Video model · Kuaishou ·
Kling Video v3 is a video model by Kuaishou. In Endless it costs from 27 credits per video and is available on every plan.
Kling Video v3 runs Kuaishou's Kling 3.0 for 3 to 15 second clips from text or an image, with native audio, motion control and a 4K mode.
Kling Video v3 runs Kling 3.0, the generation Kuaishou released in February 2026. It turns a prompt or a starting image into clips of 3 to 15 seconds and follows multi-shot instructions, such as shot-reverse-shot dialogue or cross-cutting, inside one generation. Native audio includes speech in English, Chinese, Japanese, Korean and Spanish, and each character in a scene can speak a different language.
Motion control takes the movement from a reference video and applies it to the character in your image, keeping that character's appearance. It suits dance trends, gestures and performance clips. You can also set an end image so the shot lands where you planned, and switch to the 4K mode when a clip is ready for delivery.
Kling Video O3 is the Omni model of the same 3.0 generation and adds reference-to-video and editing of existing footage. Choose v3 to animate an image or transfer motion, and O3 when you need to change a clip you already have. Both sit in the same workspace, in chat and on the canvas.
Kling 3.0 responds best to prompts written as directions to a scene. Kuaishou's guide and fal's prompting guide both describe a model that understands cinematic language: shot types, camera movement, coverage and pacing. A list of visual traits works less well than a description of what happens, in what order, and how the camera behaves while it happens.
A structure that works for most clips:
[Setting and light]. [Main subject, described once and clearly].
[Action, in the order it happens]. [Camera: framing and movement].
[Sound: ambience, effects, music, dialogue].
Define your main subjects at the start and keep describing them with the same words. If the woman is "the woman in the red scarf" in the first sentence, keep calling her that. Changing the label halfway through makes it harder for the model to keep the character consistent.
Describe motion explicitly, for both the subject and the camera. Words such as tracking, following, orbiting, pushing in, pulling back, tilting up, panning and holding still all work. For long takes, define the relationship between camera and subject: "the camera follows her in a medium shot and stops when she stops." Kuaishou's own long-take examples are written exactly this way.
Clips run from 3 to 15 seconds. Longer durations work when the prompt describes progression: what happens first, what changes in the middle, and where the shot ends. A 15-second prompt with a single action tends to feel slow; give it two or three beats.
Kling 3.0 can cut between shots inside one generation. Label each shot and give it a framing, a subject and a motion: "Shot 1, wide shot of... Shot 2, close-up of...". fal's guide mentions storyboards of up to six shots. Kuaishou notes that when a scene suits a single shot better, the model may adjust. Keep the subject description identical across shots so the same character appears in each one.
Native audio is on by default and can be switched off. When it is on, write the soundscape into the prompt: ambience, physical sounds, music. For dialogue, the guides recommend:
[Lead detective] and [Suspect]. Avoid pronouns.Kuaishou lists speech in Chinese, English, Japanese, Korean and Spanish, with characters able to switch languages in the same scene, and says other languages are translated into English. Accents and dialects can be requested by naming them, for example "with an Indian accent" or "in Cantonese". fal's model card adds that English speech is best written in lowercase, with uppercase kept for acronyms and proper nouns.
With a starting image, the picture is the starting point. Describe how the scene develops from it: what moves, how the camera moves and what changes in the environment. You do not need to describe the image again. Kuaishou says the model keeps text from the source image, such as signs, captions and logos, which helps for product shots. You can also set a last frame so the clip ends where you planned; describe the path between the two pictures.
On-screen text works from a prompt too. Put the exact words in quotes, for example a label reading "NOIR 9".
Attach a motion reference video together with a character image, and the movement from the video is applied to the character in the image. The prompt is optional here and is best used to describe the setting or style. fal's model card asks for a character image with clear body proportions and no occlusion, taking up more than 5% of the frame, and a reference video showing a realistic character with the full body or upper body and head visible.
