pith. machine review for the scientific record. sign in

arxiv: 2605.09279 · v1 · submitted 2026-05-10 · 💻 cs.GR · cs.CV· cs.MM· cs.NI· eess.IV

Recognition: 2 theorem links

· Lean Theorem

CAGS: Color-Adaptive Volumetric Video Streaming with Dynamic 3D Gaussian Splatting

Cong Zhang, Daheng Yin, Fang Dong, Fangxin Wang, Isaac Ding, Jiangchuan Liu, Jianxin Shi, Miao Zhang, Yili Jin, Zhaowu Huang

Authors on Pith no claims yet

Pith reviewed 2026-05-12 04:07 UTC · model grok-4.3

classification 💻 cs.GR cs.CVcs.MMcs.NIeess.IV
keywords volumetric video streaming3D Gaussian Splattingadaptive streamingcolor correctionvector quantizationlevels of detailphotorealistic rendering
0
0 comments X

The pith

CAGS corrects color distortions from compressed 3D Gaussians with low-resolution reference images to enable high-quality adaptive volumetric video streaming.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents CAGS as a system for streaming volumetric videos based on 3D Gaussian Splatting under varying network conditions. It establishes levels of detail through vector quantization on Gaussian attributes and restores color accuracy in the rendered output by applying a low-resolution reference image sent from the server. This method targets the limitations of density-based detail selection, which produces visible artifacts in Gaussian scenes, and shows that aggressive compression mainly introduces correctable color shifts rather than structural damage. A reader would care because it directly supports real-time photorealistic remote 3D interaction over typical internet links without requiring constant high bandwidth.

Core claim

The authors show that vector quantization can serve as a suitable mechanism for creating levels of detail in Gaussian representations, and that color distortions introduced by heavy attribute compression are effectively removed by rendering a low-resolution reference image on the server and using it for client-side restoration in the final view.

What carries the argument

The Color-Adaptive scheme, which combines vector quantization to select Gaussian attribute levels of detail with server-rendered low-resolution reference images that correct color errors during client-side rendering.

If this is right

  • The system delivers 5 to 20 dB higher PSNR than prior adaptive streaming approaches when bandwidth varies.
  • Rendering and restoration run substantially faster than existing scalable Gaussian compression pipelines.
  • The approach works without modification across multiple existing Gaussian representation formats.
  • It supports low-latency photorealistic interaction in applications such as telepresence and remote operation over heterogeneous networks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same reference-based correction could be tested on other attribute-heavy 3D representations that suffer color shift under compression.
  • Client-side restoration might integrate with view-dependent rendering to further cut transmitted data in multi-user sessions.
  • If the reference image itself can be progressively refined, the method might support graceful quality scaling beyond binary LoD switches.
  • Deployment in edge computing environments could reduce server load by shifting more restoration work to capable clients.

Load-bearing premise

Aggressive compression of Gaussian attributes produces mainly color distortions that a low-resolution reference image can reliably fix in the final rendered frame.

What would settle it

A controlled test under real fluctuating bandwidth where the reference-image correction yields no measurable PSNR gain or where density-based LoD methods produce comparable quality to the proposed VQ approach.

Figures

Figures reproduced from arXiv: 2605.09279 by Cong Zhang, Daheng Yin, Fang Dong, Fangxin Wang, Isaac Ding, Jiangchuan Liu, Jianxin Shi, Miao Zhang, Yili Jin, Zhaowu Huang.

Figure 1
Figure 1. Figure 1: Overview of a typical adaptive VV streaming pipeline. The server [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visual comparison of density-based LoD used in LTS [Sun et al [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of CAGS. The server predicts the viewport, selects tiles and LoDs, renders a low-resolution reference image from the highest-quality layer of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of compressed frame size and PSNR for super-resolution, [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of Scalable Vector Quantization (L0: the base layer quantized data; L1–L3: enhancement layers quantized data; C0–C3: corresponding [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: PRPA aligns the reference image in three steps: [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Visualization of PRPA results with different reference FoVs. A small [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Throughput of the selected network trace from the 5Gophers dataset [PITH_FULL_IMAGE:figures/full_fig_p007_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of visual quality under fixed bandwidth. The y-axis ranges vary across subplots while maintaining equal scale spans for each metric [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of visual quality under fluctuating bandwidth. The y-axis ranges vary across subplots while maintaining equal scale spans for each metric [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Visual results of the frame 64 in “coffee martini” under 30Mbps. Additional visual and video results are included in supplementary materials. [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗
Figure 15
Figure 15. Figure 15: Visual quality versus frame size across LoDs constructed by interleaving SVQ layers, evaluated on the first frame of each video. Each point corresponds [PITH_FULL_IMAGE:figures/full_fig_p014_15.png] view at source ↗
Figure 17
Figure 17. Figure 17: Comparison of visual quality under fixed bandwidth. The y-axis ranges vary across subplots while maintaining equal scale spans for each metric [PITH_FULL_IMAGE:figures/full_fig_p017_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Comparison of visual quality under fluctuating bandwidth. The y-axis ranges vary across subplots while maintaining equal scale spans for each metric [PITH_FULL_IMAGE:figures/full_fig_p018_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Comparison of visual quality under fixed bandwidth on volumetric videos prepared by Dynamic 3DGS [Luiten et al [PITH_FULL_IMAGE:figures/full_fig_p019_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Comparison of visual quality under fluctuating bandwidth on volumetric videos prepared by Dynamic 3DGS [Luiten et al [PITH_FULL_IMAGE:figures/full_fig_p020_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Comparison of visual quality under fixed bandwidth on volumetric videos prepared by HiCoM [Gao et al [PITH_FULL_IMAGE:figures/full_fig_p021_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Comparison of visual quality under fluctuating bandwidth on volumetric videos prepared by HiCoM [Gao et al [PITH_FULL_IMAGE:figures/full_fig_p022_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Comparison of visual quality under fixed bandwidth on volumetric videos prepared by 4DGS [Wu et al [PITH_FULL_IMAGE:figures/full_fig_p023_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: Comparison of visual quality under fluctuating bandwidth on volumetric videos prepared by 4DGS [Wu et al [PITH_FULL_IMAGE:figures/full_fig_p024_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: Visual results of the frame 30 under 30Mbps. [PITH_FULL_IMAGE:figures/full_fig_p025_25.png] view at source ↗
read the original abstract

