APIMODELSAPIMODELS

Kling 4.0 15-30s & Multi-Shot Prompts

More than 270 Kling 4.0 video prompts that other people published: creators on X, the kling4.video prompt library, fal.ai's Kling 4 prompting guide and a few Civitai posts. Each one is quoted in full, word for word, credited to its author and linked to the original, and the card plays the clip it produced where the source has one. We did not write any of them.

← All 275 Kling 4.0 prompts

Last updated: · Kling 4.0 API — pricing & playground · API docs

TL;DR

Kling 4.0 is the newest video model in Kuaishou's Kling family, and every clip it makes comes with sound generated in the same pass. On apimodels.app one model id, kling-4.0, covers text-to-video; reference-to-video with up to 10 images, 5 videos, 7 named subjects and 3 voices that the prompt calls by @mention; and keyframe-to-video from a first frame, up to 8 middle frames and a last frame. Clips run 3 to 30 seconds at 720p or 1080p, prompts can be up to 10,000 characters, and it costs $0.157 per second at 720p and $0.235 at 1080p, with failed generations not charged. Kling 4.0 Flash (kling-4.0-flash) takes the same request at 720p only, 3 to 20 seconds, prompts up to 8,000 characters and a first frame as its only keyframe, for $0.094 per second. The API has no negative prompt, no multi-shot storyboard mode and no 4K. This page collects more than 270 Kling 4.0 prompts other people published, most of them on X.

Source: Kling AI (the model vendor’s official site)

Hover to preview · click for sound

Travel Couple Catch Scenes

Create a 30-second vertical 9:16 photorealistic live-action travel video with natural lighting, realistic physics and spontaneous smartphone footage. IDENTITY LOCK - @Image1: ACEL. Preserve her exact face, hairstyle, skin tone and body proportions. - @Image2: Male lead. Preserve his exact identity and appearance. Never change either face. REPEATED ACTION In every scene, the man is busy with his own activity and does not notice ACEL. She enters from the RIGHT, approaches playfully, makes the same small turn, then suddenly falls backward toward him. Only then does he react, step back, bend his knees and firmly catch her upper back and waist with his free arm. Show her actual weight being supported. They exchange natural smiles. Keep props in the correct hands. 7 SCENES & OUTFITS 1. 00–04 | Wooden courtyard: Man: brown shirt, white trousers, light-brown shoes, opening a blue umbrella. ACEL: dark-brown jacket, long grey skirt, dark hat, crossbody bag. He catches her while holding the umbrella. 2. 04–08 | Turquoise river: Man: pale-pink shirt, blue jeans, sneakers, filming with phone. ACEL: layered pale-pink dress, light head covering, dark platform shoes. 3. 08–12 | Sand dunes: Man: red shirt, white trousers, sandals, talking on phone. ACEL: flowing bright-red dress. He lowers the phone and catches her. 4. 12–16 | Yellow flower valley: Man: dark-green outdoor jacket, dark trousers, brown hiking boots, backpack, searching inside it. ACEL: grey-green jacket, white short skirt, brown boots. 5. 16–20 | Blue alpine lake: Man: bright-blue jacket, white shirt, brown trousers, blue cap, filming. ACEL: flowing blue dress, light wide-brim hat, dark shoes. 6. 20–24 | Lakeside path: Man: cream outfit, crossbody bag, dark shoes, drinking from a cup with straw. ACEL: dark-purple plaid dress, light scarf, black ankle boots. 7. 24–30 | Potala Palace: Man: red-and-cream jacket, cream trousers, crossbody bag, filming the palace. ACEL: short white dress, red hat, thick scarf, light boots. He catches her and holds the final smiling pose from 00:28–00:30, with the palace clearly visible. CAMERA & NEGATIVE RULES Use stable medium-wide/full-body smartphone framing. Keep characters and scenery visible. Simple match cuts between locations; outfit changes only between scenes. Realistic balance, body weight, hand contact and fabric movement. No face changes, wrong outfits, premature catching, floating, fake embraces, extra limbs/fingers, disappearing props, distorted hands, exaggerated acting, excessive slow motion, subtitles, UI or watermarks. GOAL: Seven playful spontaneous travel moments, consistent identities, accurate outfits and locations, and believable physical interactions. Pure photorealistic live action.

