The authors built StoryEngine around an authoritative semantic plan that tracks entity placement and story-relevant attributes.
Tracking world states improves coherence in generated video stories
StoryEngine propagates explicit entity states between shots, grounds recurring visuals in canonical references, and repairs local continuity failures.
Chinese Tech
Yingrui Wang · Zeqing Wang · Yeying Jin
Tencent · National University of Singapore
Research Digest··2 min read
Wang, Wang and Jin introduce an agentic video pipeline that represents each story event as a structured change to characters, objects and locations rather than relying primarily on prior generated frames.
Why this paper
From Tencent and National University of Singapore
In one line
StoryEngine uses a state-grounded agentic framework to produce coherent and consistent long-form video stories.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§