Character orientation decides whose orientation the result follows. "Image" keeps the orientation of your picture and follows camera movement better, with reference videos up to 10 seconds. "Video" follows the reference clip, suits complex motion and accepts clips up to 30 seconds.
In your workspace, open Video and pick Kling Video v3 in the model picker.
Write a prompt with the subject, camera and motion, or start from an image you upload or generate.
Generate, then refine in chat or on a canvas. Every clip is saved to your workspace assets.
Grouped by use case. Each one opens in Endless with Kling Video v3, ready to edit and generate.
A frosted glass perfume bottle stands on a slab of wet black marble at dusk. The camera orbits it slowly in a close-up as warm side light slides across the gold lettering reading "NOIR 9". A single drop of water runs down the glass. Soft piano in the background.
Macro shot of a mechanical watch on a leather strap resting on folded linen. The camera pushes in toward the second hand as it ticks forward, then tilts up to reveal the brushed steel case catching morning light. Quiet ticking sound, no music.
A runner's white sneaker lands in a shallow puddle on a city street at night, in slow motion. Water sprays outward in a crown and neon reflections ripple across the surface. The camera tracks low alongside the shoe as it lifts off again. Splash and distant traffic.
A ceramic coffee cup on a sunlit café counter. A barista's hand pours steamed milk, drawing a leaf pattern on the surface. The camera stays fixed in a top-down shot, then slowly rises as steam drifts upward. Espresso machine hiss and soft chatter.
Shot 1, wide shot of a red kayak gliding across a still mountain lake at sunrise. Shot 2, close-up of the paddle cutting the water. Shot 3, the kayaker lifts the paddle and smiles at the camera. Birdsong and gentle water sounds.
A small bookshop on a rainy afternoon. An elderly shopkeeper wipes his glasses behind the counter. [Shopkeeper, warm gravelly voice]: "You came back for the poetry one, didn't you?" Immediately, a young woman at the shelf turns around, laughing. [Young woman, bright surprised tone]: "How did you know?" Rain taps on the window.
Inside a moving subway car at night. Two coworkers stand holding the rail. The tall man in a gray coat leans closer. [Tall man, low tired voice]: "Did you send the report?" The woman with a red scarf freezes, eyes wide. [Woman in red scarf, panicked whisper]: "I thought you did." Train rumble and a station announcement in the distance.
A food truck at a summer street fair. A customer in a straw hat orders. [Customer, cheerful tone, in Spanish]: "Dos tacos de pescado, por favor." The cook flips a tortilla on the grill and grins. [Cook, fast friendly voice, in Spanish]: "¡Salen en dos minutos!" Sizzling grill and crowd noise.
A quiet hospital corridor at dawn. A doctor in green scrubs sits beside a nervous father on a bench. The doctor places a hand on his shoulder. [Doctor, calm gentle voice]: "She's awake. You can see her now." The father exhales and covers his face with his hands. [Father, voice cracking]: "Thank you."
A rooftop at sunset over a dense city. Two old friends lean on the railing. [Friend in denim jacket, nostalgic tone, British accent]: "We used to say we'd own this whole skyline." Immediately, the other friend laughs softly. [Friend in black hoodie, dry humorous voice, American accent]: "We can barely afford the roof." Wind and distant sirens.
A sushi counter in Tokyo late at night. The chef sets a plate in front of a tourist and bows slightly. [Chef, polite soft voice, in Japanese]: "どうぞ、召し上がってください。" The tourist picks up the piece with his fingers. [Tourist, enthusiastic tone, in English]: "this is incredible, thank you." Low jazz and clinking ceramic.
Shot 1, wide establishing shot of a lighthouse on a cliff during a storm, waves crashing below. Shot 2, medium shot of the keeper climbing the spiral stairs with a lantern. Shot 3, close-up of his hand switching on the lamp. Shot 4, the beam sweeps across the dark sea. Thunder and howling wind.