Volumetric video (VV) streaming enables real-time, immersive access to remote 3D environments, powering telepresence, ecological monitoring, and robotic teleoperation. These applications turn VV streaming into a real-time interface to remote physical environments, imposing new system-level demands for photorealistic scene representation, low-latency interaction, and robust performance under heterogeneous networks. 3D Gaussian Splatting (3DGS) has been widely used for real-time photorealistic rendering, offering superior visual quality and rendering performance, but it faces challenges due to bandwidth consumption. Furthermore, as the foundation of adaptive VV streaming, existing Levels of Detail (LoD) methods based on density are not well-suited to Gaussian representations, leading to visible gaps and severe quality degradation. Recent studies have also explored attribute compression techniques to reduce bandwidth consumption. Our preliminary studies reveal that aggressive attribute compression primarily causes color distortion, which can be effectively corrected in the rendered image using a reference image. Motivated by these findings, we propose a novel Color-Adaptive scheme for adaptive VV streaming that uses vector quantization (VQ) to establish LoDs and correct color distortions with low-resolution reference images. We further present CAGS, an adaptive VV streaming system compatible with diverse Gaussian representations, which integrates the Color-Adaptive scheme by rendering reference images on the streaming server and performing color restoration on the client. Extensive experiments on our prototype system demonstrate that CAGS outperforms the existing adaptive streaming systems in PSNR by 5$\sim$20 dB under fluctuating bandwidth, operates significantly faster than existing scalable Gaussian compression methods, and generalizes across different Gaussian representations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 3 minor

Summary. The paper proposes CAGS, an adaptive volumetric video streaming system based on 3D Gaussian Splatting. It introduces a Color-Adaptive scheme that uses vector quantization (VQ) to construct Levels of Detail (LoDs) and corrects color distortions from aggressive attribute compression via low-resolution reference images rendered on the server. The system is designed to be compatible with diverse Gaussian representations and is evaluated on a prototype, claiming 5-20 dB PSNR gains over existing adaptive streaming methods under fluctuating bandwidth, faster operation than scalable Gaussian compression approaches, and generalization across representations.

Significance. If the empirical claims hold, this work addresses a key practical bottleneck in real-time volumetric video by mitigating bandwidth and quality issues in 3DGS representations without relying on density-based LoD, which the authors argue is unsuitable. The color-correction insight from preliminary studies and the prototype validation across bandwidth conditions represent a concrete engineering contribution that could improve photorealism and latency in applications such as telepresence and robotic teleoperation. Explicit strengths include the claimed compatibility with multiple Gaussian variants and reported speedups.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (Experiments): The central performance claims (5~20 dB PSNR improvement, significant speedups, and generalization) are stated without accompanying details on the number of test scenes, exact bandwidth fluctuation profiles, baseline implementations, statistical significance, or error bars. This information is load-bearing for verifying the Color-Adaptive scheme's effectiveness and must be expanded to support the generalization assertion across Gaussian representations.
  2. [§3.2] §3.2 (Color-Adaptive scheme): The assumption that color distortion is the primary artifact from aggressive attribute compression and can be reliably corrected by a low-resolution reference image is central to the method, yet the paper does not appear to include an ablation isolating this correction's contribution versus the VQ-LoD construction alone. Without such quantification, it is difficult to assess whether the reported gains are attributable to the proposed color adaptation.
minor comments (3)
  1. [§2] §2 (Related Work): The discussion of prior attribute compression and LoD methods for Gaussians would benefit from explicit comparison tables or quantitative references to the specific PSNR/latency numbers reported in those works.
  2. [Figures and §4] Figure captions and §4: Several figures showing rendered results under different bandwidths lack scale bars or explicit PSNR annotations per frame, making visual assessment of the claimed quality gains harder to interpret.
  3. [§3] Notation: The definition of the VQ codebook size and the exact form of the color restoration function on the client side could be stated more formally (e.g., as an equation) to aid reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. The comments highlight important areas where additional details and analysis will strengthen the paper. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (Experiments): The central performance claims (5~20 dB PSNR improvement, significant speedups, and generalization) are stated without accompanying details on the number of test scenes, exact bandwidth fluctuation profiles, baseline implementations, statistical significance, or error bars. This information is load-bearing for verifying the Color-Adaptive scheme's effectiveness and must be expanded to support the generalization assertion across Gaussian representations.