Hover to preview · click for sound

Culinary Academy Comedy Chaos

[ENGINE DIRECTIVE] Generate a continuous 30-second hyper-realistic live-action office/academic comedy sequence using Kling 4.0 at 120fps ultra-realism in 16:9 widescreen format. Maintain 100% strict visual consistency with provided asset sheets: "Director Kang-jun" (master chef in navy coat) and "Hye-soo" (clumsy student in flour-dusted apron). Setting strictly follows "High-End Culinary Art Academy Lab". Style: High-tension K-drama comedy, sharp whip-pans, snap-zooms, exaggerated physical freezes, dramatic shadow drops, synced comedic audio hits, dynamic subtitle overlays. [AUDIO DIRECTIVE] MUSIC: High-energy K-drama comedic track featuring rapid orchestral pizzicato strings, sudden brass drops, and frantic pots-and-pans percussion syncs. SOUND DESIGN: Loud metallic pots clanking, high-pitched sneaker slide screech, chef's clipboard SLAP, dramatic gong hit, liquid splash, sharp whip-pan WHOOSH, comedic record scratch. [TIMELINE BREAKDOWN] 00:00 - 00:03 | EXTREME CLOSE-UP HOOK & FLYING SPICES (SCROLL-STOPPING HOOK) VISUAL: Extreme macro snap-zoom onto Director Kang-jun’s twitching eye and twitching jaw. Sudden whip-pan 180 degrees to Hye-soo slipping on a stray piece of butter on the floor in slow motion. ACTION: Kang-jun stands rigidly with his clipboard, inspecting. Hye-soo trips over her own feet; her tray of spices and a squeeze bottle of dark reduction go flying into the air in hyper-slow motion toward an expensive, unfinished plating dish. DIALOGUE (HYE-SOO): Ah! Director!! Watch out!! SOUND: Comedic high-pitched screech, slow-motion bass drop, frantic brass swell, metallic clang. 00:03 - 00:07 | SLIDING CATCH & CRASH ON MARBLE VISUAL: Fast tracking camera following Hye-soo sliding across the stainless-steel prep table. ACTION: Hye-soo executes a desperate, clumsy slide across the counter, catching the falling sauce bottle mid-air, but landing face-first on a pastry board, sending a huge cloud of flour into the air. A spray of dark reduction splatters directly onto the center of Kang-jun’s pristine clipboard and sleeve. DIALOGUE (KANG-JUN): What... are you doing in my exam room? SOUND: Sliding THUD, loud flour POOF, sharp record scratch, dead silence beat. 00:07 - 00:12 | SNAP-ZOOM REACTION & SHADOW DROP VISUAL: Rapid three-stage snap-zoom on Kang-jun’s deadpan, unblinking face under a sudden dramatic shadow drop as he looks down at his splattered sleeve. ACTION: Kang-jun slowly adjusts his glasses with one finger, maintaining total silence. He reaches down, dips his finger into the chaotic splash of sauce that accidentally landed on his tasting spoon, and slowly tastes it. His eyes widen by a millimeter—the mistake created a spectacular, unplanned savory-sweet infusion. DIALOGUE (HYE-SOO - WHISPERING): Please, just don't fail me... SOUND: Sharp camera shutter CLICK-CLICK-CLICK, dramatic choir chant hit, sizzle of a pan. 00:12 - 00:18 | THE WHIP-PAN INTERROGATION VISUAL: Whip-pan camera movement back and forth between Kang-jun leaning down over the table and Hye-soo lying flat on her belly, clutching the squeeze bottle. ACTION: Kang-jun slams his hands on the counter, leaning inches from her flour-sputtered face. His expression is terrifyingly intense, but his eyes are fixed purely on the sauce. DIALOGUE (KANG-JUN): This sauce ratio... how did you think of this? DIALOGUE (HYE-SOO): Huh? It just... mixed when I fell... SOUND: Heavy counter SLAM, low bass rumble, high pizzicato tension strings. 00:18 - 00:24 | EXAGGERATED FREEZE & CLASS STUN VISUAL: Dynamic low-angle wide shot. High-contrast lighting flashes as the entire background student team in the kitchen freezes in over-the-top shock. ACTION: Kang-jun suddenly lifts the spoon and his clipboard into the air, turning sharply toward the evaluation board with an epic, serious posture, pointing dramatically at Hye-soo. DIALOGUE (KANG-JUN): She’s a genius. This student gets the highest score in history! SOUND: Triumphant K-drama orchestra hit, audience gasps, thunderous comedic sting. 00:24 - 00:30 | ABSURD RESOLUTION & CHAOTIC FINISH VISUAL: Tracking hero shot moving backward as Kang-jun vigorously shakes Hye-soo's hand while she is still half-sprawled on the marble pastry table. ACTION: Kang-jun aggressively shakes her flour-coated hand with a deadpan, intense expression. Hye-soo, still on her belly holding the bottle, nods along in complete confusion as a rain of flour dust falls around them. DIALOGUE (KANG-JUN): Starting tomorrow, you are my head research assistant. DIALOGUE (HYE-SOO): What?! I just got here yesterday!! SOUND: High-energy comedic K-drama theme song playout, party horn sound, fast fade to black.

