用全局场景令牌提升3D高斯点云效率，实现快速渲染与小体积存储。
GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens
- 先建全局隐式表征，再解码几何，避免局部冗余
- 仅用1.6万高斯点即达良好画质，模型仅4MB
- 推理速度<78毫秒，适合实时应用与移动设备
高效地分配3D表示基元是3D高斯点云的核心，直接影响表示紧凑性、重建速度和渲染保真度。现有方法在迭代优化或前馈推理间存在显著权衡，主要因依赖缺乏全局感知的局部启发式分配策略。当前前馈方法多为像素对齐或体素对齐，通过将像素反投影为密集视图对齐的基元，导致3D资产中引入冗余。随着输入视角增多，表示规模扩大，全局一致性变弱。为此，我们提出GlobalSplat，基于‘对齐先行，解码后行’原则。该方法学习一个紧凑的全局隐式场景表征，编码多视角输入并提前解决跨视图对应关系，再解码显式3D几何。关键在于，此框架无需预训练像素预测主干或复用密集基线的潜在特征，即可实现紧凑且全局一致的重建。通过粗到细的训练流程逐步增加解码容量，自然防止表示膨胀。在RealEstate10K和ACID数据集上，模型以最少16,000个高斯点达到竞争性新视角合成性能，远低于密集流水线所需数量，模型大小仅为4MB。此外，GlobalSplat推理速度显著快于基线，单次前向传播耗时低于78毫秒。
原文摘要 · Abstract (English)
The efficient spatial allocation of primitives serves as the foundation of 3D Gaussian Splatting, as it directly dictates the synergy between representation compactness, reconstruction speed, and rendering fidelity. Previous solutions, whether based on iterative optimization or feed-forward inference, suffer from significant trade-offs between these goals, mainly due to the reliance on local, heuristic-driven allocation strategies that lack global scene awareness. Specifically, current feed-forward methods are largely pixel-aligned or voxel-aligned. By unprojecting pixels into dense, view-aligned primitives, they bake redundancy into the 3D asset. As more input views are added, the representation size increases and global consistency becomes fragile. To this end, we introduce GlobalSplat, a framework built on the principle of align first, decode later. Our approach learns a compact, global, latent scene representation that encodes multi-view input and resolves cross-view correspondences before decoding any explicit 3D geometry. Crucially, this formulation enables compact, globally consistent reconstructions without relying on pretrained pixel-prediction backbones or reusing latent features from dense baselines. Utilizing a coarse-to-fine training curriculum that gradually increases decoded capacity, GlobalSplat natively prevents representation bloat. On RealEstate10K and ACID, our model achieves competitive novel-view synthesis performance while utilizing as few as 16K Gaussians, significantly less than required by dense pipelines, obtaining a light 4MB footprint. Further, GlobalSplat enables significantly faster inference than the baselines, operating under 78 milliseconds in a single forward pass. Project page is available at https://r-itk.github.io/globalsplat/
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。