Kling 4.0 Omni: show it what you mean

Up to 15 in one generation: up to 10 images (30 MB each), up to 5 videos (3-30 seconds each, 30 seconds combined), up to 3 voice profiles, and up to 7 elements including 3 taken from video.

References are numbered in upload order - @Image1, @Video1 - and you assign roles in the prompt, like "@Image1 defines the scene; preserve its layout and lighting" or "@Video1: reference the action pacing." Naming roles makes results far more predictable than letting the model guess.

Character performance, actions, camera movement, and narrative pacing. White-model and depth-map videos are supported too, for precise control over motion and timing.

Yes. Feed up to 5 clips - 1 primary plus references, each 3 to 30 seconds with a 30-second combined total - and make targeted edits to expressions, movements, camera position, style, or background while the rest stays unchanged. It works best with reference images alongside.

Duration from 3 to 30 seconds in 1-second steps (default 10 seconds), resolution at 720p, 1080p, or 4K, and aspect ratios of 21:9, 16:9, 1:1, 9:16, or auto - auto follows your references.

Kling 4.0 generates from a prompt, an image, or keyframes. Omni works from a stack of references, which makes it the pick for consistent characters, recreating proven concepts, and edits to existing footage.

You can try it in Picsart AI Playground. Access follows your Picsart plan and available credits - check Picsart pricing for current details.

Open it in Picsart AI Playground on the web or desktop app, or build it into your own projects via CLI, MCP, REST API, and SDK - no third-party API keys or separate subscriptions.