Hover to preview · click for sound

Golden Hour Ski Film Edit

A 30-second, seven-shot ski film edit of Kai <Character reference> skiing one steep, untracked mountain face at golden hour, cut together from several real cameras: a drone, a helmet-mounted action camera, an action camera on a pole, a camera planted in the snow, a chase skier and a long lens across the valley. Photoreal action-sports look, real speed, hard cuts on the action, no slow motion. Kai: athletic man in his 20s, red insulated ski jacket, black ski pants, black gloves, white helmet, mirrored orange goggles, black poles, twin-tip skis. Same outfit and gear in every shot, exactly as in the character sheet. Same mountain face, same powder, and the low sun always on his left as he skis down, throwing long blue shadows across the snow. Shot 1 (0-4s), drone: high aerial view of a sunlit ridge and the whole golden face below it. Kai, small in frame, drops in off the ridge and carves his first big turn, a plume of powder behind him. The drone pushes forward and tilts down to follow. Hard cut. Shot 2 (4-8s), helmet cam: first-person view from an action camera on his helmet, ultra-wide with curved edges. His ski tips and pole baskets at the bottom of the frame, snow rushing underneath, the horizon tilting hard left and right with each carve, powder spraying across the lens. Hard cut. Shot 3 (8-12s), pole cam: an action camera on a short pole held out in his left hand, ultra-wide, pointing back at him. His goggles and helmet fill one side of the frame, his body leaning into a turn, the mountain and sky swinging behind him, snow flecks stuck to the lens. He lets out a wordless shout into the wind. Hard cut. Shot 4 (12-17s), planted cam: a camera set low in the snow just below a wind-lip, pointing up at the sky. Kai launches off the lip and flies straight over the lens, skis crossing the low sun, snow raining down onto the camera, then he lands out of frame behind it. Hard cut. Shot 5 (17-21s), chase skier: a camera on a handheld stabiliser, skiing a few metres behind him at the same speed. He drops into a tuck and straight-lines the fall line, spray from his tails hitting the lens, the slope streaking past. Hard cut. Shot 6 (21-25s), long lens: a telephoto shot from across the valley, slightly shaky, compressing the slope. Kai lays a huge carve and throws a wall of powder backlit gold by the sun. Hard cut. Shot 7 (25-30s), low static camera at the bottom of the run: he skis straight at the lens and skids a hard hockey stop right in front of it, a sheet of snow blasting the camera. As it clears, he pushes his goggles up onto his helmet, breathing hard and laughing. Look: every action-camera shot ultra-wide and crisp with curved edges, small lens flares and snow on the lens; the drone and long-lens shots clean and natural. Warm low sun, long blue shadows, glittering ice crystals in the spray, natural colour, real motion blur. Sound, in stereo, loud and up front like real action-camera audio, never quiet: constant wind roaring and buffeting the camera mics, gusting harder as he picks up speed; the crunch of snow under his skis as he drops in; friends whooping and yelling "Go!" from the ridge at the start; the hiss, scrape and chatter of his edges biting the snow on every carve; poles planting with a crisp crunch; powder spraying off his tails with a soft roar; his hard breathing, and a grunt on the landing; a loud whoosh and the clack of skis as he flies over the planted camera, then the thud of the landing; distant ravens calling across the valley; the ripping sheet of snow blasting the lens on the hockey stop, then his panting laugh and his friends cheering from up the slope. A driving instrumental track with punchy drums sits low underneath, cut on the beat, never covering the natural sound. No subtitles.

