Researchers have introduced 4DStreamCtrl, a novel system for interactive video generation that allows for real-time, 3D-consistent control over both camera movement and object trajectories. Unlike previous methods that handled these aspects separately or offline, 4DStreamCtrl unifies camera motion, object paths, and depth into a single 3D point-track representation. This approach enables joint control and depth editing within a single forward pass, achieving up to 20 FPS for 480p video generation. The system also demonstrates temporal coherence over hundreds of frames, paving the way for interactive world models and real-time visual imagination for embodied agents. AI
IMPACT Enables interactive world models and real-time visual imagination for embodied agents.
RANK_REASON Research paper detailing a new method for video generation. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →