pith. sign in

arxiv: 2601.14821 · v1 · pith:P4UY5W6Xnew · submitted 2026-01-21 · 💻 cs.CV

POTR: Post-Training 3DGS Compression

Bert Ramlot , Martijn Courteaux , Peter Lambert , Glenn Van Wallendael This is my paper

Pith reviewed 2026-05-25 06:52 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D Gaussian Splattingpost-training compressionsplat pruninglighting coefficientsnovel view synthesisrate-distortion performanceinference acceleration
0
0 comments X

The pith

POTR introduces a post-training codec that prunes 3D Gaussian Splatting models to 2-4 times fewer splats and recomputes lighting coefficients to 97 percent sparsity without any retraining.

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

The paper presents POTR as a method to compress trained 3D Gaussian Splatting scenes for lower storage and faster rendering. Its core techniques are a modified rasterizer that evaluates the removal impact of every splat at once to enable precise pruning, plus a parallel recomputation step that sparsifies the lighting coefficients. These steps produce models that require less memory and render quicker while preserving image quality better than existing post-training compressors. The approach matters for applications that need real-time novel view synthesis from large scenes where retraining is impractical. Experiments show consistent gains in rate-distortion curves and inference speed even before any optional fine-tuning is applied.

Core claim

POTR shows that post-training compression of 3D Gaussian Splatting succeeds by using a modified rasterizer to compute every splat's individual removal effect simultaneously for pruning decisions, paired with a training-free recomputation of lighting coefficients that sharply increases their sparsity, yielding 2-4 times fewer splats, 1.5-2 times faster inference, and superior rate-distortion performance over prior post-training methods.

What carries the argument

A modified 3DGS rasterizer that simultaneously calculates the removal effect of each splat to drive pruning decisions, together with a parallel recomputation procedure that reduces lighting coefficient entropy.

If this is right

  • The resulting models contain 2-4 times fewer splats than those produced by other post-training pruning methods.
  • Inference runs 1.5-2 times faster than other compressed 3DGS models at comparable quality.
  • Lighting coefficient sparsity increases from roughly 70 percent to 97 percent with only minimal quality degradation.
  • Rate-distortion performance exceeds all tested post-training compression baselines even before fine-tuning.
  • A lightweight fine-tuning pass can be added to further improve the same metrics.

Where Pith is reading between the lines

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

  • The simultaneous-effect calculation could be adapted to other explicit scene representations that support fast rasterization if their splat or point contributions can be isolated similarly.
  • Resource-limited devices running real-time view synthesis would gain the most from the reported inference speedups.
  • The training-free coefficient step might combine with existing quantization or entropy coders to reach still lower bit rates.
  • Large pre-trained scene collections could be compressed on demand without access to the original training data or compute budget.

Load-bearing premise

The modified rasterizer must produce removal-effect values accurate enough that pruning decisions do not introduce visible quality loss in the final rendered images.

What would settle it

Run the same scenes through competing post-training pruners and through POTR's rasterizer-based pruner, then measure whether the final rendered PSNR or LPIPS at matched bit rates falls below the competitors when the rasterizer step is replaced by a simpler heuristic.

Figures

Figures reproduced from arXiv: 2601.14821 by Bert Ramlot, Glenn Van Wallendael, Martijn Courteaux, Peter Lambert.

Figure 1
Figure 1. Figure 1: Mollweide projection of the first 16 real spherical [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Simplified overview of POTR. (a) Splats are removed across multiple pruning iterations based on the change in the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Normalized cumulative importance of the highest [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The mapping function m(x) for a = 10. where c(x) and c˜k(x) denote a pixel’s color before and after removing the k-th splat, respectively. The pruning difference is crucial for determining the impact of each splat’s removal on an objective quality metric. In this initial work, the squared error is used due to its simplicity and relation to the PSNR. The difference in squared error, per pixel and color chan… view at source ↗
Figure 5
Figure 5. Figure 5: Schematic representation of the POTR-FT encoder. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 8
Figure 8. Figure 8: Quantitative comparison of post-training pruning [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: A blurred, top-down view of the Truck model shows [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: POTR’s encoding time for various models. ’Miscel [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: The impact of limiting the number of pruning [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗
read the original abstract

