NeoVerse：利用野外单目视频增强 4D 世界模型
NeoVerse: Enhancing 4D World Model with in-the-wild Monocular Videos
无位姿前馈 4D 重建加在线单目退化模拟，让 4D 世界模型可以直接吃野外单目视频训练，绕开多视角 4D 数据和繁琐预处理；做动态场景重建或新视角/新轨迹视频生成的人可以把它当作可扩展的训练与推理范式。
原文依据 NeoVerse features pose-free feed-forward 4D reconstruction, online monocular degradation pattern simulation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文提出 NeoVerse，一个多功能的 4D 世界模型，能够完成 4D 重建、新轨迹视频生成以及丰富的下游应用。我们首先指出当前 4D 世界建模方法中一个普遍的可扩展性局限，其成因要么是昂贵且专门的多视角 4D 数据，要么是繁琐的训练预处理。相比之下，NeoVerse 建立在一个核心理念之上，即让整个流程可扩展到多样化的野外单目视频。具体而言，NeoVerse 具备无位姿的前馈 4D 重建、在线单目退化模式模拟以及其他良好对齐的技术。这些设计赋予 NeoVerse 多功能性以及对各类领域的泛化能力。同时，NeoVerse 在标准重建与生成基准上取得了最优性能。
英文原文摘要 858 字符
In this paper, we propose NeoVerse, a versatile 4D world model that is capable of 4D reconstruction, novel-trajectory video generation, and rich downstream applications. We first identify a common limitation of scalability in current 4D world modeling methods, caused either by expensive and specialized multi-view 4D data or by cumbersome training pre-processing. In contrast, our NeoVerse is built upon a core philosophy that makes the full pipeline scalable to diverse in-the-wild monocular videos. Specifically, NeoVerse features pose-free feed-forward 4D reconstruction, online monocular degradation pattern simulation, and other well-aligned techniques. These designs empower NeoVerse with versatility and generalization to various domains. Meanwhile, NeoVerse achieves state-of-the-art performance in standard reconstruction and generation benchmarks.