Hover to preview · click for sound

Supernatural Murder Mystery Trailer

30-Second Cinematic Thriller Trailer Create a dark, intense, photorealistic 30-second theatrical movie trailer for a supernatural psychological murder mystery. Serious adult thriller tone, realistic actors, cinematic 35mm photography, deep shadows, cold desaturated nighttime colors contrasted with warm candlelit church interiors, rain, subtle film grain, shallow depth of field, dramatic camera movement, rapid escalating editing. The trailer begins slowly and mysteriously, then builds into a frantic final 10 seconds. Maintain the same main male character and the same sister throughout every appearance.0–4 seconds: BLACK SCREEN. Complete silence except for the faint sound of a woman breathing nervously.A sudden flash reveals a terrified young woman running alone through a dark rain-soaked alley at night. Handheld camera races backward directly in front of her as she runs toward camera, repeatedly looking behind her.SMASH CUT to an extreme close-up of the MAIN CHARACTER asleep in bed, sweating and twitching violently during a nightmare.Inside the nightmare, a deep distorted male whisper says:“Three…”A huge blood-red number 3 flashes briefly against black.4–8 seconds: Rapid nightmare fragments: the frightened woman reaches a dead end. A shadowy male figure approaches behind her, but his face is NEVER visible. Extreme close-up of the killer's hand reveals a distinctive ornate silver ring.The whisper returns:“Two…”Blood-red 2.The main character suddenly wakes with a violent gasp.Hard cut to morning television news footage showing the SAME WOMAN'S photograph.Female television reporter:“Police discovered the body early this morning…”The man's face drains of color.8–12 seconds: Police interrogation room. Harsh overhead fluorescent light. The main character sits across from two suspicious detectives.Detective, coldly:“You described the murder before we found the body.”Close-up on the main character.“Because I saw it happen.”Detective:“In a dream?”Silence. Fast close-up of the detective staring at him like he no longer believes a word he says.12–16 seconds: Warm Sunday sunlight streams through stained-glass windows inside a beautiful old church. The man's younger sister walks beside him down the aisle. She gently takes his arm.SISTER:“Maybe coming back here is exactly what you need.”He gives her a conflicted look.A friendly pastor approaches and shakes the sister's hand.CLOSE-UP:The pastor's hand.The silver ring.Everything suddenly goes silent.Extreme close-up of the main character's eyes widening.FLASH CUT back to the nightmare—the killer's hand wearing THE SAME RING.16–20 seconds: Music suddenly becomes a pounding cinematic pulse.Rapid investigative montage:The main character frantically covers his apartment wall with photographs of murdered women, newspaper articles, church bulletins and handwritten notes.He searches through old records late at night.A newspaper headline:SECOND WOMAN FOUND DEADA woman nervously enters a church confession booth.A shadow moves behind the confessional screen.A detective slams a crime-scene photograph onto a table.Police lights explode across the main character's face as he runs through a wet nighttime street.Detective shouting:“STOP!”20–24 seconds: Suddenly—Silence.The main character is asleep again.His nightmare begins.His sister stands completely alone beneath a flickering streetlight.She slowly turns toward him.Terrified.A dark figure appears several yards behind her.His sister whispers:“Michael?”The distorted voice begins:“Three…”FLASH — red 3.His sister running.“Two…”FLASH — red 2.Extreme close-up of Michael screaming awake:“NO!”24–28 seconds: The trailer erupts into extremely fast cinematic cuts synchronized with thunderous impacts:Michael sprinting through pouring rain.His sister pounding desperately against a locked door.Police racing toward the church.Michael smashing through a church doorway.A pastor turning suddenly in a dark hallway.A flashlight sweeping across rows of empty pews.Michael struggling with an unseen person.His sister screaming.The ornate silver ring hitting the floor and spinning across stone.A handgun trembling in Michael's hand. Church bells violently ringing.Michael enters a dark confession booth and sits down.Everything becomes silent.From behind the wooden screen:PRIEST: “What would you like to confess?”Michael stares straight ahead.Calm now.MICHAEL: “I'm going to kill someone tonight.”28–30 seconds: BLACK.A woman's frightened whisper:“One…”A gigantic blood-red 1 flashes onto screen with a violent bass hit.TITLE:3 2 1 — RED LIGHTThe numbers briefly flicker like a countdown. Final extremely fast shot: Michael standing alone inside the dark church as red-and-blue police lights pulse through the stained-glass windows.CUT TO BLACK.TRAILER STYLE: Psychological thriller mixed with supernatural mystery and serial-killer suspense. The dreams should feel surreal and fragmented but the real world must remain grounded and photorealistic. Avoid excessive gore; create fear through shadows, implication, facial expressions, sound, and rapid glimpses. Never clearly reveal the killer's identity during the trailer—the silver ring is the major visual clue. The pastor should initially appear completely trustworthy and ordinary. Michael's emotional progression should move from confusion → fear → obsession → desperation → grim determination. The final confession-booth line should be the trailer's biggest dramatic moment.

Knight and Petrified Dragon

Knight and Petrified Dragon

[Shot 1 - 4.0s] Low tracking shot pushing forward behind the knight. Foreground: petrified kneeling statues half-buried in ash. Subject: The knight walks forward through deep gray ash, battered cape swaying in cold drafts. Background: the colossal slumped silhouette of the four-winged dragon. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera tracks forward at hip level with subtle handheld micro-shake. Cut on the knight halting. [Shot 2 - 3.0s] Low-angle close-up. Foreground: cracked flagstones buried in fine ash. Subject: The dragon’s stony jaw pinned to the floor by a thick black-iron spear; the knight enters the frame and grips the rusted shaft with both gauntlets. Background: towering petrified scales looming in darkness. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera pushes in tight on the spearhead. Cut on gauntlets locking onto the iron. [Shot 3 - 2.5s] Medium close-up, Dutch angle. Foreground: the knight's pitted visor. Subject: The knight wrenches the heavy spear upward; bright sparks shower outward as the iron blade grinds free of stone and bone, spraying ancient ash. Background: the dragon's flared nostril twitching. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera jolts backward in sudden recoil. Cut on the spear wrenching free. [Shot 4 - 4.5s] Medium profile tracking upward. Foreground: the heavy spear clattering into the ash. Subject: The archdragon’s golden slit eye snaps open; its jaw unhinges with a grinding crack, venting plumes of pale soul-smoke as its neck arches up. Background: crumbling vaulted arches trembling in shadow. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera pans upward following the rising horned skull. Cut on the jaw snapping fully open. [Shot 5 - 7.0s] Wide tracking arc moving right. Foreground: stone statues toppling and shattering on the floor. Subject: The dragon heaves onto its front talons; its four ragged stone wings snap open, smashing flanking gothic pillars into flying rubble. Background: the collapsing dome shaking violently. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera sweeps laterally through falling ash clouds. Cut on a massive wingbeat shockwave. [Shot 6 - 9.0s] Ultra-wide hero shot (The Withheld Reveal). Foreground: the knight stands minute at screen center, drawing a greatsword in a two-handed guard. Subject: The ancient dragon rears to full height; revealed chained to its neck collar is the veiled, petrified corpse of a queen, whose skeletal hand sparks with pale blue soul-fire, igniting the dragon's throat into a blinding white-blue firestorm that illuminates the entire mausoleum. Background: the ruined dome raining ash and moonlight. Pale, cold moonlight cuts downward through a fractured dome fissure on the right, casting sharp chiaroscuro edges across ash-dusted stone and dark rusted iron. The camera cranes slowly backward and upward, holding the iconic boss title-card confrontation.

