ByteDance Develops Real-Time Spatial Video AI Model, Zhang Yiming Oversees Challenge to Meta and Alphabet
ByteDance is developing an AI model for real-time spatial video generation to compete with Meta and Alphabet in robotics and autonomous systems. Founder Zhang Yiming is personally overseeing the project, coordinating resources across business units. Built on ByteDance's Seedance model, it enables interactive virtual worlds for live streaming, short dramas, and games, generating on-demand video at 20 frames per second with about 0.05 seconds latency, potentially lowering VR hardware costs.
ByteDance is preparing to launch an AI model for real-time spatial video generation, competing with Meta and Alphabet in the fields of robotics and autonomous systems. Founder Zhang Yiming is personally supervising the model's development and coordinating related work across the company's business units, allocating AI resources and computing power to the project. The model is built on ByteDance's existing film-grade video generation AI model, Seedance, and allows users to create interactive virtual worlds for live streaming, short dramas, and games.
The model is designed to generate virtual worlds that respond to voice or actions from users of Pico headsets. It can generate on-demand video at a rate of 20 frames per second, with a latency of approximately 0.05 seconds, serving spatial computing environments and interactions. By shifting the computationally intensive process of generating spatial content to the cloud, the model aims to reduce the upfront costs required for VR adoption, thereby lowering the processing power needed in headsets and opening a path for cheaper devices with simpler hardware configurations. A ByteDance spokesperson declined to comment.