Today, Kuaishou's Kling AI officially enters the 3.0 era. This update is not only a leap in model performance, but also redefines the AI video creation workflow through integrated multimodal input and output, marking the official start of the "everyone can be a director" era.
Video Capabilities: From "Clip Generation" to "Deep Storytelling"
Kling AI 3.0 has achieved a revolutionary evolution in video generation:
World-first Subject Reference: In addition to image-to-video, it now adds multi-image/video subject reference functionality. Creators can precisely lock onto the main character, props, or scene features, solving the common issue of "character drift" in AI video creation.
Intelligent Storyboarding and Ultra-long Generation: Introducing native custom storyboarding capabilities, AI can deeply understand script logic and automatically schedule shot types and camera positions. Video duration supports flexible control between 3-15 seconds, bidding farewell to fragmented stacking and achieving a cinematic narrative in one go.
All-in-one Audio-Visual: Supports multiple languages including Chinese, English, Japanese, Korean, Spanish, as well as dialects like Sichuanese and Cantonese. By extracting more than 3 seconds of voice, it can bind a unique voice for the character, achieving natural performance with synchronized mouth movements, expressions, and voice tones.
Image Creation: 4K Pixel-level Output and Image Group Flow
The image model Omni version is now available, elevating visual quality to cinematic standards:
Ultra-high Clarity: Supports native 2K/4K pixel output without the need for secondary scaling to achieve detailed image details.
Batch Image Group Creation: Supports single/multiple image-to-image group generation, allowing一键 prediction of the plot and generation of a series of coherent scene frames with a unified style, greatly improving professional creation efficiency.
High Flexibility: Supports up to 10 reference images overlaid, combining style transfer and portrait references, allowing users to customize adding, deleting, and modifying elements for high-precision consistency control.
