MediaTek and Kuaishou have collaborated to bring generative video (Gen-AI) capabilities to smartphones powered by Dimensity 9300 and Dimensity 8300 chipsets. Kuaishou’s image-to-video technology leverages the powerful 7th generation NPU in the Dimensity mobile platform to produce dynamic videos based on image prompts directly on the smartphone. This technology continues the video generation innovations that debuted at Mobile World Congress (MWC) 2024.The Kuaishou I2V-Adapter is a lightweight, plug-and-play diffusion module based on Stable Diffusion. This module can convert static images into dynamic videos without altering the original structure and pre-trained parameters of the existing text-to-video generation (T2V) model. Internal benchmarks have demonstrated that the I2V-Adapter produces image-to-video results comparable to Stable Video Diffusion, yet with only 1% of the parameter training and using just 18% of the data — a significant reduction in power use, bandwidth, and storage demands. Additionally, its decoupled design enables the solution to seamlessly work with modules such as DreamBooth, LoRa, and ControlNet, achieving customized and controllable image-to-video generation.
The global market for short videos has experienced rapid growth in recent years, unlocking new avenues of content creation for users and businesses and creating potential for further expansion. The industry's demand for innovative and efficient video generation technology is increasingly urgent. The Kuaishou app now boasts 379.9 million daily active users and over RMB 1 billion in annual e-commerce transactions.
With this in mind, MediaTek and Kuaishou have developed this convenient video generation tool to promote the prosperity and creativity of the short video ecosystem. Both companies will continue to deepen their collaboration on generative AI tools, with a focus on on-device processing.
