Scaling4D：借助大规模单目视频推动视频新视角合成的前沿
Scaling4D: Pushing the Frontier of Video Novel View Synthesis through Large-Scale Monocular Videos
把单目视频的新视角合成重新表述为「对应关系引导的生成」任务，从理论上消除 inpainting 式训练与推理之间的鸿沟；做动态场景重建或视频生成的研究者可以直接借用这一训练范式与合成数据管线来扩展数据规模。
原文依据 reformulating VNVS into a general correspondence-guided generation task
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视频新视角合成（VNVS）旨在从单视角视频渲染动态场景的任意新视角，但其算法训练面临一大挑战：缺乏大规模多视角视频数据集。已有方法通常将单目数据框定为图像修复任务来训练，这往往导致训练与推理之间的鸿沟以及视觉伪影。合成多视角数据虽能部分缓解数据稀缺问题，但其高昂的采集成本与有限的多样性限制了可扩展性。为解决这些问题，我们提出 Scaling4D，一种在理论上弥合训练-推理鸿沟、同时利用大规模单目视频进行训练的新策略。具体而言，我们从更高层次的视角审视该问题，将 VNVS 重新表述为一种通用的对应关系引导生成任务。此外，结合大量真实世界数据，我们建立了一套与训练策略相集成的合成数据管线，以提升精度。定性与定量结果均表明性能与训练数据量呈正相关，验证了其可扩展性。
英文原文摘要 1152 字符
Video Novel View Synthesis (VNVS) aims to render arbitrary novel viewpoints of dynamic scenes from a single-view video, but its algorithmic training faces a major challenge: the lack of large-scale multi-view video datasets. Prior methods often train on monocular data by framing it as an inpainting task, which typically leads to a training-inference gap and visual artifacts. While synthetic multi-view data can partially alleviate the data scarcity issue, its high acquisition costs and limited diversity restrict scalability. To address these problems, we propose Scaling4D, a novel strategy that theoretically bridges the training-inference gap while leveraging large-scale monocular videos for training. Specifically, we take a higher-level perspective on the problem, reformulating VNVS into a general correspondence-guided generation task. Furthermore, in conjunction with extensive real-world data, we establish a synthetic data pipeline integrated with our training strategy to enhance precision. Qualitative and quantitative results demonstrate a positive correlation between performance and training data volume, confirming the scalability.