Pith. sign in

REVIEW 5 major objections 5 minor 6 cited by

HEMGS: A Hybrid Entropy Model for 3D Gaussian Splatting Data Compression

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that a hybrid entropy model compresses 3D Gaussian Splatting scenes to about 40% smaller files while maintaining rendering quality, using a single model that covers many bitrates via a rate hyperparameter.

desk verdict A useful incremental advance in 3DGS compression with a plausible new architecture, but the headline 40% saving is computed at lower rendering quality and needs a matched-quality comparison before it should be quoted. read the letter →

arxiv 2411.18473 v2 pith:OK5YOJQZ submitted 2024-11-27 cs.CV

Lei Liu , Zhenghao Chen , Wei Jiang , Wei Wang , Dong Xu This is my paper · ORCID
classification cs.CV
keywords 3DGaussianSplattingcompressionentropymodelvariable-ratehyperpriorautoregressivescene-agnosticpriorlossy-losslesscoding
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

3D Gaussian Splatting is a popular way to represent a scene as a cloud of tiny colored ellipsoids, but the resulting files are heavy. This paper claims that a hybrid entropy model, HEMGS, can shrink those files by about 40% compared with the current best baseline, HAC, while rendering the same image quality. The model works on anchor points—control points that summarize groups of Gaussians—and compresses each anchor attribute with a mix of lossy quantization and lossless entropy coding. A single trained model covers many compression rates, selected by a hyperparameter λ, so users can pick the storage-quality trade-off without retraining. If the claim holds, storing and streaming 3D scenes becomes much cheaper.

What carries the argument

The load-bearing mechanism is HEMGS, a hybrid entropy model with three cooperating parts that run per anchor attribute. First, the variable-rate predictor takes the rate hyperparameter λ, broadcasts it to the feature dimension, and outputs a learned quantization step that controls how coarsely the attribute is quantized. Second, the hyperprior network computes a prior feature from already-compressed location and attribute data, using a frozen scene-agnostic extractor (PointNet++) plus a scene-specific hash-grid trained on the current scene. Third, the autoregressive network selects context elements adaptively: with a maximum receptive field of 25×25×25 voxels, it retains all anchors in sparse areas but only the nearest 20 in dense areas. The prior and context features are combined to estimate the Gaussian distribution of each coding element, which is then fed to an arithmetic coder. This design is what lets HEMGS do lossy rate control and lossless entropy coding in one pass.

What would settle it

Compress a synthetic scene containing two spatially separated clusters of identical anchors whose attributes are correlated with each other's cluster. If the nearest-n context cannot exploit that cross-cluster redundancy, the losslessly encoded size will be no smaller than a no-context baseline, showing that the adaptive context selection fails on non-local dependencies.

Watch

Extended reading notes

Core claim

The paper's central claim is that a single neural network, combining a variable-rate predictor, a hyperprior network, and an autoregressive network, can outperform specialized 3DGS compression methods on all four benchmarks. The variable-rate predictor produces a learned quantization step from a user-specified λ, replacing the need for multiple rate-specific models. The hyperprior network fuses a scene-agnostic feature (from a frozen pre-trained 3D feature extractor, PointNet++) with a scene-specific hash-grid prior, while the autoregressive network adapts its context window to local anchor density. Together these produce accurate per-element probability estimates for the arithmetic coder, yielding roughly 40% average storage reduction over HAC at equal PSNR, and a single model that spans 16 rate points.

Load-bearing premise

The compression gain assumes that a raster-order traversal of anchors, plus the nearest already-encoded anchors, captures the statistical dependencies between attribute values; if the important dependencies are long-range or non-local, the lossless entropy coding improvement would fail for such scenes.

Editorial extensions

If this is right

  • With one trained model, HEMGS can produce 16 rate-distortion points by varying λ, so applications can switch storage budgets without retraining or storing multiple models.
  • On the four benchmark datasets, HEMGS stores scenes in about 40% less space than HAC at matched PSNR; on Mip-NeRF360 it reports BDBR savings of 52.52% over HAC and 66.47% over Context-GS.
  • The autoregressive network contributes a 16.10% storage saving on the playroom scene (Table 3), and the scene-agnostic prior adds another 6.95% saving, showing both components pull their weight.
  • Progressive coding that compresses locations, then local features, then scaling and offset uses previously compressed data as priors, cutting offset storage by 34% relative to HAC (Table 5).

