing slowly, someone asleep with their hood up, a gate desk far off with screens too small and distant to read. Around 9 s she sips her coffee. Around 15 s she glances over her shoulder toward the gate desk as an announcement plays, listens, then turns back. Around 23 s she yawns behind her hand mid-sentence and laughs at herself. [Sadie: casual young woman's voice, quiet and a little croaky from the early start, keeping it down because she's in a public place, recorded on the phone's built-in mic]: "Okay, it is way too early. ... I've been up since four. ... But I'm surprising my best friend for the weekend, and she has no idea, so... worth it. ... Wait, is that us? ... No. Not us. ... Okay. Forty minutes. I'm getting another coffee. Don't judge me." Audio as captured by the phone mic in a huge, hard-surfaced airport terminal, not studio clean: her voice soft and close on top, with the wide, echoey hum of the terminal all around it; at around 15 s a two-tone chime and then a boarding announcement over the PA, distant, echoing and unintelligible; suitcase wheels rolling past on the hard floor; a coffee machine hissing at a kiosk; low murmured conversations from nearby seats; a child chattering a few rows away; a jet engine whining faintly through the glass; her sweatshirt rustling when she moves. No subtitles, no captions, no on-screen text.

悬停预览 · 点击有声

阳光街道漫步自拍

"Raw vertical 9:16 selfie video shot on a phone's front camera held at arm's length, one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft image, a little digital noise and compression, deep focus with the background as sharp as her face, auto-exposure brightening and dimming as she walks between sun and tree shade, no colour grading. Handheld and imperfect: the phone bobs and sways with each step, tilts off level, drifts until the top of her head is cut off, and she re-centres it. [Sadie], a young woman in her early 20s with long honey-blonde waves blowing across her face, light natural makeup, small gold hoop earrings, a cropped cream knit top and light-wash jeans, walks forward along one long, straight, tree-lined downtown street in bright late-morning sun, a plain iced coffee cup with no logo in her free hand. She never stops, turns around or turns a corner; behind her is the same street she has already walked, with a few passers-by, parked cars and shopfronts with unreadable signs. Around 12 s a flower stall slides past beside her and she glances at it; around 17 s she passes a bakery window, glances at it and laughs.[Sadie: casual young woman's voice, a little breathless from walking, recorded on the phone's built-in mic]: "Okay, walk with me. I'm meeting my sister for brunch and I'm, like, ten minutes early for once. ... I love this street in the morning, before all the people. ... Oh, the flower stall's out again, so pretty. ... And that bakery... no. Be strong. ... Okay, it's just up here. I'll show you what we get. Bye!"Audio as captured by the phone mic, not studio clean: her voice close and slightly compressed, light wind buffeting the mic, traffic passing, a bus pulling away, footsteps, snippets of passer-by chatter, all mixed under and around her voice."

悬停预览 · 点击有声

女性面部特写眼神

9:16で画面固定で顔面だけが映っている20代前半の女性。女性の髪型ミディアムショートで肌が白く透き通っている。顔はやや斜め正面に映っている。二重で大きな茶色い目をしている。自然光が女性に向けられて明るい。スタートは女性の目線がカメラに向けられている。 女性の額の前に葉っぱがピンボケして映っていて少し揺れている。 0s-2s:カメラを見ている。 2s-4s:伏し目がちになり一瞬右に目を逸らす 4s-5s:目線が下に下がる。 5s-6s:瞬きをしてカメラ目線に戻る 6s-10s:カメラ目線のまま見続ける。

来自 @wataru_ai_film看原帖
悬停预览 · 点击有声

伦敦街头漫步介绍

Feature a woman with long light-brown hair, a white ribbed V-neck top, and light blue jeans. Make a single, roughly 14-second handheld walking introduction on a brick-lined London street. Frame her in a medium close-up with direct eye contact, keeping the camera moving backward at her walking pace. Allow a gentle handheld bob rather than perfectly stabilized movement.

来自 fal.ai看原帖
悬停预览 · 点击有声

开朗博主自我介绍

A vlogger-style video in which the woman cheerfully introduces herself to the audience. Maintain a natural, authentic look and realistic image quality.

来自 fal.ai看原帖

夜晚城市自拍漫步

10秒夜晚城市自拍视频,9:16竖版,真实抖音/小红书夜游vlog质感。一个明确成年的东亚女生【参考图】边走边用前置手机拍自己,画面下方始终能隐约看到伸出的自拍手臂作为锚点。她长发被夜风吹动,穿普通浅色针织外套和碎花连衣裙,清爽漂亮,日常出门感。场景是夜晚建筑雨棚下的人行道,停着的车、玻璃墙反光、远处楼灯和路灯形成散景,所有车牌和招牌不可读。镜头编码:Z4中近景逐渐靠近到Z3近景,Y5轻微俯视,X1正面自拍,手机前置广角轻微边缘畸变,手持微震,画面轻微倾斜。0-2秒:镜头从略低和轻微晃动开始,快速抬起找到她的脸,她正在笑,头发扫过脸颊,前2秒立刻有自拍手臂+夜风头发hook;2-5秒:她一边走一边看向镜头,嘴角上扬,背景车灯拖影,暗部有噪点和H.264压缩块;5-7秒:手机抬成更高角度,她侧头看向街边,再看回镜头,眼神短暂锁定0.5秒;7-9秒:她靠近镜头小跑半步,笑得更开,画面短暂虚焦;9-10秒:焦点恢复,停在她开心回看镜头的表情。声音为轻快中文流行节拍、脚步声、风声、远处车流。禁止文字字幕LOGO水印,禁止清晰车牌和路人脸,禁止过度美颜、棚拍灯光、低俗身体特写、未成年感、脸部漂移。

