WildWorld: A Large-Scale Dataset for Dynamic World Modeling with Actions and Explicit State toward Generative ARPG
2026-03-24 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created WildWorld, a huge dataset from a realistic video game, to help computers learn how actions affect game worlds over time. Unlike past datasets, WildWorld includes detailed state information and many different actions, which helps separate what actually changes the game from just pixel changes in the video. They also made WildBench, a way to test how well models follow actions and keep track of game states. Their experiments show it's still hard for models to understand complex actions and keep things consistent for a long time, suggesting future models need to pay close attention to underlying states.
dynamical systemsreinforcement learningaction-conditioned dynamicsvideo world modelslatent statesstate annotationsphotorealistic datasetlong-horizon consistencystate-aware video generationMonster Hunter: Wilds
Authors
Zhen Li, Zian Meng, Shuwei Shi, Wenshuo Peng, Yuwei Wu, Bo Zheng, Chuanhao Li, Kaipeng Zhang
Abstract
Dynamical systems theory and reinforcement learning view world evolution as latent-state dynamics driven by actions, with visual observations providing partial information about the state. Recent video world models attempt to learn this action-conditioned dynamics from data. However, existing datasets rarely match the requirement: they typically lack diverse and semantically meaningful action spaces, and actions are directly tied to visual observations rather than mediated by underlying states. As a result, actions are often entangled with pixel-level changes, making it difficult for models to learn structured world dynamics and maintain consistent evolution over long horizons. In this paper, we propose WildWorld, a large-scale action-conditioned world modeling dataset with explicit state annotations, automatically collected from a photorealistic AAA action role-playing game (Monster Hunter: Wilds). WildWorld contains over 108 million frames and features more than 450 actions, including movement, attacks, and skill casting, together with synchronized per-frame annotations of character skeletons, world states, camera poses, and depth maps. We further derive WildBench to evaluate models through Action Following and State Alignment. Extensive experiments reveal persistent challenges in modeling semantically rich actions and maintaining long-horizon state consistency, highlighting the need for state-aware video generation. The project page is https://shandaai.github.io/wildworld-project/.