    Authors: We agree that the experimental section would benefit from greater specificity to support the reported claims. In the revised manuscript, we will expand §4 to explicitly state the number of test scenes used (five representative volumetric video sequences), provide the exact bandwidth fluctuation profiles (including the real-world network traces employed), detail the baseline implementations (including how existing adaptive streaming and scalable Gaussian compression methods were reimplemented for fair comparison), and include error bars along with statistical significance testing (e.g., paired t-tests) for the PSNR and runtime results. These additions will also reinforce the generalization claims across Gaussian representations by reporting per-representation breakdowns. revision: yes

  2. Referee: [§3.2] §3.2 (Color-Adaptive scheme): The assumption that color distortion is the primary artifact from aggressive attribute compression and can be reliably corrected by a low-resolution reference image is central to the method, yet the paper does not appear to include an ablation isolating this correction's contribution versus the VQ-LoD construction alone. Without such quantification, it is difficult to assess whether the reported gains are attributable to the proposed color adaptation.

    Authors: We acknowledge that an explicit ablation would help isolate the contribution of the color-correction step. We will add a new ablation study (either as an extension to §4 or a dedicated subsection) that compares three configurations while holding the VQ-LoD construction fixed: (1) VQ-LoD without color correction, (2) VQ-LoD with the proposed low-resolution reference image correction, and (3) the full CAGS pipeline. This will quantify the incremental PSNR and perceptual gains attributable to color adaptation under varying bandwidth conditions. revision: yes

Circularity Check

0 steps flagged

No significant circularity; claims rest on independent prototype experiments

full rationale

The paper describes a Color-Adaptive scheme for volumetric video streaming motivated by preliminary observations on attribute compression effects, then validates performance via prototype experiments comparing PSNR gains, speed, and generalization against existing adaptive streaming and compression methods under fluctuating bandwidth. No derivation chain, equations, fitted parameters renamed as predictions, or load-bearing self-citations appear in the provided text. The central claims are externally falsifiable through the reported empirical comparisons rather than reducing to inputs by construction, making the argument self-contained against benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; the approach relies on standard vector quantization, 3D Gaussian Splatting rendering, and reference image correction without introducing new postulates.

pith-pipeline@v0.9.0 · 5637 in / 1174 out tokens · 47909 ms · 2026-05-12T04:07:04.170546+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

Reference graph

Works this paper leans on

73 extracted references · 73 canonical work pages

  1. [1]

    year 2024

    author person Vukasin Bozic , person Abdelaziz Djelouah , person Yang Zhang , person Radu Timofte , person Markus Gross , and person Christopher Schroers . year 2024 . Versatile Vision Foundation Model for Image and Video Colorization . In booktitle ACM SIGGRAPH 2024 Conference Papers ( series SIGGRAPH '24 ) . pages 1--11

    work page 2024

  2. [2]

    year 2025 a

    author person Yihang Chen , person Qianyi Wu , person Weiyao Lin , person Mehrtash Harandi , and person Jianfei Cai . year 2025 a . HAC : Hash-Grid Assisted Context for 3D Gaussian Splatting Compression . In booktitle Computer Vision -- ECCV 2024 , editor person Ale s Leonardis , person Elisa Ricci , person Stefan Roth , person Olga Russakovsky , person T...

    work page 2025

  3. [3]

    Hac++: To- wards 100x compression of 3d gaussian splatting,

    author person Yihang Chen , person Qianyi Wu , person Weiyao Lin , person Mehrtash Harandi , and person Jianfei Cai . year 2025 b . title HAC ++: Towards 100X Compression of 3D Gaussian Splatting . https://doi.org/10.48550/arXiv.2501.12255 doi: 10.48550/arXiv.2501.12255

    work page doi:10.48550/arxiv.2501.12255 2025

  4. [4]

    year 2024

    author person Xiaoyan Cong , person Yue Wu , person Qifeng Chen , and person Chenyang Lei . year 2024 . Automatic Controllable Colorization via Imagination . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 2609--2619

    work page 2024

  5. [5]

    year 2024

    author person Jiadi Cui , person Junming Cao , person Fuqiang Zhao , person Zhipeng He , person Yifan Chen , person Yuhui Zhong , person Lan Xu , person Yujiao Shi , person Yingliang Zhang , and person Jingyi Yu . year 2024 . LetsGo : Large-Scale Garage Modeling and Rendering via LiDAR-Assisted Gaussian Primitives . In booktitle SIGGRAPH Asia 2024 Confere...

