Kuaishou Launches Kling AI's “Multi-Image Reference” Feature For AI Video Consistency
KUALA LUMPUR, Jan 24 (Bernama) -- Kuaishou Technology (Kuaishou), a content community and social platform, has announced the global launch of Kling AI’s “multi-image reference” feature, which aims to resolve a key challenge in artificial intelligence (AI) video generation—maintaining visual consistency.
This innovative feature enables AI to analyse and integrate diverse subjects from multiple images and creatively generate composite videos based on users’ text descriptions, empowering creators to produce seamless and professional-quality videos with greater precision, according to Kuaishou in a statement.
The “multi-image reference” feature is available through the Kling AI 1.6 model, which enables users to upload images of the same subject (a person, animal, movie character, or object), and the model will reference these images to generate a video with a consistent style, ensuring the subject remains visually consistent across multiple scenes.
For example, users can upload images of a cat, a jacket, and sunglasses, and the AI will generate a video based on a prompt such as “A humanoid cat in a jacket and sunglasses turns on stage and poses for the camera.” This level of precision helps creators produce professional-quality videos.
Additionally, users can define specific scenes and actions, giving them more creative control. For instance, by uploading reference images of an elderly man, a coffee cup, and a café setting, users can generate a video of a cartoon-style elderly man sipping coffee in a café.
The feature also allows for dynamic interactions between different characters, such as generating a scene with a boy petting a Corgi based on reference images and a text prompt.
Since its initial rollout on June 6, 2024, Kling AI has undergone dozens of extensive iterations and upgrades, including enhanced motion quality with smoother motion and more natural expressions, improved colour accuracy, and lighting dynamics.
Kling AI now boasts a user base of over six million, with more than 65 million videos and 175 million images generated. The platform is also expanding its monetisation efforts, offering membership services for individuals and application programming interface (API) solutions for businesses.
-- BERNAMA