去 X 看成片

纽约街头旅行短片

Create a 30-second, 16:9 landscape travel vlog set on the streets of New York. All audible dialogue must be in English. Character and environment Use Image 2 as the reference for the adult female protagonist. Preserve her face, long wavy hair, black sunglasses, long black coat, black outfit, and tall black boots. The reference shows multiple views of the same person; only one protagonist appears in the video. The story takes place in Manhattan. Use Image 1 as a reference for the architecture, street depth, older building façades, and storefront awnings. Do not copy its elevated camera angle. Show contemporary New York in natural daylight, with yellow taxis, pedestrians, and street-level shops. A small film crew is shooting an action scene beside the sidewalk. Cameras, lighting stands, crew members, and a monitor occupy a designated curbside filming area. The protagonist initially overlooks this equipment. A crew member mistakes her for an actor wearing a black coat and directs her to a position on set. She assumes she is simply being asked to clear the walkway, cooperates, and accidentally enters the shot. Camera style The entire video consists of footage recorded by the protagonist using her phone’s front-facing camera. She holds the same phone horizontally in her right hand throughout, approximately an arm’s length from her face and slightly above chest height. Frame her from the head to the chest or waist, using a wide view that preserves the street and background action. Her right forearm occasionally enters the bottom of the frame; the phone itself remains outside the image. When addressing viewers, she looks into the phone lens. When speaking to crew members, she turns toward them before naturally looking back at the lens. Avoid an uninterrupted, mechanical stare into the camera. Maintain authentic handheld selfie footage: walking bounce, small framing changes from arm fatigue, a brief tilt when something unexpected happens, and deliberate corrections as she reframes herself. Keep the background recognizable without strong portrait-mode blur. Never switch to third-person coverage, aerial footage, or cinematic reverse angles. 0–5 seconds | Enjoying a New York morning The protagonist is already visible in the first selfie frame, walking lightly along a Manhattan sidewalk. Her hair and open coat move naturally with her steps. She looks into the lens with a small smile and says casually: “New York, no plans. Just enjoying the day.” She briefly glances toward a storefront, then returns her attention to the lens. Ordinary traffic and footsteps fill the background. 5–10 seconds | Mistaken for an actor A clear, businesslike crew member’s voice calls from offscreen left: “Black coat! Over here, please!” She turns toward the voice, slows down, and points to herself with her free left hand: “Me?” The crew member enters the left edge of the frame, wearing a headset and a crew badge. He points toward a position near the curb and nods. Assuming she is blocking the walkway, she politely takes two steps sideways. The phone rotates slightly with her body, revealing camera equipment and lighting stands behind her. She continues holding up her phone, still unaware that she has been mistaken for an actor. 10–16 seconds | Accidentally entering the action scene Hard cut forward a few seconds, maintaining the same selfie perspective. She now stands on the indicated spot, her face on the left side of the frame, with the filming area clearly visible to her right and behind her. Two adult stunt performers run past approximately three meters behind her. The leading performer stops and turns. The pursuer throws a staged punch with visible clearance; the other performer dodges sideways, followed by a brief arm grab and escape. Their movements have a rehearsed rhythm. Neither touches the protagonist. The sudden running footsteps make her shoulders flinch, interrupting her attempt to speak. She looks behind her, then back into the phone lens and says quietly, with genuine uncertainty: “Wait… am I in the movie?” Her phone-holding hand dips briefly, tilting the image. She then tries to fit both her face and the action behind her back into the frame. Her left hand rises halfway, as though she wants to ask where she should stand but does not know whom to ask. She neither screams nor joins the fight. 16–24 seconds | “Cut!” and the misunderstanding A loud male voice calls from offscreen: “Cut!” The two performers stop in succession, relax, and return toward their starting positions. Do not make everyone freeze simultaneously. An assistant director wearing a headset and carrying a folded call sheet approaches briskly from the right background. He stops approximately one meter beside her. She rotates her phone-holding arm to include both of them naturally in the same selfie frame. Looking mildly puzzled, he asks: “You’re our actor, right?” She turns toward him, gently shakes her head, and points toward her phone with her left hand: “No, just filming my vlog.” Realizing his mistake, the assistant director relaxes his shoulders, smiles apologetically, and gestures toward the clear sidewalk: “My mistake. Sorry!” Only after hearing the apology does she relax, nod, and let out a brief laugh. The exchange flows naturally, with small reaction pauses and no overlapping dialogue. 24–30 seconds | Leaving the set and sharing her reaction Hard cut to her walking along the same sidewalk again, having left the filming position. The crew remains visible behind her, resetting the scene. She exhales, looks into the lens with some lingering disbelief, then begins to laugh: “I came for a walk. Apparently, I’m an action star now.” Deliver the first sentence as an earnest recap. Let the self-deprecating amusement emerge during the second. Leave approximately one second at the end as she lightly shakes her head and keeps walking. End directly. Performance and sound Give the protagonist one consistent, natural, clear adult female English voice with a medium-low register and conversational delivery. She begins relaxed; after accidentally entering the scene, her speech becomes shorter and quieter. Once the misunderstanding is resolved, her delivery includes an exhale and spontaneous amusement. Avoid announcer-style delivery, continuous recitation, or forced comic acting. Her sunglasses remain on her face throughout. Express emotion through her mouth, chin, head turns, shoulders, and pauses rather than close-ups of eyes hidden behind the lenses. The crew member and assistant director have distinct adult male voices with natural American accents. Each line is spoken only by its assigned character. The offscreen “Cut!” must not be spoken by the protagonist. Add no other intelligible dialogue. Preserve New York traffic, footsteps, wind, clothing movement, running performers, and subdued crew ambience. Keep dialogue intelligible. No music, narration, subtitles, or text overlays. The chase and fight must remain recognizable as a staged film production: no weapons, blood, wounds, explosions, or actual injuries. The protagonist accidentally enters the shot and becomes involved in an identity mix-up; she is not attacked. Use real-time motion throughout and only the two specified hard cuts. Maintain continuity of identity, wardrobe, phone-holding hand, and street geography.