    work page 2024

  6. [6]

    year 2023

    author person Koen de Koning , person Jeroen Broekhuijsen , person Ingolf K \"u hn , person Otso Ovaskainen , person Franziska Taubert , person Dag Endresen , person Dmitry Schigel , and person Volker Grimm . year 2023 . Digital twins: dynamic model-data fusion for ecology . journal Trends in ecology & evolution volume 38 , number 10 ( year 2023 ), pages 916--926

    work page 2023

  7. [7]

    year 2024

    author person Zhiwen Fan , person Kevin Wang , person Kairun Wen , person Zehao Zhu , person Dejia Xu , and person Zhangyang Wang . year 2024 . LightGaussian : Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS . journal Advances in Neural Information Processing Systems volume 37 ( year 2024 ), pages 140138--140158

    work page 2024

  8. [8]

    year 2004

    author person Christoph Fehn . year 2004 . Depth-image-based rendering (DIBR), compression, and transmission for a new approach on 3D-TV . In booktitle Stereoscopic Displays and Virtual Reality Systems XI , Vol. volume 5291 . International Society for Optics and Photonics, publisher SPIE , pages 93 -- 104

    work page 2004

  9. [9]

    year 2024

    author person Qiankun Gao , person Jiarui Meng , person Chengxiang Wen , person Jie Chen , and person Jian Zhang . year 2024 . HiCoM : Hierarchical Coherent Motion for Dynamic Streamable Scenes with 3D Gaussian Splatting . In booktitle The Thirty-eighth Annual Conference on Neural Information Processing Systems

    work page 2024

  10. [10]

    year 2021

    author person Danilo Gasques , person Janet G Johnson , person Tommy Sharkey , person Yuanyuan Feng , person Ru Wang , person Zhuoqun Robin Xu , person Enrique Zavala , person Yifei Zhang , person Wanze Xie , person Xinming Zhang , et al . year 2021 . Artemis: A collaborative mixed-reality system for immersive surgical telementoring . In booktitle Proceed...

    work page 2021

  11. [11]

    Gersho and person Y

    author person A. Gersho and person Y. Shoham . year 1984 . Hierarchical vector quantization of speech with dynamic codebook allocation . In booktitle ICASSP '84. IEEE International Conference on Acoustics, Speech, and Signal Processing , Vol. volume 9 . pages 416--419

    work page 1984

  12. [12]

    year 2024

    author person Sharath Girish , person Tianye Li , person Amrita Mazumdar , person Abhinav Shrivastava , person David Luebke , and person Shalini De Mello . year 2024 . QUEEN : QUantized Efficient ENcoding of Dynamic Gaussians for Streaming Free-viewpoint Videos . In booktitle The Thirty-eighth Annual Conference on Neural Information Processing Systems

    work page 2024

  13. [13]

    year 2017

    author person Google . year 2017 . title Draco 3D Graphics Compression . howpublished https://github.com/google/draco

    work page 2017

  14. [14]

    year 2023

    author person Yongjie Guan , person Xueyu Hou , person Nan Wu , person Bo Han , and person Tao Han . year 2023 . MetaStream : Live Volumetric Content Capture , Creation , Delivery , and Rendering in Real Time . In booktitle Proceedings of the 29th Annual International Conference on Mobile Computing and Networking . Number 29. pages 1--15

    work page 2023

  15. [15]

    year 2022

    author person Serhan G \"u l , person Cornelius Hellge , and person Peter Eisert . year 2022 . Latency Compensation Through Image Warping For Remote Rendering-Based Volumetric Video Streaming . In booktitle 2022 IEEE International Conference on Image Processing ( ICIP ) . pages 2026--2030

    work page 2022

  16. [16]

    year 2020

    author person Bo Han , person Yu Liu , and person Feng Qian . year 2020 . ViVo : Visibility-Aware Mobile Volumetric Video Streaming . In booktitle Proceedings of the 26th Annual International Conference on Mobile Computing and Networking . pages 1--13

    work page 2020

  17. [17]

    year 2024

    author person W Hazeleger , person JPM Aerts , person Peter Bauer , person MFP Bierkens , person Gustau Camps-Valls , person MM Dekker , person FJ Doblas-Reyes , person Veronika Eyring , person C Finkenauer , person Arthur Grundner , et al . year 2024 . Digital twins of the Earth with and for humans . journal Communications earth & environment volume 5 , ...

    work page 2024

  18. [18]

    year 2025

    author person Jonas Hein , person Nicola Cavalcanti , person Daniel Suter , person Lukas Zingg , person Fabio Carrillo , person Lilian Calvet , person Mazda Farshad , person Nassir Navab , person Marc Pollefeys , and person Philipp Fürnstahl . year 2025 . Next-generation surgical navigation: Marker-less multi-view 6DoF pose estimation of surgical instrume...