3D Gaussian Splatting (3DGS) has recently emerged as a promising contender to Neural Radiance Fields (NeRF) in 3D scene reconstruction and real-time novel view synthesis. 3DGS outperforms NeRF in training and inference speed but has substantially higher storage requirements. To remedy this downside, we propose POTR, a post-training 3DGS codec built on two novel techniques. First, POTR introduces a novel pruning approach that uses a modified 3DGS rasterizer to efficiently calculate every splat's individual removal effect simultaneously. This technique results in 2-4x fewer splats than other post-training pruning techniques and as a result also significantly accelerates inference with experiments demonstrating 1.5-2x faster inference than other compressed models. Second, we propose a novel method to recompute lighting coefficients, significantly reducing their entropy without using any form of training. Our fast and highly parallel approach especially increases AC lighting coefficient sparsity, with experiments demonstrating increases from 70% to 97%, with minimal loss in quality. Finally, we extend POTR with a simple fine-tuning scheme to further enhance pruning, inference, and rate-distortion performance. Experiments demonstrate that POTR, even without fine-tuning, consistently outperforms all other post-training compression techniques in both rate-distortion performance and inference speed.

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 / 1 minor

Summary. The manuscript introduces POTR, a post-training compression method for 3D Gaussian Splatting (3DGS). It proposes two core techniques: (1) a pruning approach that employs a modified 3DGS rasterizer to compute the individual removal effect of every splat simultaneously in one forward pass, yielding 2-4× fewer splats and 1.5-2× faster inference; (2) a training-free recomputation of lighting coefficients that increases AC coefficient sparsity from 70% to 97% with minimal quality loss. An optional fine-tuning stage is also described. The central experimental claim is that POTR (even without fine-tuning) outperforms all other post-training compression techniques in rate-distortion performance and inference speed.

Significance. If the pruning and coefficient-recomputation claims are validated with rigorous experiments, the work would provide a practical, training-free compression pipeline for 3DGS that directly addresses storage and speed bottlenecks in real-time novel-view synthesis. The parallel, one-pass pruning and entropy-reduction steps would be particularly useful for deployment scenarios where retraining is costly. No machine-checked proofs or open code are mentioned, but the emphasis on post-training methods without fine-tuning is a clear potential strength if the modified rasterizer is shown to be accurate.

major comments (2)
  1. [Abstract] Abstract (and method description): The headline claim that POTR 'consistently outperforms all other post-training compression techniques in both rate-distortion performance and inference speed' is stated without any quantitative results, dataset names, baseline descriptions, or error metrics. This prevents verification of the central experimental result.
  2. [Abstract / §3] Pruning approach (Abstract and §3): The modified 3DGS rasterizer is asserted to 'efficiently calculate every splat's individual removal effect simultaneously,' yet no equation, pseudocode, or derivation is supplied showing how marginal contributions are computed while correctly handling alpha blending, depth ordering, and covariance overlap. Without such detail or validation against sequential ablation ground truth, it is impossible to confirm that the reported 2-4× sparsity is achieved without introducing approximation errors that would degrade final image quality.
minor comments (1)
  1. [Abstract] The abstract repeatedly uses 'experiments demonstrate' without referencing specific tables or figures; cross-references to quantitative results should be added once the full experimental section is reviewed.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful comments, which highlight important areas for improvement in clarity and detail. We address each major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (and method description): The headline claim that POTR 'consistently outperforms all other post-training compression techniques in both rate-distortion performance and inference speed' is stated without any quantitative results, dataset names, baseline descriptions, or error metrics. This prevents verification of the central experimental result.

    Authors: We concur that the abstract's headline claim lacks supporting quantitative evidence, which is a valid concern for verifiability. In the revised manuscript, we will incorporate specific quantitative results into the abstract, including dataset names (e.g., Mip-NeRF 360), baseline methods, and metrics such as PSNR, SSIM, and inference FPS to substantiate the rate-distortion and speed improvements. revision: yes

  2. Referee: [Abstract / §3] Pruning approach (Abstract and §3): The modified 3DGS rasterizer is asserted to 'efficiently calculate every splat's individual removal effect simultaneously,' yet no equation, pseudocode, or derivation is supplied showing how marginal contributions are computed while correctly handling alpha blending, depth ordering, and covariance overlap. Without such detail or validation against sequential ablation ground truth, it is impossible to confirm that the reported 2-4× sparsity is achieved without introducing approximation errors that would degrade final image quality.

