ByteDance is currently developing an AI model designed for real-time spatial video generation. The project is under the direct supervision of the company founder, Zhang Yiming, who is coordinating internal resources to support the initiative.
The new system is built upon the existing Seedance video model. It aims to create virtual environments that react to user movements and voice commands. Potential use cases for the technology include gaming, livestreaming, short-form entertainment, robotics, and autonomous systems.
Technical specifications indicate the model could achieve a latency of approximately 0.05 seconds while maintaining a frame rate of 20 frames per second. ByteDance plans to leverage its own cloud infrastructure to manage the significant computational requirements of generating spatial content.
The company intends to debut the model as early as next month, though the timeline remains subject to change. This development aligns with broader industry efforts to create interactive world models, similar to Google's Genie project.