去 X 看成片

中西部政治自拍视频

20-SECOND VERTICAL VLOG (9:16) — "VOTE RED: WHY I'M SHOWING UP" FORMAT: 9:16, 30fps, handheld selfie-style smartphone footage, one continuous take with two jump cuts. Natural daylight, slightly imperfect UGC realism, front-camera wide lens, light auto-exposure shifts. CREATOR (fictional, no real person): Marcus, 28, Black American man, short fade, olive bomber jacket over a plain white tee, small silver chain. Warm, direct, conversational energy. Same wardrobe and face in every shot. WORLD: Small-town main street in the American Midwest, late afternoon golden light, brick storefronts, a hardware store, an American flag hanging from a porch, a few parked pickups. A neighborhood polling place with a plain hand-painted sign stands behind him in the last shot. SHOT 1 (0-6s): Selfie close-up, Marcus walking down the sidewalk, arm extended, breath visible in cool air. He smiles at camera and says: "Midterms are November third. I'm not sitting this one out, and neither should you." Passing cars and footsteps in the background. SHOT 2 (6-13s): Jump cut. He stops beside a diner window, leaning on the frame, gesturing with his free hand. He says: "Rent, groceries, gas, they all add up. I'm voting red because I want lower costs, safer streets, and a say in where this country goes." A waitress inside laughs at something off-camera. SHOT 3 (13-18s): Jump cut. Camera flips to the street, panning across neighbors chatting, a kid on a bike, a man opening the polling place door. Marcus narrates over his shoulder: "Whatever you believe, your vote is the loudest thing you own." SHOT 4 (18-20s): Camera flips back to selfie. Marcus stands in front of the polling place, holds up a paper "I Voted" sticker, grins, and says: "Check your registration. Show up. Vote red." He gives a firm nod. LAST FRAME: Marcus centered, smiling, holding the sticker toward camera, polling place softly blurred behind him. No on-screen text, no captions, no subtitles, no watermarks, no logos, no readable signage anywhere in frame. AUDIO (diegetic only): Marcus's natural voice with slight phone-mic compression, footsteps on pavement, distant traffic, wind, faint street chatter, a door opening. No music, no voiceover, no sound effects that aren't in the scene. CONSTRAINTS: No real public figures, no candidate names, no party logos, no campaign slogans on signs, no ballots or voter information shown. Faces stable, hands natural with five fingers, no morphing, no flicker.

去 X 看成片

猫咪推杯掉落

Realistic 16:9 home video, 10 seconds, a bright modern kitchen in daylight, static camera at counter height. A fluffy orange tabby cat sits on the wooden floor, staring up at the white kitchen counter, its tail swishing. It wiggles its back legs and jumps, landing smoothly on top of the counter on all four paws. It walks slowly along the counter to a white coffee mug standing near the edge. It looks straight at the camera, then pushes the mug with one paw until it tips over the edge and falls out of frame. A loud crash of the mug breaking on the floor. The cat sits down calmly on the counter, unbothered. Sounds: soft paws on wood, a light thud as it lands, gentle footsteps on the counter, the crash of the mug, a quiet meow.

去 X 看成片

夜市炒锅火焰

Vertical handheld shot for a social feed. A street-food cook tosses noodles in a flaming wok at a night market; flames leap up, he slides the noodles onto a banana leaf, looks into the camera with a grin and asks: "Spicy?" Sound: sizzling oil, crackling flames, market noise.

去 原帖 看成片
来自 kling4.video看原帖

2010年代精品店直播带货

Create a 30second ultra realistic video featuring the character from @.jpg (maintaining her exact pinkish purple bob hair, heterochromia eyes one green and one amber, face markings, and facial features) hosting a live shopping session inside a trendy boutique in 2010. The video is captured entirely through an early smartphone front-facing camera (like iPhone 4 era), featuring authentic 2010 mobile livestream aesthetics: heavy pixelation, low resolution, digital compression artifacts, grainy texture, and slightly shaky handheld movements. Scene 1 — Welcome to the Live (0–5 seconds) Choi Yu-na stands inside a 2010 style women's boutique with neon signage and racks filled with early 2010s fashion trends, wearing her signature black bomber jacket over a black top and gray pleated skirt. She holds her early smartphone at arm's length, smiling into the low res pixelated camera and greeting her viewers. Dialogue: “Hey guys! Welcome to my mobile live stream! I'm here at the boutique today, and you guys have to see these new arrivals!” She briefly pans the pixelated camera toward the clothing racks, then returns it to her face. Scene 2 — Show the First Dress (5–10 seconds) She picks up a trendy 2010 color-block party dress from the rack, holding it up close to the low res camera lens to show its fabric and design. Dialogue: “Look at this color block dress! It's so hot right now, and the fabric is super cute for a night out!” She squints slightly at the screen as if reading early mobile chat comments. Scene 3 — First Outfit Try On (10–16 seconds) A pixelated, glitchy transition shows her wearing the same color-block dress, standing in front of a full-length boutique mirror while maintaining her exact face, hair, and unique eye colors. She poses and takes a few steps for the camera. Dialogue: “Okay everyone, here it is on! How does it look? I'm totally loving this style!” She smiles and waves at the viewer names popping up on her phone screen. Scene 4 — Second Outfit Reveal (16–23 seconds) A quick, low-res transition brings her back wearing a classic 2010 sequin top and skinny jeans set. She adjusts her hair, steps forward, and shows off the outfit. Dialogue: “And check out this sequin top and denim combo! Perfect for weekend hangs with friends!” Her delivery feels spontaneous, casual, and true to early social media video creators. Scene 5 — Final Sales Pitch (23–30 seconds) She stands near the boutique display wearing the second outfit while holding the color-block dress on a hanger beside her, smiling brightly into the phone. Dialogue: “Alright guys, let me know in the comments which one you like best! Send a text message or leave a comment to grab yours before they sell out. Bye!”