    Authors: This comment accurately points out the absence of technical details on the pruning mechanism. We will revise §3 to include the derivation, equations, and pseudocode for the simultaneous effect calculation, explicitly addressing alpha blending, depth ordering, and covariance overlaps. Furthermore, we will add a validation subsection comparing the parallel method to sequential ablation to demonstrate the accuracy of the approximation and its impact on quality. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper describes two algorithmic contributions—a modified rasterizer for simultaneous per-splat pruning effect scoring and a training-free lighting coefficient recomputation—followed by experimental comparisons. No equations, fitted parameters, or self-citations are shown that reduce any claimed performance gain to a definition, a renamed input, or a prior result by the same authors. The derivation chain consists of proposed procedures whose validity is asserted via external benchmarks rather than internal self-reference, satisfying the self-contained criterion.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated in the provided text.

pith-pipeline@v0.9.0 · 5774 in / 1050 out tokens · 36203 ms · 2026-05-25T06:52:02.288658+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. MesonGS++: Post-training Compression of 3D Gaussian Splatting with Hyperparameter Searching

    cs.CV 2026-04 unverdicted novelty 6.0

    MesonGS++ achieves over 34x compression of 3D Gaussian Splatting models with preserved or improved PSNR by using size-aware joint optimization of pruning and quantization hyperparameters via discrete sampling and 0-1 ...

  2. MesonGS++: Post-training Compression of 3D Gaussian Splatting with Hyperparameter Searching

    cs.CV 2026-04 unverdicted novelty 5.0

    MesonGS++ achieves over 34x compression of 3D Gaussian Splatting models post-training while preserving or exceeding original rendering quality through size-aware hyperparameter optimization.

Reference graph

Works this paper leans on

80 extracted references · 80 canonical work pages · cited by 1 Pith paper

  1. [1]

    Multiview video sequence analysis, compres- sion, and virtual viewpoint synthesis,

    R.-S. Wang and Y . Wang, “Multiview video sequence analysis, compres- sion, and virtual viewpoint synthesis,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 10, no. 3, pp. 397–410, 2000

    work page 2000

  2. [2]

    Stream-centric stereo matching and view synthesis: A high-speed approach on gpus,

    J. Lu, S. Rogmans, G. Lafruit, and F. Catthoor, “Stream-centric stereo matching and view synthesis: A high-speed approach on gpus,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 19, no. 11, pp. 1598–1611, 2009

    work page 2009

  3. [3]

    Fewarnet: An efficient few-shot view synthesis network based on trend regularization,

    C. Song, S. Wang, J. Wei, and Y . Zhao, “Fewarnet: An efficient few-shot view synthesis network based on trend regularization,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 10, pp. 9264–9280, 2024

    work page 2024

  4. [4]

    Multiview video coding using view interpolation and color correction,

    K. Yamamoto, M. Kitahara, H. Kimata, T. Yendo, T. Fujii, M. Tanimoto, S. Shimizu, K. Kamikura, and Y . Yashima, “Multiview video coding using view interpolation and color correction,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 17, no. 11, pp. 1436– 1449, 2007. 13

    work page 2007

  5. [5]

    Multiview plus depth video coding with temporal prediction view synthesis,

    A. I. Purica, E. G. Mora, B. Pesquet-Popescu, M. Cagnazzo, and B. Ionescu, “Multiview plus depth video coding with temporal prediction view synthesis,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 26, no. 2, pp. 360–374, 2016

    work page 2016

  6. [6]

    Uav-satellite view syn- thesis for cross-view geo-localization,

    X. Tian, J. Shao, D. Ouyang, and H. T. Shen, “Uav-satellite view syn- thesis for cross-view geo-localization,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 7, pp. 4804–4815, 2022

    work page 2022

  7. [7]

    Steered mixture-of-experts for light field images and video: Representation and coding,

    R. Verhack, T. Sikora, G. Van Wallendael, and P. Lambert, “Steered mixture-of-experts for light field images and video: Representation and coding,”IEEE Transactions on Multimedia, vol. 22, no. 3, pp. 579–593, 2020

    work page 2020

  8. [8]

    Opendibr: Open real-time depth-image-based renderer of light field videos for vr,

    J. Artois, M. Courteaux, G. Wallendael, and P. Lambert, “Opendibr: Open real-time depth-image-based renderer of light field videos for vr,” Multimedia Tools and Applications, vol. 83, pp. 25 797–25 815, 08 2023

    work page 2023

  9. [9]

    Dimensionality reduction for the real-time light-field view synthesis of kernel-based models,

    M. Courteaux, H. Mareen, B. Ramlot, P. Lambert, and G. Van Wal- lendael, “Dimensionality reduction for the real-time light-field view synthesis of kernel-based models,”Electronics, vol. 13, no. 20, 2024

    work page 2024

  10. [10]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” inECCV, 2020

    work page 2020

  11. [11]