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The adaptive context selection (all anchors in sparse regions, nearest 20 in dense ones) is a natural candidate for other sparse 3D data such as point clouds or LiDAR scans, where density varies similarly and fixed receptive fields are inefficient.
  • Because the scene-agnostic network is frozen and only the scene-specific hash-grid is overfit, HEMGS may compress a new scene with less per-scene optimization than methods that retrain the whole entropy model, which could speed deployment.
  • A worthwhile test is whether the variable-rate predictor's learned quantization step transfers to other neural compression domains (images, video, or point clouds), where rate-conditioned retraining is currently the norm; the same λ-to-step mechanism could replace that.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes HEMGS, a hybrid entropy model for compressing anchor-based 3D Gaussian Splatting (3DGS) data. The framework combines (i) a variable-rate predictor that takes a user-specified lambda as input to produce a learned quantization step, enabling multiple rate-distortion trade-offs with a single model; (ii) a hyperprior network that fuses scene-agnostic features from a pretrained PointNet++ with scene-specific features from a hash grid; and (iii) an autoregressive network with an adaptive context selection mechanism that chooses the nearest n already-encoded anchors within a large receptive field. These components are integrated into a hybrid lossy-lossless compression pipeline that first quantizes anchor locations, local features, scaling, and offsets, then entropy-codes the quantized values with arithmetic coding. Experiments on Synthetic-NeRF, Tank&Temples, Mip-NeRF360, and DeepBlending report about a 40% average size reduction over the HAC baseline while maintaining rendering quality, as well as variable-rate operation with a single trained model.

Significance. If the claimed results hold, HEMGS would be a meaningful advance in 3DGS storage and transmission: a single model covering multiple rate points is practically valuable, and the reported size reductions over HAC are substantial. The paper has clear strengths: it reports actual compressed file sizes from arithmetic coding rather than cross-entropy estimates; it includes ablations isolating the scene-agnostic network, the autoregressive network, and the adaptive context selection; and it demonstrates variable-rate behavior with a single model. However, the central quantitative claim is not yet fully supported by the evidence in the main text. The headline 40% savings is computed at lower rendering quality in several datasets, the BD-rate evaluation covers only two of the four benchmarks, and the adaptive context model is validated on a single scene. The significance is therefore conditional on a more complete, quality-matched evaluation.