去 X 看成片

路边皮卡时尚短片

Style: Cinematic Narrative Storytelling, vlog editorial hybrid with dynamic handheld FPV flow. Subjects: The subject from @imagem1, wearing the ecru crochet halter top, yellow floral high-waisted flared trousers, patterned bucket hat, and beaded bracelets, barefoot., Preserve the subject’s facial identity and body proportions throughout; exclude face drift and anatomical deformation. Environment: Rural roadside beside an olive-green and cream vintage pickup truck with an open door draped with a colorful zigzag blanket, bathed in golden hour sunlight with soft tree bokeh. Mood: Editorial lifestyle with contained joy, spontaneous smiles, and warm intimate glances. Continuity: Preserved subject wardrobe, truck placement, late afternoon lighting, and physical geography across all cuts. Constraints: No dialogue, no voiceover, no text overlays, no anatomical distortion, no music. Vathography: Layered depth mapping. Foreground truck door frame and chrome highlights at 0.5m, subject at 1.2m, truck interior at 1.8m, and sunlit background trees falling off into soft bokeh from 4m onward. Shot 1 (0 to 1.88 seconds) Purpose: Hook viewer with intimate facial expression and spontaneous warmth. Framing: Close-up on the subject's face, warm directional backlight. Action: The subject tilts their head slightly, locking eyes with the lens as lips part into a soft smile. Acting: Direct magnetic eye contact, subtle brow lift, relaxed natural breathing. Camera: FPV Tracking rapidly pushes in toward eye level, stabilizing gently on facial features. Constraints: Keep exact facial structure and hat texture intact. Transition: Quick cut on subject's slight head tilt to medium profile angle. Sound effects: Soft breeze rustling roadside leaves. Shot 2 (1.88 to 3.75 seconds) Purpose: Reveal wardrobe styling and truck interaction. Framing: Medium shot angled along the truck door threshold. Action: The subject adjusts the brim of the bucket hat with delicate fingertips while shifting weight. Acting: Contained smile, eyes flicking downward before returning affectionately to lens. Camera: FPV Tracking arcs low-to-high along the open door edge. Constraints: Maintain fabric texture of crochet top and floral trousers. Transition: Hard cut on hand movement settling on hat brim. Sound effects: Subtle fabric rustle and distant foliage whisper. Shot 3 (3.75 to 5.63 seconds) Purpose: Emphasize tactile textures of the zigzag blanket and vehicle chrome. Framing: Medium close-up focused on subject torso and door frame. Action: The subject rests one forearm along the truck frame, leaning gently forward. Acting: Confident editorial posture, calm eye contact, lips gently closed. Camera: FPV Tracking banks rapidly past the draped blanket toward the subject's torso. Constraints: No artificial camera shake, fluid motion path. Transition: Whip cut matching rightward camera momentum. Sound effects: Faint metallic door creak under gentle arm weight. Shot 4 (5.63 to 7.5 seconds) Purpose: Capture candid laughter and sun flare reflection. Framing: Close-up framing face and shoulders against cream truck cab. Action: The subject breaks into an uninhibited, spontaneous grin with head tilted back. Acting: Genuine laughter, crinkled eye corners, glowing sunlit cheeks. Camera: FPV Tracking dips and lifts rapidly in handheld pursuit rhythm. Constraints: Natural tooth alignment, no morphing. Transition: Direct cut on subject's outward breath. Sound effects: Quiet exhale synchronized with breeze. Shot 5 (7.5 to 9.38 seconds) Purpose: Ground the lifestyle setting via barefoot interaction with asphalt. Framing: Low-angle medium shot framing bare feet and flared trousers. Action: Bare feet flex subtly on warm pavement as heels lift in relaxed rhythm. Acting: Carefree, grounded body language. Camera: FPV Tracking glides near ground level, tilting upward toward the hemline. Constraints: Keep road surface and vehicle tires consistent. Transition: Quick cut on foot stepping back down. Sound effects: Gentle scrape of bare skin on sun-warmed asphalt. Shot 6 (9.38 to 11.25 seconds) Purpose: Re-engage direct intimate connection. Framing: Tight close-up focusing on eyes and lips. Action: The subject lowers their chin slightly, peering playfully from beneath the hat brim. Acting: Intimate half-smile, magnetic gaze locking onto the lens. Camera: FPV Tracking darts forward, executing a micro-refraction before holding focus. Constraints: Stable lighting contrast and skin highlights. Transition: Cut on eye blink. Sound effects: Faint whisper of wind through hair. Shot 7 (11.25 to 13.13 seconds) Purpose: Showcase full ensemble and spatial context. Framing: Full medium shot centered in the truck cabin doorway. Action: The subject raises both arms overhead, touching the door frame interior. Acting: Relaxed editorial pose, elongated torso, poised smile. Camera: FPV Tracking accelerates backward, expanding the frame. Constraints: Keep truck interior upholstery identical. Transition: Dynamic cut on arms reaching peak position. Sound effects: Muffled resonance of truck cab metal. Shot 8 (13.13 to 15 seconds) Purpose: Highlight crochet texture and golden hour skin tones. Framing: Medium close-up of chest and shoulder profile. Action: The subject turns torso toward the light, dropping one hand gracefully to the waist. Acting: Sophisticated editorial gaze into middle distance. Camera: FPV Tracking circles smoothly across subject's front quadrant. Constraints: Accurate knit pattern and strap placement. Transition: Cut as subject head turns back forward. Sound effects: Soft fabric sliding across skin. Shot 9 (15 to 16.88 seconds) Purpose: Inject energetic candid movement. Framing: Medium shot from passenger side perspective looking across. Action: The subject leans out of the doorway, laughing and swaying slightly. Acting: Playful spontaneity, relaxed shoulders, open expression. Camera: FPV Tracking rushes through the cab window opening. Constraints: Preserved truck cab dimensions and dashboard elements. Transition: Fast cut matching subject's forward sway. Sound effects: Gentle metallic vibration of the vintage frame. Shot 10 (16.88 to 18.75 seconds) Purpose: Macro focus on handcrafted accessories and hat brim. Framing: Tight macro shot of fingertips adjusting floral hat edge. Action: Fingers lightly pinch the stitched brim fabric, rotating slightly. Acting: Deliberate, graceful hand coordination. Camera: FPV Tracking zooms in tight along the arm's trajectory. Constraints: Natural hand anatomy, exact bracelet beads. Transition: Cut on hand releasing brim. Sound effects: Micro-friction of woven cotton fabric. Shot 11 (18.75 to 20.63 seconds) Purpose: Dynamic visual perspective of yellow floral trousers. Framing: Diagonal low-to-high medium shot. Action: The subject shifts weight between legs, causing pant leg fabric to drape elegantly. Acting: Poised, statuesque hip alignment and serene expression. Camera: FPV Tracking banks steeply upward from asphalt to waist. Constraints: Consistent print and flare silhouette. Transition: Snap cut on hip shift settling. Sound effects: Soft foot pivot on pavement. Shot 12 (20.63 to 22.5 seconds) Purpose: Intimate over-the-shoulder connection. Framing: Close-up over truck door frame toward subject's face. Action: The subject peeks over the window edge, catching the lens with an affectionate glance. Acting: Tender, knowing smirk, soft eye squint. Camera: FPV Tracking loops over chrome mirror and door rim. Constraints: Chrome reflection stability. Transition: Hard cut on subtle smile widening. Sound effects: Ambient tree canopy rustle. Shot 13 (22.5 to 24.38 seconds) Purpose: Reveal overall silhouette against late afternoon sun. Framing: Wide medium shot backlit by warm rim lighting. Action: The subject steps one foot back onto the truck floorboard, extending posture. Acting: Confident fashion posture, chin lifted, peaceful smile. Camera: FPV Tracking pulls out diagonally, introducing golden lens flare. Constraints: Flare must not wash out wardrobe contrast. Transition: Cut on foot contact with metal cab floor. Sound effects: Muted clink of foot meeting metal floorboard. Shot 14 (24.38 to 26.25 seconds) Purpose: Rapid kinetic push-in to spontaneous joy. Framing: Medium close-up transitioning rapidly to close-up. Action: The subject laughs softly, shaking their head with genuine ease. Acting: Spontaneous joy, relaxed jaw, shining eyes. Camera: FPV Tracking dives quickly toward subject's face. Constraints: Keep facial bone structure intact during rapid motion. Transition: Quick cut at closest proximity point. Sound effects: Soft breathy laugh exhale. Shot 15 (26.25 to 28.13 seconds) Purpose: Gentle deceleration and contemplative editorial pose. Framing: Medium close-up centered on subject framed by green door frame. Action: The subject lowers arms, resting thumbs lightly at trouser waistband. Acting: Serene, composed editorial stare directly into lens. Camera: FPV Tracking decelerates, executing a slow stabilizing float. Constraints: Precise finger articulation, no drifting features. Transition: Cut on deep chest breath. Sound effects: Subtle evening breeze rising. Shot 16 (28.13 to 30 seconds) Purpose: Final inviting emotional payoff. Framing: Close-up on radiant smiling face. Action: The subject offers a lingering, intimate smile before looking softly off-camera. Acting: Contained warmth, genuine happiness, natural blink. Camera: FPV Tracking holds steady, then gently sweeps rightward to resolve. Constraints: Maintain photographic fidelity and natural skin texture. Sound effects: Whispering leaves and calm sunset ambience fading out. Final director notes: The film must inhabit a sun-drenched, carefree 1970s editorial register with modern handheld immediacy. The tactile richness of the crochet weave, yellow floral drape, and olive-and-cream metal must pop in every frame under natural warm light. Shot 16 provides the emotional anchor, leaving the audience with an authentic feeling of golden-hour nostalgia and effortless intimacy. Audio: Diegetic sound effects: Tie every cue to a visible physical action or a real environmental source. No soundtrack. Use only synchronized diegetic sound effects and ambience supported by the scene.