    3d gaussian splatting for real-time radiance field rendering,

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering,”ACM Transactions on Graphics, vol. 42, no. 4, July 2023

    work page 2023

  12. [12]

    Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,

    J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, and P. P. Srinivasan, “Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,”ICCV, 2021

    work page 2021

  13. [13]

    Mip-nerf 360: Unbounded anti-aliased neural radiance fields,

    J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,”CVPR, 2022

    work page 2022

  14. [14]

    Zip-nerf: Anti-aliased grid-based neural radiance fields,

    ——, “Zip-nerf: Anti-aliased grid-based neural radiance fields,”ICCV, 2023

    work page 2023

  15. [15]

    A hierarchical 3d gaussian representation for real-time rendering of very large datasets,

    B. Kerbl, A. Meuleman, G. Kopanas, M. Wimmer, A. Lanvin, and G. Drettakis, “A hierarchical 3d gaussian representation for real-time rendering of very large datasets,”ACM Transactions on Graphics, vol. 43, no. 4, July 2024

    work page 2024

  16. [16]

    Citygaussian: Real-time high-quality large-scale scene rendering with gaussians,

    Y . Liu, H. Guan, C. Luo, L. Fan, J. Peng, and Z. Zhang, “Citygaussian: Real-time high-quality large-scale scene rendering with gaussians,” 2024

    work page 2024

  17. [17]

    4d gaussian splatting for real-time dynamic scene rendering,

    G. Wu, T. Yi, J. Fang, L. Xie, X. Zhang, W. Wei, W. Liu, Q. Tian, and W. Xinggang, “4d gaussian splatting for real-time dynamic scene rendering,”arXiv preprint arXiv:2310.08528, 2023

    work page arXiv 2023

  18. [18]

    Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis,

    J. Luiten, G. Kopanas, B. Leibe, and D. Ramanan, “Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis,” in3DV, 2024

    work page 2024

  19. [19]

    Gaussianspa: An

    Y . Zhang, W. Jia, W. Niu, and M. Yin, “Gaussianspa: An” optimizing- sparsifying” simplification framework for compact and high-quality 3d gaussian splatting,” inProceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 26 673–26 682

    work page 2025

  20. [20]

    Implicit gaussian splatting with efficient multi-level tri-plane representation,

    M. Wu and T. Tuytelaars, “Implicit gaussian splatting with efficient multi-level tri-plane representation,” 2024

    work page 2024

  21. [21]

    Compression of 3d gaussian splatting with optimized feature planes and standard video codecs,

    S. Lee, F. Shu, Y . Sanchez, T. Schierl, and C. Hellge, “Compression of 3d gaussian splatting with optimized feature planes and standard video codecs,”arXiv preprint arXiv:2501.03399, 2025

    work page arXiv 2025

  22. [22]

    Compact 3d scene representation via self-organizing gaussian grids,

    W. Morgenstern, F. Barthel, A. Hilsmann, and P. Eisert, “Compact 3d scene representation via self-organizing gaussian grids,”arXiv preprint arXiv:2312.13299, 2023

    work page arXiv 2023

  23. [23]

    Compgs: Efficient 3d scene representation via compressed gaussian splatting,

    X. Liu, X. Wu, P. Zhang, S. Wang, Z. Li, and S. Kwong, “Compgs: Efficient 3d scene representation via compressed gaussian splatting,” in Proceedings of the 32nd ACM International Conference on Multimedia, 2024, pp. 2936–2944

    work page 2024

  24. [24]

    Compgs++: Compressed gaussian splatting for static and dynamic scene representation,

    X. Liu, X. Wu, S. Wang, Z. Li, and S. Kwong, “Compgs++: Compressed gaussian splatting for static and dynamic scene representation,”arXiv preprint arXiv:2504.13022, 2025

    work page arXiv 2025

  25. [25]

    Hac: Hash-grid assisted context for 3d gaussian splatting compression,

    Y . Chen, Q. Wu, W. Lin, M. Harandi, and J. Cai, “Hac: Hash-grid assisted context for 3d gaussian splatting compression,” inEuropean Conference on Computer Vision, 2024

    work page 2024

  26. [26]

    Hac++: Towards 100x compression of 3d gaussian splatting,

    ——, “Hac++: Towards 100x compression of 3d gaussian splatting,” arXiv preprint arXiv:2501.12255, 2025

    work page arXiv 2025

  27. [27]