    work page 2025

  19. [19]

    year 2019

    author person Jozef Hladky , person Hans-Peter Seidel , and person Markus Steinberger . year 2019 . The camera offset space: real-time potentially visible set computations for streaming rendering . journal ACM Trans. Graph. volume 38 , number 6 ( year 2019 ), numpages 14 pages. 0730-0301

    work page 2019

  20. [20]

    year 2022

    author person Jozef Hladky , person Michael Stengel , person Nicholas Vining , person Bernhard Kerbl , person Hans-Peter Seidel , and person Markus Steinberger . year 2022 . QuadStream: A Quad-Based Scene Streaming Architecture for Novel Viewpoint Reconstruction . journal ACM Trans. Graph. volume 41 , number 6 ( year 2022 ), numpages 13 pages. 0730-0301

    work page 2022

  21. [21]

    year 2023

    author person Xudong Huang , person Wei Li , person Jie Hu , person Hanting Chen , and person Yunhe Wang . year 2023 . RefSR-NeRF: Towards High Fidelity and Super Resolution View Synthesis . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . pages 8244--8253

    work page 2023

  22. [22]

    year 2026

    author person Zhaohui Huang , person Cong Zhang , person Jianxin Shi , person Xiaoyi Fan , person Laizhong Cui , and person Jiangchuan Liu . year 2026 . ACPGS : Towards Bandwidth-Efficient Delivery of 3D Gaussian Splatting . In booktitle Proceedings of the 36th Workshop on Network and Operating System Support for Digital Audio and Video ( series NOSSDAV '...

    work page 2026

  23. [23]

    year 2025

    author person Yuheng Jiang , person Chengcheng Guo , person Yize Wu , person Yu Hong , person Shengkun Zhu , person Zhehao Shen , person Yingliang Zhang , person Shaohui Jiao , person Zhuo Su , person Lan Xu , person Marc Habermann , and person Christian Theobalt . year 2025 . Topology- Aware Optimization of Gaussian Primitives for Human-Centric Volumetri...

    work page 2025

  24. [24]

    Mathis , and person Amir Patel

    author person Daniel Joska , person Liam Clark , person Naoya Muramatsu , person Ricardo Jericevich , person Fred Nicolls , person Alexander Mathis , person Mackenzie W. Mathis , and person Amir Patel . year 2021 . title AcinoSet: A 3D Pose Estimation Dataset and Baseline Models for Cheetahs in the Wild . [arxiv] 2103.13282 [cs.CV]

    work page arXiv 2021

  25. [25]

    year 2023

    author person Bernhard Kerbl , person Georgios Kopanas , person Thomas Leimkuehler , and person George Drettakis . year 2023 . 3D Gaussian Splatting for Real-Time Radiance Field Rendering . journal ACM Transactions on Graphics volume 42 , number 4 ( year 2023 ), pages 139:1--139:14

    work page 2023

  26. [26]

    year 2024

    author person Bernhard Kerbl , person Andreas Meuleman , person Georgios Kopanas , person Michael Wimmer , person Alexandre Lanvin , and person George Drettakis . year 2024 . A Hierarchical 3D Gaussian Representation for Real-Time Rendering of Very Large Datasets . journal ACM Transactions on Graphics volume 44 , number 3 ( year 2024 )

    work page 2024

  27. [27]

    year 2025

    author person Gunjoong Kim , person Seonghoon Park , person Jeho Lee , person Chanyoung Jung , person Hyungchol Jun , and person Hojung Cha . year 2025 . Vega: Fully Immersive Mobile Volumetric Video Streaming with 3D Gaussian Splatting . In booktitle Proceedings of the 31st Annual International Conference on Mobile Computing and Networking . pages 1106--1120

    work page 2025

  28. [28]

    year 2017

    author person Christian Ledig , person Lucas Theis , person Ferenc Huszar , person Jose Caballero , person Andrew Cunningham , person Alejandro Acosta , person Andrew Aitken , person Alykhan Tejani , person Johannes Totz , person Zehan Wang , and person Wenzhe Shi . year 2017 . Photo- Realistic Single Image Super-Resolution Using a Generative Adversarial ...

    work page 2017

  29. [29]

    year 2024

    author person Joo Chan Lee , person Daniel Rho , person Xiangyu Sun , person Jong Hwan Ko , and person Eunbyung Park . year 2024 . Compact 3D Gaussian Representation for Radiance Field . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 21719--21728

    work page 2024

  30. [30]

    year 2025

    author person Hao Li , person Sicheng Li , person Xiang Gao , person Abudouaihati Batuer , person Lu Yu , and person Yiyi Liao . year 2025 . GIFStream : 4D Gaussian-based Immersive Video with Feature Stream . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition

    work page 2025

  31. [31]

    year 2024

    author person Jiyang Li , person Lechao Cheng , person Zhangye Wang , person Tingting Mu , and person Jingxuan He . year 2024 . LoopGaussian : Creating 3D Cinemagraph with Multi-view Images via Eulerian Motion Field . In booktitle Proceedings of the 32nd ACM International Conference on Multimedia ( series MM '24 ) . pages 476--485

    work page 2024

  32. [32]

    year 2022 a

    author person Lingzhi Li , person Zhen Shen , person Zhongshu Wang , person Li Shen , and person Ping Tan . year 2022 a . Streaming Radiance Fields for 3D Video Synthesis . journal Advances in Neural Information Processing Systems volume 35 ( year 2022 ), pages 13485--13498

    work page 2022

  33. [33]

    year 2022 b

    author person Tianye Li , person Mira Slavcheva , person Michael Zollh \"o fer , person Simon Green , person Christoph Lassner , person Changil Kim , person Tanner Schmidt , person Steven Lovegrove , person Michael Goesele , person Richard Newcombe , and person Zhaoyang Lv . year 2022 b . Neural 3D Video Synthesis From Multi-View Video . In booktitle Proc...

