WorldDirector: Building Controllable World Simulators with Persistent Dynamic Memory
This work addresses the problem of maintaining object identity and controllability in long-duration video generation, offering a new paradigm for building interactive world simulators.
WorldDirector introduces a video world model that decouples motion orchestration from visual generation, enabling persistent memory of dynamic objects and unrestricted viewpoint control. It achieves strict physical logic and appearance stability for objects re-entering scenes after long absences.
We present WorldDirector, a highly controllable video world model framework designed for persistent dynamic object memory and unrestricted viewpoint exploration. Unlike existing world models that entangle physical dynamics with pixel rendering and rely on continuous visual observation to sustain motion, our framework explicitly decouples semantic motion orchestration from visual generation. By leveraging an LLM to coordinate 3D trajectories with camera movements and subsequently employing these orchestrated trajectories as control signals for video generation, our approach ensures strict physical logic and appearance stability, successfully preserving the exact visual identities of dynamic entities even when they re-enter the scene after prolonged periods out of view. Experimental results demonstrate that our method supports the synthesis of complex and extended events with unprecedented controllability and persistent dynamic object memory. Project Page: https://worlddirector.github.io/