去 X 看成片

西西里小巷柠檬时尚

Style: Cinematic Narrative Storytelling with dynamic vlog realism and high-energy home-video vitality. Subjects: The subject from @imagem1, maintaining facial identity, dark wavy hair, gold wire sunglasses, delicate pendant necklace, and red linen ruffled wrap dress., Preserve the subject’s facial identity and body proportions throughout; exclude face drift and anatomical deformation. Environment: Sunny Sicilian village alleyway with warm ocher plaster walls, rustic limestone facades, and stone cobblestones drenched in golden Mediterranean sunlight. Mood: Happy; overflowing with spontaneous joy, radiant smiles, and warm exuberant body language. Continuity: Exact wardrobe, accessories, hair, props (fresh yellow lemons), and warm lighting continuity across all cuts. Constraints: No gendered terms. Never touch or lean on walls. No music or dialogue. Strictly utilize FPV Tracking movement only. Vathography: Foreground framing elements create parallax; subject occupies the central depth plane with sharp volumetric separation against soft, warm background village walls. Shot 1 (0 to 1.5 seconds) Purpose: Hook viewer with an intimate, radiant smile in motion. Framing: Close-up, eye-level, 35mm perspective. Action: The subject strides forward into sunlight, holding a bright lemon in one hand. Acting: A spontaneous Duchenne smile blooms across the subject's face, gaze locking magnetically onto the lens. Camera: Dynamic FPV Tracking moves backward rapidly, matching the subject's forward pace. Constraints: Maintain steady distance; keep face tack-sharp. Transition: Whip cut on subject's slight turn to incoming wider tracking frame. Sound effects: Soft step on stone, crisp rustle of linen fabric. Shot 2 (1.5 to 3 seconds) Purpose: Reveal environment and full silhouette in motion. Framing: Medium shot, slight low-angle. Action: The subject walks along the cobblestones, tossing and catching a lemon with light fingers. Acting: Buoyant, joyful posture; head tilted slightly back with a delighted grin. Camera: Low-slung FPV Tracking glides ahead in rapid pursuit, tilting upward subtly. Constraints: No physical contact with walls; keep movement centered. Transition: Match-cut on lemon toss to close detail. Sound effects: Rhythmic shoe taps on cobblestones, soft air displacement from lemon toss. Shot 3 (3 to 4.5 seconds) Purpose: Highlight tactile texture of fruit and linen. Framing: Extreme close-up on hands and mid-torso. Action: The subject rolls two yellow lemons between nimble fingers against the red dress. Acting: Relaxed, joyful finger articulation showcasing carefree confidence. Camera: FPV Tracking sweeps closely across the subject's hands, banking right. Constraints: Prevent motion blur on the lemons; keep wardrobe texture sharp. Transition: Fast spatial cut on camera bank into an over-the-shoulder frame. Sound effects: Subtle citrus skin squeak against fingertips, breeze fluttering ruffled sleeve. Shot 4 (4.5 to 6 seconds) Purpose: Capture over-the-shoulder perspective exploring the sunlit path. Framing: Medium close-up, high over-the-shoulder angle. Action: The subject strides ahead toward a sun-drenched stone archway. Acting: Light, brisk pace; shoulders open, joyful head turn catching backlight. Camera: High FPV Tracking follows closely behind, swooping gently over the subject's shoulder. Constraints: Maintain fluid motion path without wobbling. Transition: Direct cut as subject pivots around. Sound effects: Gentle breeze murmuring through alleyway, light footsteps. Shot 5 (6 to 7.5 seconds) Purpose: Intimate frontal face reveal with spontaneous laughter. Framing: Tight close-up, eye-level. Action: The subject spins back toward the lens mid-stride. Acting: Bright eyes crinkling behind sunglasses, lips parting in a brief, genuine laugh. Camera: FPV Tracking pulls back quickly, banking laterally to cushion the turn. Constraints: Keep facial features stable and clear despite sudden turn. Transition: Fast whip cut matching lateral camera movement. Sound effects: Rapid fabric swoosh, soft breath of laughter. Shot 6 (7.5 to 9 seconds) Purpose: Showcase vibrant color contrast and open body posture. Framing: Medium wide shot, eye-level. Action: The subject steps into open golden sunlight with arms casually outstretched. Acting: Expansive, exuberant posture; radiant grin beaming skyward. Camera: Fast FPV Tracking circles semi-orbitally around the subject. Constraints: Keep subject well clear of background architecture. Transition: Cut on circular momentum into medium framing. Sound effects: Ambient village echo, rustling ruffled linen hem. Shot 7 (9 to 10.5 seconds) Purpose: Re-engage close emotional connection with the camera. Framing: Close-up, slightly high angle. Action: The subject lifts sunglasses slightly with one hand, flashing direct eye contact. Acting: Playful, inviting eye contact, warm smile tugging at lip corners. Camera: FPV Tracking darts closer, stabilizing into a swift forward push. Constraints: Sunglasses stay balanced; eyes fully visible. Transition: Cut on sunglasses adjustment to a lower perspective. Sound effects: Subtle click of wire frames, quiet intake of breath. Shot 8 (10.5 to 12 