    work page 2022

  34. [34]

    year 2023

    author person Litian Liang , person Liuyu Bian , person Caiwei Xiao , person Jialin Zhang , person Linghao Chen , person Isabella Liu , person Fanbo Xiang , person Zhiao Huang , and person Hao Su . year 2023 . Robo360: a 3D omnispective multi-material robotic manipulation dataset . journal arXiv preprint arXiv:2312.06686 ( year 2023 )

    work page arXiv 2023

  35. [35]

    year 2023

    author person Junhua Liu , person Boxiang Zhu , person Fangxin Wang , person Yili Jin , person Wenyi Zhang , person Zihan Xu , and person Shuguang Cui . year 2023 . CaV3 : Cache-assisted Viewport Adaptive Volumetric Video Streaming . In booktitle 2023 IEEE Conference Virtual Reality and 3D User Interfaces ( VR ) . pages 173--183

    work page 2023

  36. [36]

    year 2024

    author person Xiangrui Liu , person Xinju Wu , person Pingping Zhang , person Shiqi Wang , person Zhu Li , and person Sam Kwong . year 2024 . CompGS : Efficient 3D Scene Representation via Compressed Gaussian Splatting . In booktitle Proceedings of the 32nd ACM International Conference on Multimedia ( series MM '24 ) . pages 2936--2944

    work page 2024

  37. [37]

    year 2025

    author person Edward Lu and person Anthony Rowe . year 2025 . QUASAR: Quad-based Adaptive Streaming And Rendering . journal ACM Trans. Graph. volume 44 , number 4 ( year 2025 ), numpages 18 pages. 0730-0301

    work page 2025

  38. [38]

    year 2024

    author person Tao Lu , person Mulin Yu , person Linning Xu , person Yuanbo Xiangli , person Limin Wang , person Dahua Lin , and person Bo Dai . year 2024 . Scaffold-Gs: Structured 3d Gaussians for View-Adaptive Rendering . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 20654--20664

    work page 2024

  39. [39]

    year 2024

    author person Jonathon Luiten , person Georgios Kopanas , person Bastian Leibe , and person Deva Ramanan . year 2024 . Dynamic 3D Gaussians: Tracking by Persistent Dynamic View Synthesis . In booktitle 3DV

    work page 2024

  40. [40]

    year 2022

    author person Zhengxiong Luo , person Yan Huang , person Shang Li , person Liang Wang , and person Tieniu Tan . year 2022 . Learning the Degradation Distribution for Blind Image Super-Resolution . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 6063--6072

    work page 2022

  41. [41]

    Srinivasan , person Matthew Tancik , person Jonathan T

    author person Ben Mildenhall , person Pratul P. Srinivasan , person Matthew Tancik , person Jonathan T. Barron , person Ravi Ramamoorthi , and person Ren Ng . year 2021 . NeRF : Representing Scenes as Neural Radiance Fields for View Synthesis . journal Commun. ACM volume 65 , number 1 ( year 2021 ), pages 99--106

    work page 2021

  42. [42]

    year 2020

    author person Arvind Narayanan , person Eman Ramadan , person Jason Carpenter , person Qingxu Liu , person Yu Liu , person Feng Qian , and person Zhi-Li Zhang . year 2020 . A First Look at Commercial 5G Performance on Smartphones . In booktitle Proceedings of The Web Conference 2020 ( series WWW '20 ) . pages 894--905

    work page 2020

  43. [43]

    year 2025

    author person Niantic Labs . year 2025 . title spz: File Format for 3D Gaussian Splats . howpublished https://github.com/nianticlabs/spz

    work page 2025

  44. [44]

    year 2024

    author person Panagiotis Papantonakis , person Georgios Kopanas , person Bernhard Kerbl , person Alexandre Lanvin , and person George Drettakis . year 2024 . Reducing the Memory Footprint of 3D Gaussian Splatting . journal Proceedings of the ACM on Computer Graphics and Interactive Techniques volume 7 , number 1 ( year 2024 ), pages 16:1--16:17

    work page 2024

  45. [45]

    year 2023

    author person Aras Pranckevičius . year 2023 . title Gaussian Splatting playground in Unity . howpublished https://github.com/aras-p/UnityGaussianSplatting

    work page 2023

  46. [46]

    year 2019

    author person Edgar Rojas-Mu \ n oz , person Maria Eugenia Cabrera , person Daniel Andersen , person Voicu Popescu , person Sherri Marley , person Brian Mullis , person Ben Zarzaur , and person Juan Wachs . year 2019 . Surgical telementoring without encumbrance: a comparative study of see-through augmented reality-based approaches . journal Annals of surg...

    work page 2019

  47. [47]

    year 2024

    author person Jianxin Shi , person Miao Zhang , person Linfeng Shen , person Jiangchuan Liu , person Yuan Zhang , person Lingjun Pu , and person Jingdong Xu . year 2024 . Towards Full-scene Volumetric Video Streaming via Spatially Layered Representation and NeRF Generation . In booktitle Proceedings of the 34th Edition of the Workshop on Network and Opera...

