arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2607.01290cs.CV

AnchorSplat:用于高斯泼溅的快速且结构一致的细节合成

AnchorSplat: Fast and Structure Consistent Detail Synthesis for Gaussian Splatting

MAIS&NLPR, CASIA(中国科学院自动化研究所模式识别国家重点实验室);GigaAI(极佳科技);ShanghaiTech University(上海科技大学);Tsinghua University(清华大学)

Dexu Zhu, Jiangnan Shao, Xiaofeng Wang, Junxian Duan, Jie Cao, Zheng Zhu, Huaibo Huang

AI总结:

提出AnchorSplat,一种端到端3D原生深度网络,通过点锚机制实现几何一致性,无需原始多视图图像,在3DGS-SR基准上实现比优化方法快10^5倍的细节合成。

摘要:

3D高斯泼溅(3DGS)已成为高保真渲染的强大表示方法。然而,现有资产常常受到质量瓶颈的影响,例如细节缺失和纹理噪声。先前通过2D图像处理增强这些资产的尝试引入了多视图不一致性和高计算成本。在本文中,我们提出了一种名为AnchorSplat的新型3D原生细化范式。AnchorSplat是一个直接作用于3D结构的端到端深度网络,避免了传统3D-2D-3D流水线的昂贵优化开销。关键的是,AnchorSplat是一种严格的无源解决方案,不需要原始多视图图像。所提出方法的核心是点锚机制,它通过局部偏移约束强制几何一致性,缓解了不适定映射和梯度混淆。此外,AnchorSplat用单次乘法机制替代了迭代致密化。为了促进研究,我们构建了3DGS-SR,这是该任务的第一个大规模基准。实验表明,在3DGS-SR数据集上取得了最先进的结果,吞吐量比优化方法快高达10^5倍。值得注意的是,AnchorSplat在包括生成模型输出和真实世界扫描在内的不同数据分布上表现出鲁棒的零样本泛化能力。

英文摘要:

3D Gaussian Splatting (3DGS) has emerged as a powerful representation for high-fidelity rendering. However, existing assets often suffer from quality bottlenecks such as missing details and texture noise. Prior attempts to enhance these assets via 2D image processing introduce multi-view inconsistencies and high computational costs. In this paper, we propose a novel 3D-native refinement paradigm named AnchorSplat. AnchorSplat is an end-to-end deep network operating directly on 3D structures, avoiding the expensive optimization overhead of traditional 3D-2D-3D pipelines. Crucially, AnchorSplat is a strictly source-free solution requiring no original multi-view images. Central to the proposed method is the Point Anchor Mechanism, which enforces geometric consistency via local offset constraints, mitigating ill-posed mapping and gradient confounding. Furthermore, AnchorSplat replaces iterative densification with a single-pass multiplication mechanism. To facilitate research, we construct 3DGS-SR, the first large-scale benchmark for this task. Experiments demonstrate state-of-the-art results on the 3DGS-SR dataset, with throughput up to $10^5$ times faster than optimization methods. Notably, AnchorSplat exhibits robust zero-shot generalization across diverse data distributions, including generative model outputs and real-world scans.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