seconds) Purpose: Dynamic dynamic stride along cobblestone alley. Framing: Medium shot, low-angle. Action: The subject walks briskly away from camera, swinging arms naturally. Acting: Lively, bouncy stride full of carefree energy. Camera: Low FPV Tracking chases closely behind the moving hemline. Constraints: Stable tracking along uneven stone ground. Transition: Cut on footstep impact into reverse angle. Sound effects: Firm, distinct heel taps echoing off rustic stone. Shot 9 (12 to 13.5 seconds) Purpose: Catch a candid backward glance over the shoulder. Framing: Medium close-up, dynamic eye-level. Action: The subject glances back over one shoulder toward the camera while advancing. Acting: Playful knowing smirk, eyes sparkling with pure exuberance. Camera: FPV Tracking overtakes and matches subject speed alongside the shoulder. Constraints: Clean profile without facial warping. Transition: Fast whip cut matching camera acceleration. Sound effects: Whispering summer wind, soft garment friction. Shot 10 (13.5 to 15 seconds) Purpose: Accentuate bright Mediterranean colors and tactile fruit. Framing: Close-up insert tilted toward hands and chest. Action: The subject brings one lemon near the neckline, rotating it in the sun. Acting: Graceful, fluid wrist motion, expressing casual appreciation of the day. Camera: FPV Tracking arcs tightly around the hands, highlighting specular light on fruit. Constraints: Hands and fruit remain completely in focus. Transition: Cut on wrist turn to frontal close-up. Sound effects: Faint friction of linen, soft breeze. Shot 11 (15 to 16.5 seconds) Purpose: Capture spontaneous, contained joy in a brief pause. Framing: Close-up, eye-level. Action: The subject pauses momentarily mid-alley, turning to face the lens. Acting: Radiant, authentic beam; eyelids soft and cheeks lifted in pure joy. Camera: FPV Tracking glides smoothly to an intimate halt just inches away. Constraints: Preserve natural skin texture and lighting tone. Transition: Hard cut on subject's outward arm movement. Sound effects: Gentle sigh of ambient breeze, settling footsteps. Shot 12 (16.5 to 18 seconds) Purpose: Convey lively full-body celebration. Framing: Medium shot, slightly low. Action: The subject takes a joyful skip-step forward, dress fluttering. Acting: Carefree, exuberant body language, chest lifted proudly. Camera: FPV Tracking rushes backward, keeping subject framed dynamically. Constraints: Subject remains centered during rapid footwork. Transition: Cut on landing step to high angle. Sound effects: Quick double tap of shoes, ruffled skirt fluttering. Shot 13 (18 to 19.5 seconds) Purpose: Atmospheric, dynamic top-down tracking. Framing: Medium high-angle. Action: The subject walks beneath sunlit lemon tree branches hanging over a courtyard. Acting: Upward glance, open beaming expression. Camera: FPV Tracking swoops down from above, following subject's path. Constraints: Keep camera operator reflection invisible in sunglasses. Transition: Cut on subject's forward head tilt. Sound effects: Distant rustle of foliage, clear stone footfalls. Shot 14 (19.5 to 21 seconds) Purpose: Re-establish close intimate connection. Framing: Close-up, eye-level. Action: The subject turns back toward the camera with a bright nod. Acting: Intimate, welcoming smile; warm ocular engagement. Camera: FPV Tracking glides rapidly alongside subject's cheek. Constraints: Facial proportions rock-solid. Transition: Whip cut on head turn to three-quarter view. Sound effects: Soft linen rustle, warm exhalation. Shot 15 (21 to 22.5 seconds) Purpose: Build toward an energetic finale. Framing: Medium shot, eye-level. Action: The subject extends an arm toward the lens in a playful beckoning gesture. Acting: Magnetic, delighted expression, fingers lightly splayed. Camera: FPV Tracking matches the gesture, surging smoothly forward. Constraints: Hand stays in frame without clipping. Transition: Smooth cut on forward camera thrust to final hero frame. Sound effects: Swift sleeve flutter, rhythmic footsteps. Shot 16 (22.5 to 24 seconds) Purpose: Conclude with a memorable, euphoric portrait in motion. Framing: Medium close-up, golden hour backlight. Action: The subject strolls forward, holding both lemons, giving a final radiant smile. Acting: Deep authentic Duchenne smile, posture relaxed yet beaming with fulfillment. Camera: FPV Tracking decelerates gently as subject fills the frame, settling gracefully. Constraints: Maintain facial fidelity; no wall contact; clean, photographic resolution. Sound effects: Fading footstep on stone, soft summer breeze settling. Final director notes: Maintain a warm, candid Sicilian lifestyle aesthetic anchored by authentic Italian sun. The red linen dress must pop intensely against the textured ocher stone, and the subject's spontaneous smile must remain the core emotional heartbeat of every cut. Ensure every camera movement feels like an agile, responsive pursuit without losing photographic realism or facial consistency. Audio: Diegetic sound effects: Tie every cue to a visible physical action or a real environmental source. No soundtrack. Use only synchronized diegetic sound effects and ambience supported by the scene.

