MoQSplat improves streaming of 3D scenes with adaptive network transport
MoQSplat: Adaptive Progressive Streaming of 3D Gaussian Splatting via MoQ
MultimediaGraphicsNetworking and Internet Architecture
Summary
3D Gaussian Splatting can create very realistic 3D images but sending such huge 3D data over the internet is slow and clunky. The authors designed MoQSplat, a way to send smaller parts of the 3D scene through a better internet transport called Media over QUIC, which avoids common delays and lets users see the parts they're looking at in high quality first. The system automatically adjusts what data to send based on where and how users look around, improving speed and experience. They also tested different methods for sending parts of the 3D data progressively and found some ways work better than others.
What this means in practice
- •For virtual reality developers: Stream large photorealistic 3D scenes efficiently by adapting quality and regions to user viewpoint in VR applications.
- •For online gaming studios: Reduce lag and improve visual quality in games that use detailed 3D environment streaming with six degrees of freedom navigation.$Commercial implications: Enables selling smoother and more immersive 3D content streaming for multiplayer or cloud-based gaming platforms.
Authors
Emanuele Artioli, Mohammadreza Ghafari, Md Tariqul Islam, Farzad Tashtarian, Christian Rothenberg, Christian Timmerer
Abstract
3D Gaussian Splatting (3DGS) enables photorealistic novel view synthesis, but transmitting gigabyte-scale scene data remains challenging for immersive applications. Traditional HTTP Adaptive Streaming over TCP introduces Head-of-Line (HOL) blocking and coarse segmenting ill-suited to fine-grained 3DGS delivery. We propose MoQSplat, which maps 3DGS content onto the Media over QUIC (MoQ) transport hierarchy. MoQSplat partitions scenes into spatial Tracks, clusters splats into semantically coherent Groups, and constructs progressive-quality Subgroups mapped to independent QUIC streams to eliminate connection-level HOL blocking. Using a stateless, subscriber-driven adaptation loop, clients dynamically request spatial regions and quality tiers based on six degrees of freedom (6-DoF) frustum visibility, distance, and foveal alignment. We evaluate the core components on a prototype implementation, showing that opacity-based pruning outperforms scale-based pruning for progressive delivery. The source code is available at https://github.com/emanuele-artioli/MoQSplat.