    Compact3d: Smaller and faster gaussian splatting with vector quanti- zation,

    K. Navaneet, K. P. Meibodi, S. A. Koohpayegani, and H. Pirsiavash, “Compact3d: Smaller and faster gaussian splatting with vector quanti- zation,”arXiv preprint arXiv:2311.18159, 2023

    work page arXiv 2023

  28. [28]

    End- to-end rate-distortion optimized 3d gaussian representation,

    H. Wang, H. Zhu, T. He, R. Feng, J. Deng, J. Bian, and Z. Chen, “End- to-end rate-distortion optimized 3d gaussian representation,” 2024

    work page 2024

  29. [29]

    Eagles: Efficient accelerated 3d gaussians with lightweight encodings,

    S. Girish, K. Gupta, and A. Shrivastava, “Eagles: Efficient accelerated 3d gaussians with lightweight encodings,” 2024

    work page 2024

  30. [30]

    Scaffold-gs: Structured 3d gaussians for view-adaptive rendering,

    T. Lu, M. Yu, L. Xu, Y . Xiangli, L. Wang, D. Lin, and B. Dai, “Scaffold-gs: Structured 3d gaussians for view-adaptive rendering,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 20 654–20 664

    work page 2024

  31. [31]

    Compact 3d gaussian representation for radiance field,

    J. C. Lee, D. Rho, X. Sun, J. H. Ko, and E. Park, “Compact 3d gaussian representation for radiance field,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 21 719–21 728

    work page 2024

  32. [32]

    F-3dgs: Factorized coordinates and representations for 3d gaussian splatting,

    X. Sun, J. C. Lee, D. Rho, J. H. Ko, U. Ali, and E. Park, “F-3dgs: Factorized coordinates and representations for 3d gaussian splatting,” arXiv preprint arXiv:2405.17083, 2024

    work page arXiv 2024

  33. [33]

    Reducing the memory footprint of 3d gaussian splatting,

    P. Papantonakis, G. Kopanas, B. Kerbl, A. Lanvin, and G. Drettakis, “Reducing the memory footprint of 3d gaussian splatting,”Proceedings of the ACM on Computer Graphics and Interactive Techniques, vol. 7, no. 1, May 2024

    work page 2024

  34. [34]

    Contextgs: Com- pact 3d gaussian splatting with anchor level context model,

    Y . Wang, Z. Li, L. Guo, W. Yang, A. Kot, and B. Wen, “Contextgs: Com- pact 3d gaussian splatting with anchor level context model,”Advances in neural information processing systems, vol. 37, pp. 51 532–51 551, 2024

    work page 2024

  35. [35]

    Locality-aware gaussian compression for fast and high-quality rendering,

    S. Shin, J. Park, and S. Cho, “Locality-aware gaussian compression for fast and high-quality rendering,”arXiv preprint arXiv:2501.05757, 2025

    work page arXiv 2025

  36. [36]

    Lightweight predictive 3d gaussian splats,

    J. Cao, V . Goel, C. Wang, A. Kag, J. Hu, S. Korolev, C. Jiang, S. Tulyakov, and J. Ren, “Lightweight predictive 3d gaussian splats,” arXiv preprint arXiv:2406.19434, 2024

    work page arXiv 2024

  37. [37]

    Mini-splatting: Representing scenes with a constrained number of gaussians,

    G. Fang and B. Wang, “Mini-splatting: Representing scenes with a constrained number of gaussians,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 165–181

    work page 2024

  38. [38]

    Lp-3dgs: Learning to prune 3d gaussian splatting,

    Z. Zhang, T. Song, Y . Lee, L. Yang, C. Peng, R. Chellappa, and D. Fan, “Lp-3dgs: Learning to prune 3d gaussian splatting,”Advances in Neural Information Processing Systems, vol. 37, pp. 122 434–122 457, 2024

    work page 2024

  39. [39]

    Speedy-splat: Fast 3d gaussian splatting with sparse pixels and sparse primitives,

    A. Hanson, A. Tu, G. Lin, V . Singla, M. Zwicker, and T. Goldstein, “Speedy-splat: Fast 3d gaussian splatting with sparse pixels and sparse primitives,” inProceedings of the Computer Vision and Pattern Recog- nition Conference, 2025, pp. 21 537–21 546

    work page 2025

  40. [40]

    Radsplat: Radiance field-informed gaussian splatting for robust real- time rendering with 900+ fps,

