Demiurge stores accumulated scene information in a dynamic 4D point-cloud cache.
Persistent memory keeps generated 4D worlds consistent across camera paths
Demiurge repeatedly adds synthesized views and recovered geometry to a dynamic point-cloud cache, allowing later camera trajectories to reuse earlier results.
Big Tech
Hanxin Zhu · Jianqi Ma · Cong Wang · Zhihao Hu · Junyao Gao · Tianyu He · +7 more
USTC · CASIA · Tencent · Tongji University · Microsoft
Research Digest··2 min read
Zhu and colleagues present Demiurge, a system that turns one monocular video into a navigable 4D representation, meaning a 3D scene that evolves over time.
Why this paper
From Microsoft and 5 others
In one line
Demiurge constructs navigable 4D worlds from a single video by autoregressively rerendering novel views and merging them into a dynamic 4D point-cloud memory.
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.
§