major comments (5)
  1. [Section 4.3, Table 1] The claim of 'about a 40% average reduction in size while maintaining rendering quality' is not directly supported by Table 1. The low-rate rows achieve the largest size reductions (e.g., 42.8% on Mip-NeRF360, 45.5% on Tank&Temples, 42.2% on DeepBlending, 36.6% on Synthetic-NeRF), but at those operating points PSNR, SSIM, or LPIPS are worse than HAC on most datasets (e.g., Mip-NeRF360 PSNR 27.68 vs 27.77; DeepBlending PSNR 30.24 vs 30.34; Synthetic-NeRF PSNR 33.33 vs 33.71). The high-rate rows are competitive or better in some cases, but the size reductions are much smaller (12.9%–33.6%). Thus the 40% average appears to be computed at unmatched quality, not at equal distortion. The authors should either report BD-rate (or BD-PSNR) over all four datasets or clearly state the distortion level at which each size reduction is measured.
  2. [Section 4.3, Table 2] The BDBR results are reported for only two of the four benchmarks (Mip-NeRF360 and DeepBlending). The abstract claims an average storage saving of approximately 40% across all benchmarks, but without equal-quality comparisons for Tank&Temples and Synthetic-NeRF, the headline number is not verifiable from the presented data. Moreover, the two reported BDBR values for HAC differ substantially (52.52% vs 82.85%), and no per-scene RD data or uncertainty estimates are given, so the robustness of the average is unclear.
  3. [Section 3.3.3 and Table 4] The adaptive context selection algorithm, a core component of the autoregressive network, is evaluated only on the 'playroom' scene from DeepBlending. The storage difference between 'Ours' and 'Ours w/o ACS' is 0.19 MB on that single scene, and the generalization of the chosen threshold n=20 and receptive field 25x25x25 to other scenes and datasets is not demonstrated. Please provide ablations on at least one additional scene per dataset, or otherwise justify the universality of these hyperparameters.
  4. [Section 4.3, Figure 3] The rate-distortion curves omit the Synthetic-NeRF dataset, and each curve contains only two operating points per method. Since the paper claims state-of-the-art compression across four benchmarks, the RD evidence is incomplete. The variable-rate model can produce at least 16 rate points (as shown in Figure 4 for Tank&Temples); full RD curves for all four datasets with more rate points should be provided to substantiate the claim.
  5. [Section 4.2 and Table 1] The mapping between the four predefined lambda values (1e-3, 2e-3, 3e-3, 4e-3) and the 'low-rate' and 'high-rate' rows in Table 1 is not reported. Without this mapping, the reader cannot reproduce the results or understand how the size-quality trade-off is controlled. Please specify which lambda corresponds to each reported row and, if only two of the four lambda values are used for the main comparisons, explain why the other two are omitted.
minor comments (5)
  1. [Section 3.2] The description of location compression says the quantized location is 'directly losslessly encoded into a bit-stream'; please clarify the entropy coding scheme used for the location (e.g., fixed-rate arithmetic coding or a learned distribution).
  2. [Section 3.3.2] Figure 2(c) marks the scene-agnostic network as 'Frozen', but the text says it introduces 'no additional storage overhead'; please state explicitly whether the PointNet++ weights are frozen and whether any fine-tuning occurs.
  3. [Section 3.4] In Eq. (2), lambda is both the rate-distortion trade-off weight and an input to the variable-rate predictor. Please use separate symbols for these two roles to avoid ambiguity.
  4. [Table 5] The column header 'Fidlity' is misspelled; it should be 'Fidelity'.
  5. [Throughout] The paper references supplementary materials for per-scene results, the definition of Lanchor, and other implementation details, but the arXiv version does not include a supplementary file. Please ensure the supplementary material is publicly available at the time of publication.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the reported sizes come from actual arithmetic coding after optimizing a cross-entropy estimate, and no load-bearing step reduces to a fitted parameter or self-citation.

full rationale

HEMGS is an empirical compression system rather than a mathematical derivation, and its central claims are not forced by construction. The training objective (Eq. 2) minimizes a cross-entropy estimate of storage via Lanchor, but the reported sizes in Tables 1, 3, 4, and 5 are measured compressed sizes obtained after arithmetic encoding/decoding, so the bitrate results are not automatically equal to the training loss. The variable-rate predictor takes λ as a conditioning input and produces a quantization step; this is a parameterized mapping, not a fitted quantity being renamed as a prediction. The autoregressive and hyperprior networks are learned modules whose ablation is measured on held-out compressed size (Table 3), so their contribution is an empirical claim rather than a self-definitional one. The paper cites several prior works by its own authors, but these citations appear in related-work context and are not used to justify the central compression claim, to import a uniqueness theorem, or to smuggle in an ansatz. The adaptive context selection and scene-agnostic/scene-specific priors are modeling choices with stated assumptions, and their limitations would be about generality or correctness, not circularity. The approximately 40% storage reduction may be based on unmatched rate points rather than equal-distortion comparisons, but that is a benchmarking-quality concern outside the scope of circularity analysis.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method relies on standard information theory, the Scaffold-GS anchor representation, and a handful of hand-chosen hyperparameters. No new physical entities are introduced. The most fragile assumption is the transferability of a frozen PointNet++ to 3DGS anchor data, which is only ablated on one scene.