Hover to preview · click for sound

Retro Robot Family Suspense

[Shot1, 3s ] 水色のワンピースを着た女の子が、ロビンの隣にしゃがんで両手で優しく肩を支えている。心配そうに父親を見上げて話す。カメラはリリーの不安げな顔へゆっくりドリーイン。ロビンは静止したまま、光る目がわずかに点滅する。静かなサスペンス調のBGM。セリフ:「Daddy, something's wrong with Robin!」 [Shot2, 3s ] 黄色い花柄のワンピースを着た母親が、心配そうにロビンを見下ろしながら身を乗り出す。突然何かを思い出したように眉を上げ、考え込むように話す。カメラは母親の顔のクローズアップへゆっくりプッシュイン。背景のキッチンには暖かい照明が灯っている。不穏なストリングスが静かに流れる。セリフ:「He's been acting strange lately.」 [Shot3, 3s ] ベージュのスーツを着た父親が突然目を大きく見開く。慌てて上半身を起こし、信じられないという表情でロビンを凝視する。カメラは父親の驚いた顔へ素早くドリーインし、タイトなクローズアップ。表情はますます緊迫していく。鋭いオーケストラの効果音が鳴り、その後一瞬の静寂が訪れる。セリフ:「Oh, no! It can't be!」 [Shot4, 3s ] 小さく光沢のあるオレンジ色のレトロロボット、ロビンが突然起動する。丸い目が明るく輝く。頭部を2回ぎこちなく機械的に動かし、非常に短い腕がわずかに震える。カメラはローアングルから光る目へゆっくりドリーイン。ロビンの元のデザイン、体の比率、アンテナ、腹部前面のパネル、極端に短い脚を完全に維持する。柔らかな電子音と陽気な完了チャイムが響く。セリフ:「Don't worry! Kling four point oh update complete!」 [Shot5, 3s ] 家族全員が突然喜びの笑い声を上げる。父親と母親が勢いよくハイタッチし、水色のワンピースを着たリリーとピンクのワンピースを着た妹も楽しそうにハイタッチする。ロビンは非常に短い機械の両腕を上げて喜び、丸い目が明るく輝く。カメラはワイドショットから緩やかにクレーンアップしながら後退。美しいレトロフューチャーなキッチンと夜の街並みを映し出す。1950年代の陽気なスウィングジャズが始まる。セリフ:「Hooray!」

Hover to preview · click for sound

Potter Shaping a Bowl

Cut between shots of the same scene: a wide as the camera circles slowly around her at the wheel; a close-up of her wet hands pulling the wall of the bowl up; a close-up of her face as she concentrates; back to the wide as she slows the wheel and sits back to look at it. Same woman, same studio, same light in every shot. Natural studio sound, the hum of the wheel. Photoreal.

Hover to preview · click for sound

Coastal Explosion Destruction Sequence

Use the start image as the exact opening frame and preserve the identity of the coastal location: the rocky shoreline, small beach, calm bay, dirt path on the right, wooden posts in the foreground water, distant pier, and the large rock wall across the cove. Create a cinematic multi-shot environmental destruction sequence. The scene begins calm and still at dusk. A sudden bright impact appears in the distance near the far side of the cove. A powerful explosion erupts along the rocky coastline, throwing debris, dust, smoke and water high into the air. The shockwave spreads across the bay. The water reacts violently, waves push toward the shoreline, loose debris falls into the sea and smoke expands across the background. A second larger explosion follows closer to the shoreline, briefly illuminating the rock wall, beach and water. Structure the sequence as: 1. Wide establishing shot of the calm coast. 2. Distant impact and first explosion. 3. Wider shot showing the shockwave affecting the environment. 4. Closer dramatic shot with debris, smoke, water spray and falling rock fragments. 5. Final wide shot showing the same coastline covered in smoke and aftermath. Preserve the spatial layout of the bay, beach, dirt path, pier, shoreline and rock wall across every shot. The explosions must interact physically with the environment through realistic debris trajectories, water displacement, smoke expansion and changing light. Keep the scene readable rather than abstract or chaotic. Sound: distant impact, deep explosions, rock debris, water displacement, wind and waves. No background music. No captions.

