State-of-the-art video generative models produce promising visual content yet often violate basic physics principles, limiting their utility. While some attribute this deficiency to insufficient physics understanding from pre-training, we find that the shortfall in physics plausibility also stems from suboptimal inference strategies. We therefore introduce WMReward and treat improving physics plausibility of video generation as an inference-time alignment problem. In particular, we leverage the strong physics prior of a latent world model (here, VJEPA-2) as a reward to search and steer multiple candidate denoising trajectories, enabling scaling test-time compute for better generation performance. Empirically, our approach substantially improves physics plausibility across image-conditioned, multiframe-conditioned, and text-conditioned generation settings, with validation from human preference study. Notably, in the ICCV 2025 Perception Test PhysicsIQ Challenge, we achieve a final score of 62.64%, winning first place and outperforming the previous state of the art by 7.42%. Our work demonstrates the viability of using latent world models to improve physics plausibility of video generation, beyond this specific instantiation or parameterization.

翻译：当前最先进的视频生成模型能够产生视觉效果出色的内容，但常常违背基本的物理原理，这限制了其实用性。虽然部分研究将此缺陷归因于预训练阶段对物理规律的理解不足，但我们发现物理合理性的欠缺也源于次优的推理策略。因此，我们提出了WMReward方法，并将提升视频生成的物理合理性视为一个推理时对齐问题。具体而言，我们利用一个潜在世界模型（本文采用VJEPA-2）所具备的强物理先验作为奖励，通过搜索并引导多条候选去噪轨迹，实现了测试时计算资源的扩展以提升生成性能。实验表明，我们的方法在图像条件生成、多帧条件生成和文本条件生成等多种设置下，均显著提升了物理合理性，并通过人类偏好研究得到了验证。值得注意的是，在ICCV 2025感知测试PhysicsIQ挑战赛中，我们以62.64%的最终得分获得第一名，较先前最佳性能提升了7.42%。我们的工作证明了利用潜在世界模型提升视频生成物理合理性的可行性，其价值超越了特定的模型实例或参数化方案。