一句话：提出 Nebula 协同渲染框架，通过云端时序感知 LoD 搜索与客户端立体光栅化，实现大规模 3DGS 的 VR 加速。
针对大规模 3D 高斯泼溅（3DGS）在 VR 中可扩展性不足、云端高保真内容受带宽限制的问题，Nebula 提出协同渲染框架。它不在云端传输视频，而是传输 LoD 搜索后的中间结果，大幅降低云-端通信；云端引入时序感知 LoD 搜索，利用帧间时序一致性缓解不规则内存访问与冗余访问；客户端提出立体光栅化，使双眼共享大部分计算并保持比特精确质量，仅需少量硬件增强。评估显示，相比有损视频流，Nebula 实现 2.7× motion-to-photon 加速，并降低 1925% 带宽。
3D Gaussian splatting (3DGS) has drawn significant attention in the architectural community recently. However, current architectural designs often overlook the 3DGS scalability, making them fragile for extremely large-scale 3DGS. Meanwhile, the VR bandwidth requirement makes it impossible to deliver high-fidelity and smooth VR content from the cloud. We present Nebula, a coherent acceleration framework for large-scale 3DGS collaborative rendering. Instead of streaming videos, Nebula streams intermediate results after the LoD search, reducing 1925% data communication between the cloud and the client. To further enhance the motion-to-photon experience, we introduce a temporal-aware LoD search in the cloud that tames the irregular memory access and reduces redundant data access by exploiting temporal coherence across frames. On the client side, we propose a novel stereo rasterization that enables two eyes to share most computations during the stereo rendering with bit-accurate quality. With minimal hardware augmentations, Nebula achieves 2.7× motion-to-photon speedup and reduces 1925% bandwidth over lossy video streaming.
@inproceedings{ZhuLLWZLGL026,
title = {Nebula: Infinite-Scale 3D Gaussian Splatting in VR via Collaborative Rendering and Accelerated Stereo Rasterization},
author = {He Zhu and Zheng Liu and Xingyang Li and Anbang Wu and Jieru Zhao and Fangxin Liu and Yiming Gan and Jingwen Leng and Yu Feng},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790190}
}