Kling 3.0 brings enhanced realism, 2K/4K output, multimodal video & facial motion control

Kling 3.0 has launched as a major upgrade to its AI text-to-video and text-to-image models, introducing new features across both workflows. The IMAGE 3.0 Omni model unlocks a cinematic narrative experience through still frame storytelling, bolstered by the new Image Series Mode and the ability to output images in direct 2K or 4K Ultra HD resolution. These features deliver enhanced realism and consistency across generated visuals.

Building on these advances, IMAGE 3.0 also adds style transfer, portrait and character reference support, multi-image blending, and local re-editing. Users gain the flexibility to employ multiple reference images without restriction, benefiting from comprehensive upgrades to creative and visual effects.

In the video domain, Kling VIDEO 3.0's unified multimodal architecture now handles text-to-video, image-to-video, reference-to-video, video content addition or removal, as well as video modification and transformation. Following these core changes, users can generate videos up to 15 seconds in length, make use of native audio, and experience improved storyboarding, shot control, and prompt adherence. Overall realism and the quality of character performances are also improved.

Kling VIDEO 2.6 and VIDEO O1 have been transitioned to VIDEO 3.0 and VIDEO 3.0 Omni, respectively, delivering an expanded set of creative tools. Finally, an upgraded Motion Control feature enables binding of facial elements with multiple reference images or video, ensuring facial consistency for complex expressions and occlusion handling.