    M. Niemeyer, F. Manhardt, M.-J. Rakotosaona, M. Oechsle, D. Duck- worth, R. Gosula, K. Tateno, J. Bates, D. Kaeser, and F. Tombari, “Radsplat: Radiance field-informed gaussian splatting for robust real- time rendering with 900+ fps,” in2025 International Conference on 3D Vision (3DV). IEEE, 2025, pp. 134–144

    work page 2025

  41. [41]

    Mini-splatting2: Building 360 scenes within minutes via aggressive gaussian densification,

    G. Fang and B. Wang, “Mini-splatting2: Building 360 scenes within minutes via aggressive gaussian densification,”arXiv preprint arXiv:2411.12788, 2024

    work page arXiv 2024

  42. [42]

    Cat-3dgs: A context-adaptive triplane ap- proach to rate-distortion-optimized 3dgs compression,

    Y .-T. Zhan, C.-Y . Ho, H. Yang, Y .-H. Chen, J. C. Chiang, Y .-L. Liu, and W.-H. Peng, “Cat-3dgs: A context-adaptive triplane ap- proach to rate-distortion-optimized 3dgs compression,”arXiv preprint arXiv:2503.00357, 2025

    work page arXiv 2025

  43. [43]

    Pcgs: Progressive compression of 3d gaussian splatting,

    Y . Chen, M. Li, Q. Wu, W. Lin, M. Harandi, and J. Cai, “Pcgs: Progressive compression of 3d gaussian splatting,”arXiv preprint arXiv:2503.08511, 2025

    work page arXiv 2025

  44. [44]

    Megs2: Memory-efficient gaussian splatting via spherical gaussians and unified pruning,

    J. Chen, Y . Chen, Y . Zou, Y . Huang, P. Wang, Y . Liu, Y . Sun, and W. Wang, “Megs2: Memory-efficient gaussian splatting via spherical gaussians and unified pruning,”arXiv preprint arXiv:2509.07021, 2025

    work page arXiv 2025

  45. [45]

    Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps,

    Z. Fan, K. Wang, K. Wen, Z. Zhu, D. Xu, and Z. Wang, “Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps,” 2023

    work page 2023

  46. [46]

    Mesongs: Post-training compression of 3d gaussians via efficient attribute trans- formation,

    S. Xie, W. Zhang, C. Tang, Y . Bai, R. Lu, S. Ge, and Z. Wang, “Mesongs: Post-training compression of 3d gaussians via efficient attribute trans- formation,” inEuropean Conference on Computer Vision. Springer, 2024

    work page 2024

  47. [47]

    Compressed 3d gaussian splatting for accelerated novel view synthesis,

    S. Niedermayr, J. Stumpfegger, and R. Westermann, “Compressed 3d gaussian splatting for accelerated novel view synthesis,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR), June 2024, pp. 10 349–10 358

    work page 2024

  48. [48]

    Safeguardgs: 3d gaussian primitive pruning while avoiding catastrophic scene destruction,

    Y . Lee, Z. Zhang, and D. Fan, “Safeguardgs: 3d gaussian primitive pruning while avoiding catastrophic scene destruction,”arXiv preprint arXiv:2405.17793, 2024

    work page arXiv 2024

  49. [49]

    Pup 3d-gs: Principled uncertainty pruning for 3d gaussian splatting,

    A. Hanson, A. Tu, V . Singla, M. Jayawardhana, M. Zwicker, and T. Goldstein, “Pup 3d-gs: Principled uncertainty pruning for 3d gaussian splatting,” inProceedings of the Computer Vision and Pattern Recogni- tion Conference, 2025, pp. 5949–5958

    work page 2025

  50. [50]

    Entropygs: An efficient entropy coding on 3d gaussian splatting,

    Y . Huang, J. Pang, F. Zhu, and D. Tian, “Entropygs: An efficient entropy coding on 3d gaussian splatting,”arXiv preprint arXiv:2508.10227, 2025

    work page arXiv 2025

  51. [51]

    Efficientgs: Streamlining gaussian splatting for large-scale high-resolution scene representation,

    W. Liu, T. Guan, B. Zhu, L. Xu, Z. Song, D. Li, Y . Wang, and W. Yang, “Efficientgs: Streamlining gaussian splatting for large-scale high-resolution scene representation,”IEEE MultiMedia, 2025. 14

    work page 2025

  52. [52]

    Instant neural graphics primitives with a multiresolution hash encoding,

    T. M ¨uller, A. Evans, C. Schied, and A. Keller, “Instant neural graphics primitives with a multiresolution hash encoding,”ACM Trans. Graph., vol. 41, no. 4, pp. 102:1–102:15, Jul. 2022

    work page 2022

  53. [53]

    Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps,

    C. Reiser, S. Peng, Y . Liao, and A. Geiger, “Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps,” inInternational Conference on Computer Vision (ICCV), 2021

    work page 2021

  54. [54]

    Compressing explicit voxel grid representations: Fast nerfs become also small,

    C. L. Deng and E. Tartaglione, “Compressing explicit voxel grid representations: Fast nerfs become also small,” inProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), January 2023, pp. 1236–1245

    work page 2023

  55. [55]

    Compressing volumetric radiance fields to 1 mb,

    L. Li, Z. Shen, Z. Wang, L. Shen, and L. Bo, “Compressing volumetric radiance fields to 1 mb,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2023, pp. 4222–4231

    work page 2023

  56. [56]

    Tinynerf: to- wards 100× compression of voxel radiance fields,

    T. Zhao, J. Chen, C. Leng, and J. Cheng, “Tinynerf: to- wards 100× compression of voxel radiance fields,” inAAAI, ser. AAAI’23/IAAI’23/EAAI’23. AAAI Press, 2023

    work page 2023

  57. [57]

    Masked wavelet representation for compact neural radiance fields,

    D. Rho, B. Lee, S. Nam, J. C. Lee, J. H. Ko, and E. Park, “Masked wavelet representation for compact neural radiance fields,” inProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 20 680–20 690

    work page 2023

  58. [58]

    Nerfcodec: Neural feature compression meets neural radiance fields for memory-efficient scene representation,

    S. Li, H. Li, Y . Liao, and L. Yu, “Nerfcodec: Neural feature compression meets neural radiance fields for memory-efficient scene representation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 274–21 283

    work page 2024

  59. [59]

    Neural nerf compression,

    T. Pham and S. Mandt, “Neural nerf compression,” inProceedings of the 41st International Conference on Machine Learning, 2024, pp. 40 592– 40 610

    work page 2024

  60. [60]

    Ecrf: Entropy- constrained neural radiance fields compression with frequency domain optimization,

    S. Lee, F. Shu, Y . Sanchez, T. Schierl, and C. Hellge, “Ecrf: Entropy- constrained neural radiance fields compression with frequency domain optimization,” in2024 IEEE 26th International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2024, pp. 1–6

    work page 2024

  61. [61]

    Neural residual radiance fields for streamably free-viewpoint videos,

    L. Wang, Q. Hu, Q. He, Z. Wang, J. Yu, T. Tuytelaars, L. Xu, and M. Wu, “Neural residual radiance fields for streamably free-viewpoint videos,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 76–87

    work page 2023

  62. [62]

    Variable bitrate neural fields,

    T. Takikawa, A. Evans, J. Tremblay, T. M ¨uller, M. McGuire, A. Jacob- son, and S. Fidler, “Variable bitrate neural fields,” inACM SIGGRAPH 2022 Conference Proceedings, ser. SIGGRAPH ’22. New York, NY , USA: Association for Computing Machinery, 2022

    work page 2022

  63. [63]

    Binary radiance fields,

    S. Shin and J. Park, “Binary radiance fields,” inProceedings of the 37th International Conference on Neural Information Processing Systems, ser. NIPS ’23. Red Hook, NY , USA: Curran Associates Inc., 2024

    work page 2024

  64. [64]

    How far can we compress instant-ngp-based nerf?

    Y . Chen, Q. Wu, M. Harandi, and J. Cai, “How far can we compress instant-ngp-based nerf?” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2024, pp. 20 321–20 330

    work page 2024

  65. [65]

    Local geometry-based intra prediction for octree-structured geometry coding of point clouds,

    Z. Wang, S. Wan, and L. Wei, “Local geometry-based intra prediction for octree-structured geometry coding of point clouds,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 2, pp. 886– 896, 2023

    work page 2023

  66. [66]

    V oxelcontext-net: An octree based frame- work for point cloud compression,

    Z. Que, G. Lu, and D. Xu, “V oxelcontext-net: An octree based frame- work for point cloud compression,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2021, pp. 6042–6051

    work page 2021

  67. [67]

    Octsqueeze: Octree-structured entropy model for lidar compression,

    L. Huang, S. Wang, K. Wong, J. Liu, and R. Urtasun, “Octsqueeze: Octree-structured entropy model for lidar compression,” in2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 1310–1320

    work page 2020

  68. [68]

