ResStreamGS: Residual-Sensitive Dual-Stream Gaussian Splatting for Mobile Volumetric Video Streaming

Haoran Liu; Di Wu; Wang Luo; Hengyuan Na; Miao Hu; Guocong Quan

cs.GROct 15, 2026

ResStreamGS: Residual-Sensitive Dual-Stream Gaussian Splatting for Mobile Volumetric Video Streaming

Authors: Haoran Liu, Di Wu, Wang Luo, Hengyuan Na, Miao Hu, Guocong Quan

Abstract

Real-time mobile volumetric video enables immersive free-viewpoint experiences, but delivering high-quality 3D content under constrained bandwidth and computation remains challenging. Recent 3D Gaussian Splatting (3DGS) systems, such as V $^3$ , enable efficient mobile streaming by encoding time-varying Gaussian attributes with commodity video codecs and hardware decoding. However, existing pipelines treat all Gaussians as frame-varying signals, which is poorly suited to full-scene content where large portions remain temporally stable. Our empirical study shows that frame-wise adaptation and video coding of all Gaussians cause quality degradation in complex full-scene content, particularly in temporally stable regions. To address this issue, we propose ResStreamGS, a residual-sensitive dual-stream framework that processes Gaussians according to their temporal behavior. ResStreamGS identifies residual-sensitive primitives through lightweight probe adaptation and reuses stable Gaussians within each Group of Pictures (GOP). Stable Gaussians are transmitted once per GOP, while only residual-sensitive Gaussians remain in the frame-wise hardware-decodable video stream. Experiments on complex full-scene volumetric videos show that ResStreamGS improves post-codec PSNR by 7.51--8.35~~dB over a controlled V $^3$ -style all-Gaussians baseline, reduces amortized payload by up to 56.5%, and achieves 30.883~~FPS on a Qualcomm Snapdragon 778G platform.

Figures & tables

Figure 3: (a) Stage-wise PSNR degradation in nominally stable regions: the frozen configuration defines the 31.4-dB reference, while frame-wise retraining and video coding incur losses of 4.69 and 1.74 dB, respectively, totaling 6.43 dB. (b) Gaussian attribute change ratios across target frames within the GOP.

Figure 4: Overview of ResStreamGS. Each GOP is partitioned into group-level stable Gaussians and frame-wise residual-sensitive Gaussians. Blue elements indicate the stable Gaussians stream path, and orange elements indicate the residual-sensitive path.

Method	Mobile display	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	DISTS $\downarrow$	Payload (MB) $\downarrow$
Neu3D
Uncompressed Reference	–	34.60	0.9554	0.100	0.05188	107.810
4DGS	$\times$	29.70	0.9207	0.121	0.06514	0.895
GIFStream	$\times$	30.59	0.9275	0.151	0.05658	0.039
V 3	$\checkmark$	25.53	0.8276	0.320	0.16322	3.741
[0pt][0pt] ResStreamGS	$\checkmark$	33.04	0.9425	0.107	0.03756	1.627

Table 1: Quantitative comparison of reconstruction quality and amortized streaming payload on Neu3D and MeetRoom. Neu3D results are averaged over five full-scene sequences. Mobile playback indicates whether a method supports on-device playback (

\checkmark

\times

), and payload denotes the complete decoder-required representation amortized per frame. Best and second-best results among compressed methods are shown in bold and underlined , respectively.

Figure 5: Qualitative comparison across representative scenes and methods.

Figure 6: Client-side active latency comparison between V 3 and ResStreamGS during mobile playback. The total latency is decomposed into video decoding, visibility culling and sorting, and rendering.

Figure 7: Quality–payload trade-off under different residual-sensitive Gaussian ratios on cook_spinach , coffee_martini , and MeetRoom. The ratio varies from 0.5% to 100%, with payload normalized by the all-Gaussians baseline. Stars denote the operating points automatically selected by the probe-stage quality criterion.

Figure 8: Client-side active latency breakdown for ResStreamGS and its ablated variants during mobile playback. The ablations disable stable caching, view-frustum culling, and opacity pruning, respectively.

