视频模型会忘记自己看过什么
现在的视频生成模型能拍出越来越长的连续画面，但有个没人注意的硬伤：它记不住自己前面拍过什么。研究者做了个叫 WorldMemBench 的测试，让 9 个主流视频模型在 260 个场景里离开再回来，看它们能不能记住人物长相、物体位置、环境布局，以及事件进展到哪一步。结果发现：静态记忆强的模型，动态状态照样会崩；而且模型每多回访一次，记忆就再掉一截，不是只忘一次就稳定。这不是你明天能用上的东西，但它划出了一条新赛道——当视频模型从“生成画面”走向“生成世界”，记忆才是那个真正决定世界能不能持续的瓶颈。
📄 原文摘要(英文)
As video models enable longer and more realistic interactive experiences, maintaining a world consistent with its history becomes essential. This requires both preserving previously observed content and ensuring coherent world evolution over time. World memory is a critical capability to evaluate alongside visual fidelity and video duration. We introduce WorldMemBench, a benchmark for systematically evaluating world memory in video world models. Across 260 test cases, it assesses static memory through the retention of human identity and appearance, object appearance and geometry, and environmental content and geometry, and dynamic memory through event progression and state persistence. Using controlled camera departures and multi-revisit trajectories, the benchmark evaluates static content against fixed initial references and dynamic events against expected state transitions. It combines appearance and geometry metrics with event-specific assessments, and reports video quality and camera-following accuracy as complementary diagnostics. Evaluation of nine representative models shows that strong static retention does not imply reliable dynamic state evolution, while repeated revisits reveal memory degradation beyond the first return. Human studies support the alignment between our automatic evaluations and human judgments. WorldMemBench provides a systematic testbed for diagnosing memory failures and guiding the development of persistent, temporally coherent video worlds.