Dreamland: Controllable World Creation with Simulator and Generative Models

Sicheng Mo, Ziyang Leng, Leon Liu, Weizhen Wang, Honglin He, Bolei Zhou · 2025

Read Dreamland: Controllable World Creation with Simulator and Generative Models by Sicheng Mo, Ziyang Leng, Leon Liu, Weizhen Wang, Honglin He, Bolei Zhou (2025) 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-2506-08006.txt.

Description

Sicheng Mo and co-authors presented Dreamland in 2025, a frame for creating virtual worlds that users can control piece by piece. Large video models make rich, realistic scenes, but they rarely let users edit single elements. A physics simulator gives that control, yet lacks visual realism.

Dreamland joins the two, using a layered world form that holds both pixel and object meaning plus geometry. That bridge keeps control fine, cuts adaptation cost, and works with ready or future generative models. A new data set supports training and testing. The system beat baselines on image quality and control, and may help train embodied agents.

Catalog record

CREATORSicheng Mo, Ziyang Leng, Leon Liu, Weizhen Wang, Honglin He, Bolei Zhou
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-08006

Browse the full library · About the Tower of Babel