Shot 1, low-angle tracking shot behind a cyclist racing down a forest trail. Shot 2, close-up of the front wheel kicking up mud. Shot 3, first-person view over the handlebars as the trail drops. Shot 4, side shot as the cyclist lands a jump and skids to a stop.
Shot 1, close-up of a chef's hands chopping scallions on a wooden board. Shot 2, medium shot as she slides them into a smoking wok. Shot 3, overhead shot of noodles tossed in the air. Shot 4, the finished bowl placed on the pass under warm light. Kitchen clatter and sizzling.
Shot 1, profile shot of a woman reading a letter at a kitchen table, morning light. Shot 2, close-up of the letter trembling in her hand. Shot 3, reverse shot of her face as tears form. Shot 4, wide shot as she stands and walks to the window.
Shot 1, a boxer wraps her hands in a dim locker room. Shot 2, close-up of her eyes in the mirror. Shot 3, tracking shot behind her as she walks down a tunnel toward bright arena lights. Shot 4, wide shot of the crowd rising as she enters. Muffled crowd roar building.
The woman in the portrait slowly turns her head toward the camera and smiles, a light breeze moving strands of her hair. The camera pushes in gently to a close-up. Keep her clothing and the background unchanged.
The camera moves slowly forward along the street in the photo as the shop signs flicker on one by one. Pedestrians begin to walk and a tram passes from left to right. Evening city ambience.
The cat on the windowsill stretches, yawns and looks out the window as a bird lands on the ledge outside. The camera stays static. Soft purring and a single chirp.
The product box in the image opens on its own: the lid lifts, and the headphones inside rise slightly on a soft light. The camera orbits a quarter turn. Keep the printed text on the box sharp and unchanged.
Starting from this mountain landscape, clouds drift quickly across the peaks and shadows move over the valley as the light shifts from morning to afternoon. The camera tilts down slowly toward the river. Wind and distant birds.
The character performs the dance from the reference video on a bright studio floor, keeping the outfit and face from the image. Smooth, confident movement.
The illustrated mascot repeats the waving and bowing gestures from the reference clip, standing in front of a shop entrance. Keep its colors and proportions.
The woman in the photo copies the yoga sequence from the reference video on a beach at sunrise, slow and controlled, with waves behind her.
The knight in the image walks forward and draws his sword with the same timing as the performer in the reference video, in a misty castle courtyard.
The child in the portrait mimics the jumping and clapping from the reference clip in a sunny backyard, laughing as she moves.
One continuous shot with no cuts. The camera follows a waiter in a medium shot as he carries a tray through a crowded restaurant kitchen, pauses when he pauses at the pass, then pushes through swinging doors into the candlelit dining room. Kitchen noise fades into soft jazz.
A single unbroken 15-second shot inside an abandoned greenhouse. The camera glides slowly between overgrown plants as dust floats in shafts of light. At the end, it reveals a girl sitting on a rusted bench, reading. Birds outside, creaking glass.
A slow aerial shot moves over a fishing village at dawn. Boats leave the harbor one by one, leaving white trails on the water. The camera keeps rising until the whole bay is visible. Gulls and engine hum.
A rainy neon alley at night, one continuous handheld shot. A man in a long coat walks away from the camera, stops under a flickering sign, lights a cigarette and turns to look back. Rain on metal, a distant train.
A static wide shot of a wheat field as a summer storm approaches. Dark clouds roll in, the wheat bends in waves, and the first lightning strike flashes on the horizon. Rising wind and low thunder.
by Kuaishou
Generate new clips or edit existing footage with references
From 27 credits per video
by ByteDance
30-second clips in one pass, guided by up to 50 references
From 51 credits per video
by ByteDance
Clips of 4 to 15 seconds steered by image, video and audio references
From 16 credits per video
Kling Video v3 next to other models: price, maker and plan side by side.
Start free, create a project, and bring your brand’s references together. Or book a demo to explore Endless with your team.