free parameters (4)
  • Lambda set {1e-3, 2e-3, 3e-3, 4e-3} = 1e-3 to 4e-3
    Predefined rate-control hyperparameters input to the variable-rate predictor and used as Lagrangian weights in the loss; chosen by hand, not fitted to data.
  • Autoregressive context threshold n=20 = 20
    Maximum number of nearest encoded elements used as context in dense areas; chosen by hand and not swept.
  • Receptive field size 25x25x25 voxels = 25x25x25
    Maximum context window for the autoregressive model; chosen by hand and not swept.
  • Anchor location quantization bit depth = 16 bits
    Fixed quantization for anchor locations; standard practice but a free choice.
assumptions (4)
  • standard math Arithmetic coding of a symbol stream using the learned distribution p achieves the cross-entropy lower bound exactly (asymptotically).
    Section 3.1 uses cross-entropy as the storage lower bound and relies on arithmetic coding to approach it.
  • domain assumption The anchor-based representation (Scaffold-GS) is a valid and efficient base for 3DGS compression.
    The entire framework is built on anchors; if anchors were replaced, the method would not apply.
  • ad hoc to paper A PointNet++ model pretrained on generic 3D tasks provides transferable features for anchor location distributions in 3DGS scenes.
    Section 3.3.2 claims scene-agnostic generalization without evidence beyond a single-scene ablation.
  • domain assumption The adaptive quantization procedure from HAC is correctly integrated and introduces no extra distortion beyond the learned step.
    Section 3.2 invokes HAC's adaptive quantization; the paper does not re-derive or validate it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HEMGS: A Hybrid Entropy Model for 3D Gaussian Splatting Data Compression." pith.science (2026). https://pith.science/paper/OK5YOJQZ

@misc{pith2026241118473,
  author       = {Pith},
  title        = {Pith review of: HEMGS: A Hybrid Entropy Model for 3D Gaussian Splatting Data Compression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OK5YOJQZ}},
  note         = {Machine review of arXiv:2411.18473}
}
abstract

In this work, we propose a novel compression framework for 3D Gaussian Splatting (3DGS) data. Building on anchor-based 3DGS methodologies, our approach compresses all attributes within each anchor by introducing a novel Hybrid Entropy Model for 3D Gaussian Splatting (HEMGS) to achieve hybrid lossy-lossless compression. It consists of three main components: a variable-rate predictor, a hyperprior network, and an autoregressive network. First, unlike previous methods that adopt multiple models to achieve multi-rate lossy compression, thereby increasing training overhead, our variable-rate predictor enables variable-rate compression with a single model and a hyperparameter $\lambda$ by producing a learned Quantization Step feature for versatile lossy compression. Second, to improve lossless compression, the hyperprior network captures both scene-agnostic and scene-specific features to generate a prior feature, while the autoregressive network employs an adaptive context selection algorithm with flexible receptive fields to produce a contextual feature. By integrating these two features, HEMGS can accurately estimate the distribution of the current coding element within each attribute, enabling improved entropy coding and reduced storage. We integrate HEMGS into a compression framework, and experimental results on four benchmarks indicate that HEMGS achieves about a 40% average reduction in size while maintaining rendering quality over baseline methods and achieving state-of-the-art compression results.

Figures

Figures reproduced from arXiv: 2411.18473 by the authors.

Figure 1
Figure 1. The overview of our 3DGS data compression framework, which integrates HEMGS with other additional coding components. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. (a) The details of our HEMGS, which comprises a variable-rate predictor, a hyperprior network, and an autoregressive network. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The Rate-Distortion (RD) curves on three benchmark datasets, including Mip-NeRF360, Tank&Temples, and DeepBlending. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The Rate-Distortion (RD) curves of our variable-rate [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 6 Pith papers

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

  1. ECoNGS: Efficient Compressive Neural Gaussian Splats for Volume Visualization

    cs.CV 2026-07 conditional novelty 6.0 of 10

    ECoNGS compresses volume-visualization scenes into entropy-coded neural Gaussian splats that are up to 6x smaller, train up to 6x faster, and render more accurately than the prior iVR-GS method.

  2. NanoGS: Training-Free Gaussian Splat Simplification

    cs.CV 2026-03 conditional novelty 6.0 of 10

    A training-free, CPU-based graph-merging method reduces 3D Gaussian Splat primitive counts by orders of magnitude while maintaining higher rendering fidelity than prior compaction methods.

  3. Smol-GS: Compact Representations for Abstract 3D Gaussian Splatting

    cs.CV 2025-11 conditional novelty 6.0 of 10

    A 3D Gaussian Splatting compression method that stores coordinates in an occupancy octree and represents appearance with 8-d learned features, reaching ~4.7-6.4 MB per scene at near-SOTA quality.

