Kling 4.0 AI Video Generator

Kling 4.0 supports up to 10 keyframe inputs, giving creators more control over important moments throughout a video. Instead of relying on one starting image and hoping the generation develops in the right direction, you can use keyframes to define how a scene should evolve. Two keyframes can establish the beginning and ending. More keyframes can guide important transitions, camera positions, character actions, or visual changes throughout the story. Kling 4.0 also improves consistency between these keyframes, helping characters, environments, styles, and other visual elements stay more stable across the sequence.

Kling 4.0 can use up to 15 references across a generation. These references can include images, videos, audio, and visual elements. The model supports up to 10 image references and up to 5 video references, with a combined video duration of up to 30 seconds. Each reference can serve a different purpose: Images can help define how a character, product, environment, or visual style should look, videos can guide movement, camera language, pacing, or actions, audio can help maintain voice characteristics, and elements can help preserve important subjects throughout the video. This gives creators more flexibility when a single text prompt is not enough to explain exactly what they want. For branded content, references can help maintain a recognizable product or visual identity. For character-driven content, they can help keep appearance more consistent. For serialized content, multiple references can help maintain continuity across episodes or scenes.

AI video creation does not always need to start from zero. Kling 4.0 includes video editing capabilities that can understand and modify existing footage. It supports up to five video inputs, including one primary video and additional reference videos. The model can use the spatial and temporal information from these clips to better understand what should be changed and how the change should look. Instead of regenerating an entire video because one element is wrong, creators can make more targeted edits to specific areas, characters, actions, or visual details while keeping the rest of the footage intact. This can be useful for: changing a specific visual element, refining a character action, adjusting an effect, modifying part of an advertisement, improving one section of an existing AI video. Kling 4.0 supports individual input videos up to 30 seconds, with a total input duration of up to 30 seconds.

Kling 4.0 supports voice generation across nine languages: Chinese, English, Japanese, Korean, Spanish, Portuguese, German, French, and Hindi. It also supports different dialects and accents, including Cantonese and Sichuanese, as well as American, British, Indian English, and several Chinese regional accents. This can be useful for global marketing, localization, multilingual storytelling, game characters, branded content, and international ecommerce. Kling 4.0 also supports text generation across nine languages. This can help create readable text inside scenes, including product packaging, storefront signs, subtitles, ad copy, and other visual text elements. The model is designed to integrate that text with the perspective and style of the surrounding environment.