Hover to preview · click for sound

Third-Person Gameplay Sequence

Use the supplied game images to establish the heroine, environment, and interface styling. Create a 30-second third-person action-adventure gameplay sequence. Keep the woman’s light-brown jacket, green cargo trousers, backpack, and curved blade consistent, with the camera following behind her at playable character height.

by fal.aiSource
Hover to preview · click for sound

Ship Piano Multishot Keyframes

Use the supplied keyframes to guide the wide ship view, piano inserts, and pianist close-up. Keep the ship, instrument, and performer consistent across the cuts.

by fal.aiSource

Sunset Residential Lane Goodbye

Create one complete 30-second, ultra-realistic Japanese romantic drama. Follow the five timed scenes below within a single finished film. The emotion comes from a woman who tries to say goodbye, then finds the courage to admit she wants to meet again. MAIN SUBJECT Ella is the sole visible character: a beautiful adult Japanese woman with straight dark hair and bangs, natural skin texture, subtle everyday makeup, and believable expressions. Her wardrobe for this film is a white blouse, a pale blue cardigan, simple neutral trousers, and ordinary low shoes. Keep the same facial identity, facial proportions, apparent adult age, eyes, skin tone, hair, clothes, and body proportions throughout every shot. Use small changes around her eyes and mouth to convey emotion; avoid exaggerated crying. LOCATION A quiet fork in a Japanese residential lane at sunset. Modest houses, concrete-block garden walls, plants at entrances, overhead utility wires, and a distant railway crossing establish a lived-in Japanese neighborhood. Keep the street layout and sunset direction coherent. No landmark, crowds, prominent brands, or readable advertisements. CAMERA / VISUAL STYLE Naturalistic live-action romance, intimate eye-level framing, soft warm sunset, realistic depth of field, restrained contrast. The camera occupies the emotional position of the unseen person saying goodbye to her. No visible male actor. Begin in a medium shot, hold the physical distance as she leaves, and let her approach create the final close-up. Subtle human handheld movement, clean motivated edits, no orbiting, drone shots, artificial beauty filters, or slow motion. This is a dramatic scene, not a fashion montage. 00:00–00:05 — A POLITE GOODBYE Ella stands at the fork facing the unseen person near the camera. She makes a slightly forced but gentle smile and gives one small wave. In natural Japanese, with synchronized lips, she says: 「じゃあね。……元気でね」. She lowers her hand and begins to turn away. 00:05–00:11 — THE SMILE DISAPPEARS She walks several normal steps away. The camera stays at the original position and does not chase her. A restrained three-quarter view reveals that her social smile has faded. She looks down very briefly, trying to maintain composure. Only ordinary footsteps and neighborhood ambience. 00:11–00:17 — SHE CHANGES HER MIND Her feet stop naturally. A small breath moves her shoulders. She hesitates, turns her head and then her body back toward the camera, and softly calls: 「……ねえ」. Leave silence around the word. No tears falling, no melodramatic gesture. 00:17–00:24 — RETURNING She walks back toward the same camera along the same path, stopping at a comfortable conversational distance. Her expression combines shyness and resolve. Looking toward the unseen person's eyes beside the lens, she says: 「さっきの、言い直してもいい?」. Do not rush her approach or dialogue. 00:24–00:30 — THE HONEST WORDS An intimate close-up as she stays still. Her eyes become slightly moist, and a vulnerable smile appears. At approximately 00:25–00:27, she says quietly: 「また、会いたい」. Hold her face and the silence through the final seconds as she waits for a reply. End on this living expression without a title card or a fade to black. PHYSICAL REALISM / CONTINUITY Realistic walking speed, grounded feet, anatomically correct hands, consistent spatial direction, and natural breathing. No extra fingers, duplicated people, facial morphing, sliding feet, teleportation, or changing clothing. Her return follows the path she just walked. One person and one continuous emotional event; no unrelated insert shots. AUDIO Ella speaks only the four specified Japanese lines, in one consistent natural adult female voice. Accurate Japanese lip synchronization, quiet breaths, shoes on pavement, distant traffic, and a faint railway crossing. No male reply, narrator, music, subtitles, or added text. Preserve roughly two seconds of emotional breathing space after the final line. FINAL FEEL A small, believable romantic turning point. Her beauty is visible in her face and presence, while her uncertainty makes the scene human. The final result should feel like thirty seconds extracted from a convincing Japanese live-action drama.