去 X 看成片

阁楼试穿西装旋转

"Handheld UGC try-on clip in a sunlit loft studio. The woman swings the camel wool blazer off her shoulder, slips both arms into the sleeves one at a time, and tugs the lapels straight. Then she does one full slow spin on the spot so the pleated emerald satin skirt flares out and settles back down, and she ends facing the camera with a small smile. The wool creases at the elbows as she bends her arms, and the satin pleats catch the window light as they swing. The camera stays at the same shin-to-head framing the whole time. Realistic fabric physics, natural light, ambient room sound only, no music, no speech, no text on screen."

去 X 看成片

怎么写好 Kling 4.0 提示词

  1. 1每份素材都用 @ 点名. API 上的素材按它在请求里的位置编号:image_urls 里的是 @image_1、@image_2,video_urls 里的是 @video_1,帧是 @first_frame、@keyframe_1、@last_frame,elements 里的主体和 voices 里的音色用 @名字。提示词里每个 @ 都要对得上同一个请求里真有的素材,对不上的请求会在扣费前直接被拒。
  2. 2「附上的图」要改成 @ 点名. fal.ai 指南里不少提示词写的是「用 Video 1 的运镜,用附上的图片定场景」,离开那几个文件就跑不起来。在 API 上,把视频放进 video_urls、图片放进 image_urls,再按同样的顺序把说法改成 @video_1、@image_1。
  3. 3角色交给具名主体来锁. 与其每句话都重新描述一遍那张脸,不如在 elements 里交一次:2 到 4 张不同角度的图、正面那张放第一,或者一段 3 到 8 秒、一镜到底的人物视频。之后她每次出场都写 @Mia。本页长提示词里反复强调「全程同一个人」,这一步就是在请求层面做同一件事。
  4. 4声音关不掉,那就写清楚. Kling 4.0 的声音没法关,所以效果好的提示词都把声音写出来:kling4.video 的例子结尾都有一行「Sound:」,台词放进引号并写明语种,比如 says in English: "Sorry, was that oat milk or almond?"。想让某个角色用指定的嗓音,Kling 4.0 用 voices 绑一段音频,Flash 用 elements[].voice_url。带参考视频时,keep_original_sound 可以保留它们的原声。
  5. 5分镜写在正文里,时间要加得起来. 本页常见的写法,是在一条提示词里写「Shot 1 (0-5s):… Shot 2 (5-10s):…」。API 没有单独的分镜字段,这些分镜就是普通的提示词文字:每个镜头只放一个动作,所有时间段加起来要等于你请求的时长,3 到 30 秒(Flash 最多 20 秒)。
  6. 6画幅和时长写进参数,别写在提示词里. X 上的长提示词常常一开头就是「Format: 16:9, 30 seconds, 24fps, 1080p」,还有几条要 4K。在 API 上这些都是参数:duration、resolution(720p 或 1080p,没有 4K)和 aspect_ratio(auto、16:9、9:16、1:1、21:9)。在请求里设好,免得文字和参数打架。
  7. 7定路线用关键帧,定长相用参考. 关键帧管镜头往哪走:一张首帧、最多 8 张中间帧、一张尾帧,总张数受时长限制,3 秒最多 4 张,9 秒及以上最多 10 张。参考素材管东西长什么样。两者不能放进同一个请求,Flash 只收首帧。

