AB
AiBoss
project

Keling 3.0 Model - A new generation of multimodal AI creation model launched by Kuaishou Keling

Keling AI 3.0 is Kuaishou's new generation multimodal AI creation model, enabling an 'All-in-One' native creation workflow. Model version updates include the introduction of Video 3.0, supporting AI intelligent scene creation, 15-second long video generation, multilingual audio...

What is the Keling 3.0 model?

Keling AI 3.0 is Kuaishou's new generation multimodal AI creation model, realizing an "All-in-One" native creation workflow. Model version updates include: Video 3.0 supports AI intelligent storyboarding, 15-second long video generation, multilingual lip-syncing (including dialects), and image-based video subject references; Video 3.0 Omni enhances all-around references and audio cloning; Image 3.0 supports the fusion and free editing of 10 reference images; and Image 3.0 Omni provides 2K/4K native ultra-high-definition output and batch image creation. The model covers the entire chain from generation to editing, significantly lowering the barrier to professional video production and ushering in an era of AI creation where "everyone can direct."

Main functions of the Keling 3.0 model

  • Video 3.0
    • Intelligent storyboardThe AI-powered intelligent shot composition function can automatically adjust the shot size and camera position, generating multi-shot narrative videos with a cinematic feel with just one click.
    • Main ReferenceThe image-to-video + subject reference function is a world first, supporting multiple images or videos as subject references to firmly lock the visual core and prevent the image from going astray.
    • Multilingual lip-syncingThe all-in-one audio-visual function supports the generation of multiple languages including Chinese, English, Japanese, Korean, Spanish, and dialects such as Sichuanese and Cantonese, with natural and smooth lip movements and expressions that feel completely natural.
    • Text authenticity guaranteedThe original sound-level text function can achieve high-fidelity preservation of fonts, meeting the needs of business scenarios for clear and rigorous information transmission.
    • Extra long durationThe 15-second extended generation function supports flexible duration settings from 3 to 15 seconds, which can accommodate more complex narrative logic to complete the creation of a complete story.
  • Video 3.0 Omni
    • Reference upgradeThe All-in-One Reference 3.0 feature significantly improves subject similarity compared to the O1 version, and is more sensitive and accurate in responding to complex text commands.
    • Timbre CloningThe All-in-One Theme 3.0 supports uploading 3-8 second character videos to extract character features and original voice, perfectly restoring appearance, figure, and charm.
    • Custom storyboardThe Storyboard Narrative 3.0 feature adds native custom storyboard capabilities, and the free duration control is upgraded to 15 seconds, allowing for pixel-level modification.
  • Image 3.0
    • Multiple images lockedThe consistency enhancement feature supports up to 10 reference images, accurately locking in core elements and color tones to achieve style consistency across multiple images.
    • Free editorThe free multi-reference image function integrates style transfer, portrait reference and other multiple image functions, and allows you to directly edit elements, add, delete and modify them without switching functions.
    • Upgraded textureThe comprehensive effect upgrade feature enhances the realism of portraits and optimizes to cinematic color tones, resulting in richer image details and a blockbuster-quality feel.
  • Image 3.0 Omni
    • Light and shadow reconstructionThe in-depth narrative function enables film-level light and shadow reconstruction, clearly deconstructing the audiovisual elements in the prompts to efficiently support professional needs.
    • Batch Image GroupThe image creation function supports creating images in all formats, including single images or multiple images, and allows for batch adjustments and optimizations to create a complete visual system.
    • Native Ultra HDThe native ultra-high definition function supports 2K or 4K pixel-level direct output, eliminating the need for secondary magnification to render delicate and full images with ultimate detail.
    • Realistic AugmentationThe texture enhancement feature comprehensively improves the realism of the image, maintains stable details, and achieves a dual improvement in creative efficiency and work quality.

How to use the Keling 3.0 model

Visit the Keling AI official website; Black Gold members can enjoy advanced experience benefits (web version only), and all features will be available soon.

Application scenarios of the Keling 3.0 model

  • Film and television production fieldIt can quickly produce cinematic short dramas, commercials, and trailers, significantly reducing the cost of professional film and television production.
  • E-commerce marketing fieldBatch production of multilingual product demonstration videos and sales content to improve conversion rates and market reach.
  • Social mediaCreate consistent personal brand content and coherent short videos to enhance account recognition and user engagement.
  • Education and training sectorTo create high-quality multilingual teaching materials and scenario simulation videos, optimize the online learning experience, and reduce course development costs.
  • Game and Animation FieldUsing multi-image reference locking and image-to-video conversion features, we can maintain consistency in character design and quickly convert original artwork into dynamic cutscenes, accelerating the game development process.