Kuaishou ships Kling 3.0: native 4K 60fps video, 6-shot storyboarding, 5-language lip sync
- Kuaishou Technology released Kling 3.0, a video generation platform built on a unified Multimodal Visual Language (MVL) architecture that integrates video, audio, and image generation into one model and keeps character identity and audio continuity across shots in a single generation pass.
- Output reaches native 4K (3840x2160) at 60fps with 16-bit HDR color and no upscaling, with clips up to 15 seconds long.
- A Multi-Shot AI Director mode generates up to 6 connected camera shots with consistent characters per sequence, alongside Elements character locking from up to 4 reference images and coreference for 3 or more characters in one scene.
- Motion Control AI exposes three control surfaces: Curve Dolly for drawn 3D camera trajectories, Multi-Elements for independent movement paths across 2 to 4 subjects, and adjustable Camera Shake, plus 6-axis camera control with intensity values from -10 to 10 and full API support.
- Native lip sync covers 5 languages: English (American, British, Indian accents), Chinese, Japanese, Korean, and Spanish, using frame-accurate phoneme mapping.