    work page 2024

  48. [48]

    year 2025

    author person Yuang Shi , person G \'e raldine Morin , person Simone Gasparini , and person Wei Tsang Ooi . year 2025 . LapisGS : Layered Progressive 3D Gaussian Splatting for Adaptive Streaming . In booktitle International Conference on 3D Vision 2025

    work page 2025

  49. [49]

    year 2025

    author person Yuan-Chun Sun , person Yuang Shi , person Cheng-Tse Lee , person Mufeng Zhu , person Wei Tsang Ooi , person Yao Liu , person Chun-Ying Huang , and person Cheng-Hsin Hsu . year 2025 . LTS : A DASH Streaming System for Dynamic Multi-Layer 3D Gaussian Splatting Scenes . In booktitle Proceedings of the 16th ACM Multimedia Systems Conference ( se...

    work page 2025

  50. [50]

    year 2022

    author person Towaki Takikawa , person Alex Evans , person Jonathan Tremblay , person Thomas M \"u ller , person Morgan McGuire , person Alec Jacobson , and person Sanja Fidler . year 2022 . Variable Bitrate Neural Fields . In booktitle Special Interest Group on Computer Graphics and Interactive Techniques Conference Proceedings . pages 1--9

    work page 2022

  51. [51]

    year 2018

    author person Fei Tao , person He Zhang , person Ang Liu , and person Andrew YC Nee . year 2018 . Digital twin in industry: State-of-the-art . journal IEEE Transactions on industrial informatics volume 15 , number 4 ( year 2018 ), pages 2405--2415

    work page 2018

  52. [52]

    year 2022

    author person Chen Wang , person Xian Wu , person Yuan-Chen Guo , person Song-Hai Zhang , person Yu-Wing Tai , and person Shi-Min Hu . year 2022 . NeRF-SR : High Quality Neural Radiance Fields Using Supersampling . In booktitle Proceedings of the 30th ACM International Conference on Multimedia ( series MM '22 ) . pages 6445--6454

    work page 2022

  53. [53]

    year 2024 a

    author person Penghao Wang , person Zhirui Zhang , person Liao Wang , person Kaixin Yao , person Siyuan Xie , person Jingyi Yu , person Minye Wu , and person Lan Xu . year 2024 a . V 3: Viewing Volumetric Videos on Mobiles via Streamable 2D Dynamic Gaussians . In booktitle SIGGRAPH Asia 2024 Conference Papers

    work page 2024

  54. [54]

    year 2024 b

    author person Yizong Wang , person Dong Zhao , person Huanhuan Zhang , person Teng Gao , person Zixuan Guo , person Chenghao Huang , and person Huadong Ma . year 2024 b . Bandwidth- Efficient Mobile Volumetric Video Streaming by Exploiting Inter-Frame Correlation . journal IEEE Transactions on Mobile Computing volume 23 , number 10 ( year 2024 ), pages 9410--9423

    work page 2024

  55. [55]

    year 2024

    author person Ole Wegen , person Willy Scheibel , person Matthias Trapp , person Rico Richter , and person Jurgen Dollner . year 2024 . A Survey on Non-photorealistic Rendering Approaches for Point Cloud Visualization . journal IEEE Transactions on Visualization and Computer Graphics ( year 2024 ), pages 1--20 . 1941-0506

    work page 2024

  56. [56]

    year 2024

    author person Guanjun Wu , person Taoran Yi , person Jiemin Fang , person Lingxi Xie , person Xiaopeng Zhang , person Wei Wei , person Wenyu Liu , person Qi Tian , and person Xinggang Wang . year 2024 . 4D Gaussian Splatting for Real-Time Dynamic Scene Rendering . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recogni...

    work page 2024

  57. [57]

    year 2023

    author person Jiangkai Wu , person Yu Guan , person Qi Mao , person Yong Cui , person Zongming Guo , and person Xinggong Zhang . year 2023 . ZGaming : Zero-Latency 3D Cloud Gaming by Image Prediction . In booktitle Proceedings of the ACM SIGCOMM 2023 Conference ( series ACM SIGCOMM '23 ) . pages 710--723

    work page 2023

  58. [58]

    year 2025

    author person Shuzhao Xie , person Jiahang Liu , person Weixiang Zhang , person Shijia Ge , person Sicheng Pan , person Chen Tang , person Yunpeng Bai , person Cong Zhang , person Xiaoyi Fan , and person Zhi Wang . year 2025 . SizeGS : Size-aware Compression of 3D Gaussian Splatting via Mixed Integer Programming . In booktitle Proceedings of the 33rd ACM ...

