Make a scene, not just a moving frame. Write the action, direct up to six shots, add your own visual references and choose whether the moment needs sound.
3–15 seconds 720p to 4K Native audio optional
Kling 3.0 is Kuaishou’s AI video model for turning text or images into short moving scenes. On this site, you can start with a prompt, guide the opening or ending with your own images, and direct the action shot by shot.
These two clips were generated with this site’s Kling 3.0 model. The first is one action with sound; the second uses two timed shots. Play either clip, then inspect the exact prompt beside it.
Watch the camera pull back as the artist turns the cup. Unmute to hear workshop ambience and the final tap. The prompt gives the camera one move and reserves the sound for a concrete action.
A cinematic portrait-format film of an adult ceramic artist in a sunlit workshop. Close-up: clay-covered hands lift a freshly glazed deep-blue cup from a wooden bench. The camera smoothly pulls back to reveal the artist turning the cup toward the window, warm light gliding across the glaze. She gently sets it on a saucer. Natural workshop ambience, a soft ceramic tap, no music, no speech, no subtitles. Detailed materials, believable hand motion, gentle pacing.
Make it yours: Replace the object and hand action first. Keep the camera move and sound cue specific to your new scene.
Watch the cut at two seconds: the brass watering can carries across both shots. The storyboard specifies a wide opening, a close-up ending and exactly how long each beat lasts.
A cinematic two-shot journey through a miniature greenhouse at sunrise. Maintain the same brass watering can and green plants across the shots. Rich glass reflections, natural camera movement, realistic water, no text or logos.
Make it yours: Swap the watering can for one distinctive object, then keep it named in both shots so the sequence has a visual link.
Describe a subject, what changes, where it happens and how the camera moves.
Upload an opening image, with an optional ending image, to guide the start and destination. The output framing may follow your image.
Supply multiple views of a character or object when it must stay recognizable. Add up to three Elements, with two to four images each.
Order up to six shots and assign each a duration. Their seconds must add up to the 3–15 second clip length.
Add dialogue, ambience or a clear sound cue to the prompt, then switch audio on. Check the updated credit quote before generating.
Want a feature video or role-based image references? Those are inputs for Kling 3.0 Omni, a separate model.
A useful short-video prompt tells the model what changes over time. The workshop example works because it names a subject, an action, a camera move and one audible moment, all within five seconds.
Start in the generatorStart with one subject and one achievable change: an artist turns a cup, or water lands on a fern.
Choose a push-in, pull-back or close-up. For a sequence, divide the action into timed shots in Create.
Add your own frames or Elements only if the scene needs them. Pick duration, output and audio; review the updated credit quote before submitting once.
Play the result in Create. If continuity fails, simplify the action or repeat the identifying detail in each storyboard shot.
Start at 720p to test an idea, then increase resolution if your final use needs it. This example uses the site’s current Kling 3.0 credit rules for a five-second clip.
The exact quote updates in Create as you change settings. Credits are used only when you submit a generation.
| Output | Silent | With audio |
|---|---|---|
| 720p | 68 | 101 |
| 1080p | 90 | 135 |
| 4K | 429 | 429 |
Yes. Add a first frame in Create to guide the opening; an optional end frame can guide where the clip finishes. The output framing can follow the source image rather than the aspect ratio selected in the form. You provide the image—our text-only examples above do not use one.
Elements let you provide multiple views of a recurring character or object. This Kling 3.0 workflow accepts up to three Elements, each with two to four images. Use them when identity matters across a scene; they are separate from the first and last frames.
Yes. Storyboard lets you write up to six ordered shots whose durations add up to the clip length. Native audio is optional. Describe the sound you want and check the audio setting and updated quote in Create before submitting.
This site currently offers whole-number durations from 3 to 15 seconds and 720p, 1080p or 4K output for Kling 3.0. Available framing and the final image-guided aspect ratio depend on the input mode.
No. This page uses Kling 3.0. Omni is a separate model with role-based image references and a feature-video input workflow. Its input rules and audio pricing differ, so check the model selected in Create before you generate.
The quote updates with the model, duration, resolution and audio setting in Create. A generation uses credits when submitted; simply reading an example or opening a draft does not. Completed videos appear in the Create conversation and My Assets.
Start with one clear action. Add more control only when your idea needs it.
Create with Kling 3.0