提示词公式

画幅和时长放进参数(duration、resolution、aspect_ratio)+ 主体,或用 @名字 点名具名主体 + 每个分镜一个动作、标好时间 + 镜头运动 + 光线和场景 + 每份参考素材都用 @ 点名 + 一行「Sound:」,台词放进引号并写明语种

例(本库收录的原文,原样引用,作者 @Alin_Reaper05,发在 X):An exploration ship maneuvers through a vast cloud of charged gas and floating debris. Inside the cockpit, the pilot moves the controls while the captain braces against the console and an alien crew member monitors the storm. The camera moves from the crew toward the forward window, cinematic tension, realistic motion.

常见问题

Kling 4.0 是什么?Kling 4.0 Flash 有什么不同?

可灵 Kling 4.0 是快手可灵系列最新的视频模型,每条视频都带同步生成的声音,时长 3 到 30 秒,720p 或 1080p。Kling 4.0 Flash 是它的轻量档,端点和请求格式都一样,但只有 720p、3 到 20 秒,提示词最长 8000 字符,关键帧只收首帧,也不收单独的音色列表。要 1080p、超过 20 秒、要中间帧或尾帧、要单独的音色,就用 Kling 4.0。

这些提示词是谁写的?

都是别人写的:X 和 Civitai 上的创作者、kling4.video 的提示词库,以及 fal.ai 的 Kling 4 提示词指南,我们一个字都没写、也没改。每张卡片都逐字全文引用提示词、写明作者、链回出处。提示词和视频归原作者所有;我们自己写的说明和整理的数据按 CC BY 4.0 授权。要更正署名或撤下某一条,发邮件给我们,或者在 GitHub 仓库提 issue。

在哪能跑这些提示词?多少钱?

在 apimodels.app 上,模型名 kling-4.0 或 kling-4.0-flash,接口 POST /v1/video/generations,按秒计费。Kling 4.0 不带视频输入时,720p $0.157/秒、1080p $0.235/秒:720p 5 秒 $0.785,1080p 10 秒 $2.35,720p 30 秒 $4.71。Flash $0.094/秒:5 秒 $0.47,10 秒 $0.94。带参考视频时,Kling 4.0 是 $0.094 或 $0.141 每计费秒,Flash 是 $0.063 每计费秒;计费秒数在输入视频不长于成片时是成片 × 2,更长时是输入 + 成片。所以 720p 出 10 秒、带一段 5 秒参考视频,Kling 4.0 是 $1.88,Flash 是 $1.26。参考图和关键帧不另外收费,失败不扣费。

本页的提示词放到 API 上失败,为什么?

超出限制的请求会在建任务之前返回 400,不扣费。常见原因有:提示词里的 @ 找不到对应的素材;提示词超过 10000 字符(Flash 是 8000,本页最长的那条就超了);时长不在 3 到 30 秒之内(Flash 是 3 到 20 秒);要 4K,目前没有;帧和参考素材混用;参考图有一边不到 300 像素,或者宽高比不在 0.4 到 2.5 之间;参考视频读不出时长。API 也没有负面提示词、多镜头分镜和视频续写,围绕这些功能写的提示词,要改写成一条普通的提示词。

要用作者当时的参考图或参考视频吗?

只有提示词点到它们的时候才需要。纯文字的提示词直接就能跑。写着「附上的图片」「Video 1」,或者点名参考素材里某个角色的提示词,就要你自己准备素材,放进 image_urls、video_urls 或 elements,再用 @ 点名。只上传你有权使用的人像和声音。

Kling 4.0 的提示词能写多长?

Kling 4.0 最长 10000 字符,Flash 最长 8000。本页的提示词从一句话到将近 1 万字符都有,长的那些多半把字数花在固定的人物描述、带时间的分镜表和声音清单上,而不是形容词。如果长提示词里描述的参考素材你并没有传,把那一段删掉。

这些提示词和视频能商用吗?

提示词和视频归原作者所有,商用前先看原帖怎么说。你自己生成的视频,按你用来生成的那个服务的条款执行,也只用你有权使用的人像和声音。

在 apimodels.app 上跑这些提示词

Kling 4.0 已经上线,和另外 180+ 个图像、视频、语音、大模型共用一个 API、一个 key,按量计费、注册送 $0.10 试用额度。

这份提示词库在 GitHub 上是开源的

整份合集以 Markdown 和 JSON 放在 GitHub 上:每条提示词都带作者署名和出处链接(X、kling4.video、fal.ai 或 Civitai)。提示词逐字全文引用,归原作者所有。视频也归作者所有,只链原帖、不转存。我们自己写的说明和整理的数据按 CC BY 4.0 授权。要更正署名或撤下某一条,提 issue 即可。

在 GitHub 上查看

其它提示词库