Ranking score	PSNR (dB) $\uparrow$	Payload (MB/frame) $\downarrow$
Random	29.947	1.805
Position	31.193	1.846
+ Rotation	31.573	1.861
+ Scale	31.387	1.848
+ Opacity	31.337	1.844
+ Color	31.322	1.859

Table 2: Ablation of Gaussian ranking cues on coffee_martini .

Configuration	Latency (ms) $\downarrow$	FPS $\uparrow$	Memory (MiB) $\downarrow$
V 3	379.95	2.614	518.4
w/o stable caching	87.80	10.392	470.4
w/o frustum culling	51.86	16.859	656.2
w/o opacity pruning	34.48	24.531	325.0
Full ResStreamGS	27.44	30.883	331.9

Table 3: Mobile playback performance and component ablations.

3D Gaussian Splatting

3D Gaussian Splatting represents scenes as optimizable 3D Gaussian primitives for fast, photorealistic rendering from new viewpoints.

See all 1,019 papers

GSStream: 3D Gaussian Splatting based Volumetric Scene Streaming System
Mar 10, 2026Zhiye Tang, Qiudan Zhang, Lei Zhang +3Streaming 3D Reconstruction 3D Gaussian Splatting
College of Computer Science and Software Engineering, Shenzhen University, Shenzhen, 518060, China. · Department of Computer Science, City University of Hong Kong, Kowloon, Hong Kong SAR, China. · School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan 430074, China.
DecoGS: Adaptive Static-Dynamic Decoupling of 3D Gaussians for Free-Viewpoint Video Streaming
Sep 15, 2026Idil Sulo, Alexey Supikov, Ilke Demir +13D Gaussian Splatting Streaming 3D Reconstruction
University of Bonn, Almetra · V3DEO · Cauth AI +1
EvoGS: Constructing Continuous-Layered Gaussian Splatting with Evolution Tree for Scalable 3D Streaming
Jun 5, 2026Yuang Shi, Simone Gasparini, Géraldine Morin +13DGS Compression 3D Gaussian Splatting
National University of Singapore, Singapore · IRIT - University of Toulouse, France · IPAL, IRL2955, Singapore

Mar 10, 2026J/K move · Enter open · S save

GSStream: 3D Gaussian Splatting based Volumetric Scene Streaming System

Zhiye Tang, Qiudan Zhang, Lei Zhang, Junhui Hou+2

College of Computer Science and Software Engineering, Shenzhen University, Shenzhen, 518060, China. · Department of Computer Science, City University of Hong Kong, Kowloon, Hong Kong SAR, China. · School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan 430074, China.

Streaming 3D Reconstruction 3D Gaussian Splatting

Open paper AskPDF arXiv

Recently, the 3D Gaussian splatting (3DGS) technique for real-time radiance field rendering has revolutionized the field of volumetric scene representation, providing users with an immersive experience. But in return, it also poses a large amount of data volume, which is extremely bandwidth-intensive. Cutting-edge researchers have tried to introduce different approaches and construct multiple variants for 3DGS to obtain a more compact scene representation, but it is still challenging for real-time distribution. In this paper, we propose GSStream, a novel volumetric scene streaming system to support 3DGS data format. Specifically, GSStream integrates a collaborative viewport prediction module to better predict users' future behaviors by learning collaborative priors and historical priors from multiple users and users' viewport sequences and a deep reinforcement learning (DRL)-based bitrate adaptation module to tackle the state and action space variability challenge of the bitrate adaptation problem, achieving efficient volumetric scene delivery. Besides, we first build a user viewport trajectory dataset for volumetric scenes to support the training and streaming simulation. Extensive experiments prove that our proposed GSStream system outperforms existing representative volumetric scene streaming systems in visual quality and network usage. Demo video: https://youtu.be/3WEe8PN8yvA.

Abstract

3D Gaussian Splatting

Explore similar work

GSStream: 3D Gaussian Splatting based Volumetric Scene Streaming System

DecoGS: Adaptive Static-Dynamic Decoupling of 3D Gaussians for Free-Viewpoint Video Streaming

EvoGS: Constructing Continuous-Layered Gaussian Splatting with Evolution Tree for Scalable 3D Streaming