    work page doi:10.1145/3746027.3755370 2025

  59. [59]

    year 2024 a

    author person Jiawei Xu , person Zexin Fan , person Jian Yang , and person Jij Xie . year 2024 a . Grid4D : 4D Decomposed Hash Encoding for High-Fidelity Dynamic Gaussian Splatting . In booktitle Proceedings of the 38th International Conference on Neural Information Processing Systems ( series NIPS '24 , Vol. volume 37 ) . pages 123787--123811 . 979-8-3313-1438-5

    work page 2024

  60. [60]

    year 2024 b

    author person Zhen Xu , person Yinghao Xu , person Zhiyuan Yu , person Sida Peng , person Jiaming Sun , person Hujun Bao , and person Xiaowei Zhou . year 2024 b . Representing Long Volumetric Video with Temporal Gaussian Hierarchy . journal ACM Trans. Graph. volume 43 , number 6 ( year 2024 ), pages 171:1--171:18

    work page 2024

  61. [61]

    year 2024 b

    author person Jinbo Yan , person Rui Peng , person Luyang Tang , and person Ronggang Wang . year 2024 b . 4D Gaussian Splatting with Scale-aware Residual Field and Adaptive Optimization for Real-time Rendering of Temporally Complex Dynamic Scenes . In booktitle Proceedings of the 32nd ACM International Conference on Multimedia ( series MM '24 ) . pages 7871--7880

    work page 2024

  62. [62]

    year 2024 a

    author person Zhiwen Yan , person Weng Fei Low , person Yu Chen , and person Gim Hee Lee . year 2024 a . Multi- Scale 3D Gaussian Splatting for Anti-Aliased Rendering . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 20923--20931

    work page 2024

  63. [63]

    year 2025

    author person Daheng Yin , person Isaac Ding , person Yili Jin , person Jianxin Shi , and person Jiangchuan Liu . year 2025 . TrackerSplat : Exploiting Point Tracking for Fast and Robust Dynamic 3D Gaussians Reconstruction . In booktitle Proceedings of the SIGGRAPH Asia 2025 Conference Papers ( series SA Conference Papers '25 ) . pages 1--11

    work page 2025

  64. [64]

    year 2024

    author person Daheng Yin , person Jianxin Shi , person Miao Zhang , person Zhaowu Huang , person Jiangchuan Liu , and person Fang Dong . year 2024 . FSVFG : Towards Immersive Full-Scene Volumetric Video Streaming with Adaptive Feature Grid . In booktitle Proceedings of the 32nd ACM International Conference on Multimedia ( series MM '24 ) . pages 11089--11098

    work page 2024

  65. [65]

    year 2022

    author person Anlan Zhang , person Chendong Wang , person Bo Han , and person Feng Qian . year 2022 . YuZu : Neural-Enhanced Volumetric Video Streaming . In booktitle 19th USENIX Symposium on Networked Systems Design and Implementation ( NSDI 22) . pages 137--154

    work page 2022

  66. [66]

    year 2024

    author person Anlan Zhang , person Chendong Wang , person Yuming Hu , person Ahmad Hassan , person Zejun Zhang , person Bo Han , person Feng Qian , and person Shichang Xu . year 2024 . Habitus: Boosting Mobile Immersive Content Delivery through Full-body Pose Tracking and Multipath Networking . In booktitle 21st USENIX Symposium on Networked Systems Desig...

    work page 2024

  67. [67]

    Sander , person Lu Yuan , person Amine Bermak , and person Dong Chen

    author person Bo Zhang , person Mingming He , person Jing Liao , person Pedro V. Sander , person Lu Yuan , person Amine Bermak , and person Dong Chen . year 2019 . Deep Exemplar-Based Video Colorization . In booktitle Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition . pages 8052--8061

    work page 2019

  68. [68]

    year 2021

    author person Jiakai Zhang , person Xinhang Liu , person Xinyi Ye , person Fuqiang Zhao , person Yanshun Zhang , person Minye Wu , person Yingliang Zhang , person Lan Xu , and person Jingyi Yu . year 2021 . Editable Free-Viewpoint Video Using a Layered Neural Representation . journal ACM Transactions on Graphics volume 40 , number 4 ( year 2021 ), pages 1...

    work page 2021

  69. [69]

    year 2018

    author person Richard Zhang , person Phillip Isola , person Alexei A Efros , person Eli Shechtman , and person Oliver Wang . year 2018 . The Unreasonable Effectiveness of Deep Features as a Perceptual Metric . In booktitle CVPR

    work page 2018

  70. [70]

    year 2024

    author person Hongliang Zhong , person Jingbo Zhang , and person Jing Liao . year 2024 . VQ-NeRF : Neural Reflectance Decomposition and Editing With Vector Quantization . journal IEEE Transactions on Visualization and Computer Graphics volume 30 , number 9 ( year 2024 ), pages 6247--6260 . 1941-0506

    work page 2024

  71. [71]

    year 2024

    author person Zhi Zhou , person Junke Zhu , and person Zhangjin Huang . year 2024 . Gaussian Splatting with Neural Basis Extension . In booktitle Proceedings of the 32nd ACM International Conference on Multimedia ( series MM '24 ) . pages 6043--6052

    work page 2024

  72. [72]

    year 2025

    author person Mufeng Zhu , person Mingju Liu , person Cunxi Yu , person Cheng-Hsin Hsu , and person Yao Liu . year 2025 . SGSS : Streaming 6- DoF Navigation of Gaussian Splat Scenes . In booktitle Proceedings of the 16th ACM Multimedia Systems Conference ( series MMSys '25 ) . pages 46--56

    work page 2025

  73. [73]

    year 2024

    author person Walter Zimmer , person Gerhard Arya Wardana , person Suren Sritharan , person Xingcheng Zhou , person Rui Song , and person Alois C Knoll . year 2024 . Tumtraf v2x cooperative perception dataset . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . pages 22668--22677

    work page 2024