Watch on X

Train Window Farewell

Two shots. Shot 1 (0-5s): close-up through a train window, a woman in a camel coat presses her palm to the glass, eyes wet. Shot 2 (5-10s): reverse angle from inside the carriage, a man on the snowy platform walks alongside as the train starts moving, then stops and raises his hand. Sound: a train horn, wheels starting to turn, a muffled station announcement.

Watch on the original post
by kling4.videoSource

Romanian Folklore Dark Comedy

KLING 4.0 — 20-SECOND ROMANIAN FOLKLORE DARK COMEDY, 16:9, FIVE SHOTS. VISUAL STYLE: Authentic 1977 Romanian-Czechoslovak live-action fairy-tale movie, imaginative handmade costumes, grotesque practical prosthetic makeup, soft 35mm film grain, faded analog color, atmospheric Carpathian woodland. CHARACTERS: BABA CLOANȚA: An ancient, hunched Romanian witch with a gigantic hooked nose, warts, enormous pointed ears, almost toothless mouth and glittering mischievous eyes. Multiple layers of ragged wool, old beads, dried herbs and tiny copper bells. Irritable, cunning, unexpectedly affectionate. THE PRINCE: A handsome but painfully nervous young man, age 25, wearing a faded blue medieval doublet, a velvet cap and an oversized sword. Frightened of practically everything. SETTING: Outside a ramshackle wooden witch's cottage deep in a misty Romanian forest. Twisted roots, mossy stones, bundles of dried herbs, a rough wooden counter. DIALOGUE: Romanian only, naturally spoken with authentic Romanian pronunciation. 0:00–0:03 — MEDIUM CLOSE-UP: The prince rushes toward Baba Cloanța, breathless. He cries, "Mă urmărește un blestem!" His eyes dart around the trees. 0:03–0:07 — CLOSE-UP: The witch grins and holds up a tiny antique brass bell on a red string. She says mysteriously, "Sună când ești în primejdie." 0:07–0:11 — DETAIL SHOT: The prince takes the bell. Instantly it begins ringing violently in his hand. He looks at it, terrified. 0:11–0:15 — MEDIUM TWO-SHOT: The prince frantically shakes the bell and asks, "E stricat?" Baba Cloanța tilts her enormous head and studies him. 0:15–0:20 — EXTREME CLOSE-UP: The witch leans toward the prince until her grotesque face nearly fills the frame. She smiles with all three remaining teeth and whispers tenderly, "Nu, dragule. Sunt chiar aici." The prince slowly realizes what she means. His eyes widen. The bell rings faster. Cut to black at 20 seconds. PERFORMANCE: Distinctive practical-makeup facial expressions, convincing frightened reactions, excellent timing. The witch is genuinely amused, never cartoonishly frantic. CAMERA: Natural 1970s film coverage, restrained movements, expressive close-ups, clear continuity. AUDIO: Romanian dialogue, creaking woodland branches, distant crows, wind, footsteps, increasingly frantic small brass bell. No music, no narrator, no subtitles.

Watch on X

How to write great Kling 4.0 prompts

  1. 1Call every input by its @name. On the API, references are numbered by their position in the request: @image_1 and @image_2 for image_urls, @video_1 for video_urls, @first_frame, @keyframe_1 and @last_frame for frames, and @Name for a subject in elements or a voice in voices. Every mention has to match an input in the same request, or the request is rejected before anything is charged.
  2. 2Turn "the supplied image" into a mention. Many prompts from fal.ai's guide say things like "Use Video 1 for the camera movement and the supplied images for the scene", and they only work with those files attached. On the API, pass the video in video_urls and the images in image_urls, then rewrite the references as @video_1 and @image_1 in the same order.
  3. 3Lock a character as a named subject. Instead of describing the same face in every sentence, send it once in elements: 2 to 4 views with the main one first, or one 3 to 8 second single-shot clip of the person. Then write @Mia every time she appears. It does at the request level what long prompts here try to do with "the same person throughout".
  4. 4Script the sound, since it is always on. Sound cannot be turned off on Kling 4.0, so the prompts that land write it out: the kling4.video examples end with a "Sound:" line, and dialogue is quoted with the language named, as in says in English: "Sorry, was that oat milk or almond?" To give a character a set voice, bind a clip with voices on Kling 4.0 or elements[].voice_url on Flash. With reference videos, keep_original_sound keeps their own audio.
  5. 5Write timed shots as text, and make them add up. A common pattern here is "Shot 1 (0-5s): … Shot 2 (5-10s): …" inside one prompt. The API has no separate storyboard field, so those shots travel as plain prompt text: give each one a single action, and make the time ranges add up to the duration you request, 3 to 30 seconds (up to 20 on Flash).
  6. 6Move format lines into the request. Long X prompts often open with "Format: 16:9, 30 seconds, 24fps, 1080p", and a few ask for 4K. On the API those are settings: duration, resolution (720p or 1080p; 4K is not available) and aspect_ratio (auto, 16:9, 9:16, 1:1 or 21:9). Set them in the request so the text and the settings do not disagree.
  7. 7Keyframes for a planned path, references for a look. Keyframes fix where the shot goes: a first frame, up to 8 middle frames and a last frame, with the total capped by the duration, from 4 frames at 3 seconds to 10 from 9 seconds up. References fix what things look like. You cannot send both in one request, and Flash takes a first frame only.