    3d point cloud compression: A survey,

    C. Cao, M. Preda, and T. Zaharia, “3d point cloud compression: A survey,” inProceedings of the 24th International Conference on 3D Web Technology, ser. Web3D ’19. New York, NY , USA: Association for Computing Machinery, 2019, p. 1–9

    work page 2019

  69. [69]

    Compression of 3d point clouds using a region-adaptive hierarchical transform,

    R. L. de Queiroz and P. A. Chou, “Compression of 3d point clouds using a region-adaptive hierarchical transform,”IEEE Transactions on Image Processing, vol. 25, no. 8, pp. 3947–3956, 2016

    work page 2016

  70. [70]

    Octree-Based Progres- sive Geometry Coding of Point Clouds,

    Y . Huang, J. Peng, C.-C. J. Kuo, and M. Gopi, “Octree-Based Progres- sive Geometry Coding of Point Clouds,” inSymposium on Point-Based Graphics, M. Botsch, B. Chen, M. Pauly, and M. Zwicker, Eds. The Eurographics Association, 2006

    work page 2006

  71. [71]

    Hierarchical segmentation based point cloud attribute compression,

    K. Zhang, W. Zhu, and Y . Xu, “Hierarchical segmentation based point cloud attribute compression,” in2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018, pp. 3131– 3135

    work page 2018

  72. [72]

    Emerging mpeg standards for point cloud compression,

    S. Schwarz, M. Preda, V . Baroncini, M. Budagavi, P. Cesar, P. A. Chou, R. A. Cohen, M. Krivoku ´ca, S. Lasserre, Z. Li, J. Llach, K. Mammou, R. Mekuria, O. Nakagami, E. Siahaan, A. Tabatabai, A. M. Tourapis, and V . Zakharchenko, “Emerging mpeg standards for point cloud compression,”IEEE Journal on Emerging and Selected Topics in Circuits and Systems, vo...

    work page 2019

  73. [73]

    3dgs.zip: A survey on 3d gaussian splatting compres- sion methods,

    M. T. Bagdasarian, P. Knoll, F. Barthel, A. Hilsmann, P. Eisert, and W. Morgenstern, “3dgs.zip: A survey on 3d gaussian splatting compres- sion methods,” 2024

    work page 2024

  74. [74]

    Structure-from-motion revisited,

    J. L. Sch ¨onberger and J.-M. Frahm, “Structure-from-motion revisited,” inConference on Computer Vision and Pattern Recognition (CVPR), 2016

    work page 2016

  75. [75]

    Deep blending for free-viewpoint image-based rendering,

    P. Hedman, J. Philip, T. Price, J.-M. Frahm, G. Drettakis, and G. Bros- tow, “Deep blending for free-viewpoint image-based rendering,”ACM Transactions on Graphics (SIGGRAPH Asia Conference Proceedings), vol. 37, no. 6, November 2018

    work page 2018

  76. [76]

    Tanks and temples: Benchmarking large-scale scene reconstruction,

    A. Knapitsch, J. Park, Q.-Y . Zhou, and V . Koltun, “Tanks and temples: Benchmarking large-scale scene reconstruction,”ACM Transactions on Graphics, vol. 36, no. 4, 2017

    work page 2017

  77. [77]

    Image quality assess- ment: from error visibility to structural similarity,

    Z. Wang, A. Bovik, H. Sheikh, and E. Simoncelli, “Image quality assess- ment: from error visibility to structural similarity,”IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600–612, 2004

    work page 2004

  78. [78]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, 2018, pp. 586–595

    work page 2018

  79. [79]

    Georgs: Geomet- ric regularization for real-time novel view synthesis from sparse inputs,

    Z. Liu, J. Su, G. Cai, Y . Chen, B. Zeng, and Z. Wang, “Georgs: Geomet- ric regularization for real-time novel view synthesis from sparse inputs,” IEEE Transactions on Circuits and Systems for Video Technology, pp. 1–1, 2024

    work page 2024

  80. [80]

    Dngaussian: Optimizing sparse-view 3d gaussian radiance fields with global-local depth normalization,

    J. Li, J. Zhang, X. Bai, J. Zheng, X. Ning, J. Zhou, and L. Gu, “Dngaussian: Optimizing sparse-view 3d gaussian radiance fields with global-local depth normalization,”arXiv preprint arXiv:2403.06912, 2024. Bert Ramlotreceived the M.Sc. degree in com- puter science engineering from Ghent University, Belgium, in 2024, where he is currently pursuing the Ph.D...

    work page arXiv 2024