Captain Safari: A World Engine with Pose-Aligned 3D Memory

TL;DR

Captain Safari introduces a pose-conditioned world engine that uses a persistent memory and retrieval mechanism to generate long, 3D-consistent videos with accurate camera control, outperforming existing methods on a new drone video dataset.

Contribution

The paper presents a novel pose-conditioned world engine with a dynamic memory retrieval system, enabling stable, long-horizon, 3D-consistent video synthesis under complex camera trajectories.

Findings

Outperforms state-of-the-art camera-controlled video generators in quality and consistency.

Reduces error metrics such as MEt3R and FVD significantly.

Achieves higher preference scores in human evaluations.

Abstract

World engines aim to synthesize long, 3D-consistent videos that support interactive exploration of a scene under user-controlled camera motion. However, existing systems struggle under aggressive 6-DoF trajectories and complex outdoor layouts: they lose long-range geometric coherence, deviate from the target path, or collapse into overly conservative motion. To this end, we introduce Captain Safari, a pose-conditioned world engine that generates videos by retrieving from a persistent world memory. Given a camera path, our method maintains a dynamic local memory and uses a retriever to fetch pose-aligned world tokens, which then condition video generation along the trajectory. This design enables the model to maintain stable 3D structure while accurately executing challenging camera maneuvers. To evaluate this setting, we curate OpenSafari, a new in-the-wild FPV dataset containing high-dynamic drone videos with verified camera trajectories, constructed through a multi-stage geometric and kinematic validation pipeline. Across video quality, 3D consistency, and trajectory following, Captain Safari substantially outperforms state-of-the-art camera-controlled generators. It reduces MEt3R from 0.3703 to 0.3690, improves AUC@30 from 0.181 to 0.200, and yields substantially lower FVD than all camera-controlled baselines. More importantly, in a 50-participant, 5-way human study where annotators select the best result among five anonymized models, 67.6% of preferences favor our method across all axes. Our results demonstrate that pose-conditioned world memory is a powerful mechanism for long-horizon, controllable video generation and provide OpenSafari as a challenging new benchmark for future world-engine research.