Prompt formula

Format as parameters (duration, resolution, aspect_ratio) + subject, or @Name for a named subject + one action per timed shot + camera move + light and setting + an @mention for every reference + a "Sound:" line with quoted dialogue and its language

Example from the collection, quoted as posted (@Alin_Reaper05 on 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.

Frequently asked questions

What is Kling 4.0, and how is Kling 4.0 Flash different?

Kling 4.0 is the newest video model in Kuaishou's Kling family; it generates sound with every clip and runs 3 to 30 seconds at 720p or 1080p. Kling 4.0 Flash is its lighter tier on the same endpoint and request shape, but with 720p only, 3 to 20 seconds, prompts up to 8,000 characters, a first frame as the only keyframe and no standalone voices list. Choose Kling 4.0 when you need 1080p, more than 20 seconds, middle or last keyframes, or standalone voices.

Who wrote these prompts?

Other people did: creators on X and Civitai, the kling4.video prompt library and fal.ai's Kling 4 prompting guide, and we did not write or edit any of them. Each card quotes the prompt in full, word for word, names the author and links the original. The prompts and clips belong to their creators; our own notes and data are licensed CC BY 4.0. To correct a credit or have an entry removed, email us or open an issue on the GitHub repository.

Where can I run these prompts, and what does it cost?

On apimodels.app, as kling-4.0 or kling-4.0-flash on POST /v1/video/generations, billed per second. Kling 4.0 without video input is $0.157/s at 720p and $0.235/s at 1080p: 5 seconds at 720p is $0.785, 10 seconds at 1080p $2.35 and 30 seconds at 720p $4.71. Flash is $0.094/s: 5 seconds $0.47, 10 seconds $0.94. With a reference video the rate is $0.094/s or $0.141/s on Kling 4.0 and $0.063/s on Flash, per billed second: the output length × 2 when the input video is no longer than the output, or input + output when it is longer, so 10 seconds at 720p with a 5-second reference video is $1.88 on Kling 4.0 and $1.26 on Flash. Reference images and keyframes are not billed separately, and failed generations are not charged.

Why does a prompt from this page fail on the API?

Requests outside the limits get a 400 before a task is created, and nothing is charged. Causes include an @mention with no matching input; a prompt over 10,000 characters (8,000 on Flash, which the longest prompt here exceeds); a length outside 3 to 30 seconds (3 to 20 on Flash); 4K, which is not available; frames mixed with references; reference images under 300 px on a side or outside a 0.4 to 2.5 aspect ratio; and a reference video whose length cannot be read. The API also has no negative-prompt field, no multi-shot storyboard mode and no video extension, so prompts written around those features need to be rewritten as one plain prompt.

Do I need the images or videos the creator used?

Only for prompts that point at them. Text-only prompts run as they are. Prompts that say "the supplied image" or "Video 1", or that name a character from a reference, need your own files in image_urls, video_urls or elements, each called with @. Only upload faces and voices you have the rights to use.

How long can a Kling 4.0 prompt be?

Up to 10,000 characters on Kling 4.0 and 8,000 on Flash. The prompts here run from one sentence to almost 10,000 characters, and the long ones tend to spend their words on a fixed character description, a timed shot list and a sound list rather than on adjectives. If a long prompt describes a reference you are not sending, cut that part.

Can I use these prompts and videos commercially?

The prompts and clips belong to their creators, so check the original post before reusing a prompt commercially. For clips you generate yourself, follow the terms of the service you generate them with, and only use faces and voices you have the rights to.

Run these prompts on apimodels.app

Kling 4.0 is live here alongside 180+ other image, video, audio and language models — one API, one key, pay as you go, with $0.10 of trial credit on sign-up.

This library is open source on GitHub

The whole collection is on GitHub as Markdown and JSON: every prompt with its author's credit and a link to the original on X, kling4.video, fal.ai or Civitai. Prompts are quoted in full, word for word, and belong to their creators. The clips stay with their creators and are linked, not re-hosted. Our own notes and data are licensed CC BY 4.0. Open an issue to correct a credit or have a prompt removed.

View on GitHub

Other prompt libraries