MetaEarth3D: Unlocking World-scale 3D Generation with Spatially Scalable Generative Modeling
Jinqi Cao, Zhiping Yu, Baihong Lin, Chenyang Liu, Zhenwei Shi, Zhengxia Zou · 2026
Read MetaEarth3D: Unlocking World-scale 3D Generation with Spatially Scalable Generative Modeling by Jinqi Cao, Zhiping Yu, Baihong Lin, Chenyang Liu, Zhenwei Shi, Zhengxia Zou (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-22828.txt.
Description
Cao and five colleagues present MetaEarth3D, a generative foundation model that treats spatial scale as a new axis of intelligence rather than only scaling parameters and data. It generates spatially consistent, unbounded three-dimensional scenes at planetary scale, from large terrains and cities down to street blocks. Trained on 10 million real-world images from around the globe, it shows strong visual and geospatial realism. Beyond generation, it serves as a data engine for virtual environments in Earth observation and simulation.
Catalog record
| CREATOR | Jinqi Cao, Zhiping Yu, Baihong Lin, Chenyang Liu, Zhenwei Shi, Zhengxia Zou |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 94335 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.22828