Grounding World Simulation Models in a Real-World Metropolis

TL;DR

This paper introduces Seoul World Model (SWM), a city-scale video generation model grounded in real-world Seoul data, capable of producing long, diverse, and spatially faithful urban videos with temporal consistency.

Contribution

The paper presents SWM, a novel city-scale world model that integrates retrieval-augmented conditioning, synthetic data generation, and a Virtual Lookahead Sink for stable, long-horizon urban video synthesis.

Findings

SWM outperforms existing models in urban video fidelity.

Supports diverse camera trajectories and text prompts.

Achieves long-horizon, temporally consistent city videos.

Abstract

What if a world simulation model could render not an imagined environment but a city that actually exists? Prior generative world models synthesize visually plausible yet artificial environments by imagining all content. We present Seoul World Model (SWM), a city-scale world model grounded in the real city of Seoul. SWM anchors autoregressive video generation through retrieval-augmented conditioning on nearby street-view images. However, this design introduces several challenges, including temporal misalignment between retrieved references and the dynamic target scene, limited trajectory diversity and data sparsity from vehicle-mounted captures at sparse intervals. We address these challenges through cross-temporal pairing, a large-scale synthetic dataset enabling diverse camera trajectories, and a view interpolation pipeline that synthesizes coherent training videos from sparse street-view images. We further introduce a Virtual Lookahead Sink to stabilize long-horizon generation by continuously re-grounding each chunk to a retrieved image at a future location. We evaluate SWM against recent video world models across three cities: Seoul, Busan, and Ann Arbor. SWM outperforms existing methods in generating spatially faithful, temporally consistent, long-horizon videos grounded in actual urban environments over trajectories reaching hundreds of meters, while supporting diverse camera movements and text-prompted scenario variations.