  4. A Novel Benchmark and Dataset for Efficient 3D Gaussian Splatting with Gaussian Point Cloud Compression

    cs.GR 2025-05 conditional novelty 6.0 of 10

    GausPcgc adapts learned point cloud compression to Gaussian Splatting anchor positions, and the new GausPcc-1K dataset improves position bitrate by 8.2% over G-PCC v23 in the paper's benchmark.

  5. 4DGS-CC: A Contextual Coding Framework for 4D Gaussian Splatting Data Compression

    cs.CE 2025-04 conditional novelty 6.0 of 10

    4DGS-CC compresses 4D Gaussian Splatting data by about 12x with maintained rendering quality by entropy-coding quantized neural voxels and a vector-quantized codebook.

  6. Efficient Geometry Compression and Communication for 3D Gaussian Splatting Point Clouds

    cs.MM 2025-09 conditional novelty 3.0 of 10

    Integrating AVS PCRM geometry coding into the i3DV Gaussian platform, with Morton-code alignment, saves 10-25% total bitrate without changing rendering quality.

Reference graph

Works this paper leans on

42 extracted references · 32 canonical work pages · cited by 6 Pith papers

  1. [1]

    Variational image compression with a scale hyperprior

    Johannes Ball ´e, David Minnen, Saurabh Singh, Sung Jin Hwang, and Nick Johnston. Variational image compression with a scale hyperprior. ICLR, 2018. 1, 2, 3

    work page 2018

  2. [2]

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

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. In CVPR, pages 5470– 5479, 2022. 2, 5, 6, 8

    work page 2022

  3. [3]

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

    Yihang Chen, Qianyi Wu, Weiyao Lin, Mehrtash Harandi, and Jianfei Cai. Hac: Hash-grid assisted context for 3d gaussian splatting compression. In ECCV, pages 422–438. Springer, 2025. 1, 2, 3, 4, 5, 6, 7, 8

    work page 2025

  4. [4]

    Ex- ploiting intra-slice and inter-slice redundancy for learning- based lossless volumetric image compression

    Zhenghao Chen, Shuhang Gu, Guo Lu, and Dong Xu. Ex- ploiting intra-slice and inter-slice redundancy for learning- based lossless volumetric image compression. IEEE TIP , 31:1697–1707, 2022. 3

    work page 2022

  5. [5]

    Lsvc: A learning-based stereo video compres- sion framework

    Zhenghao Chen, Guo Lu, Zhihao Hu, Shan Liu, Wei Jiang, and Dong Xu. Lsvc: A learning-based stereo video compres- sion framework. In CVPR, pages 6073–6082, 2022

    work page 2022

  6. [6]

    Neural video compression with spatio- temporal cross-covariance transformers

    Zhenghao Chen, Lucas Relic, Roberto Azevedo, Yang Zhang, Markus Gross, Dong Xu, Luping Zhou, and Christo- pher Schroers. Neural video compression with spatio- temporal cross-covariance transformers. In ACM MM, pages 8543–8551, 2023

    work page 2023

  7. [7]

    Group-aware parameter-efficient updating for content- adaptive neural video compression

    Zhenghao Chen, Luping Zhou, Zhihao Hu, and Dong Xu. Group-aware parameter-efficient updating for content- adaptive neural video compression. In ACM MM , pages 11022–11031, 2024

    work page 2024

  8. [8]

    Learned image compression with discretized gaussian mixture likelihoods and attention modules

    Zhengxue Cheng, Heming Sun, Masaru Takeuchi, and Jiro Katto. Learned image compression with discretized gaussian mixture likelihoods and attention modules. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7939–7948, 2020. 1, 2, 3, 5

    work page 2020

Show all 42 references
  1. [9]

    Elements of information theory

    Thomas M Cover. Elements of information theory . John Wiley & Sons, 1999. 3

    1999

  2. [10]

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

    Zhiwen Fan, Kevin Wang, Kairun Wen, Zehao Zhu, De- jia Xu, and Zhangyang Wang. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps. arXiv preprint arXiv:2311.17245, 2023. 1, 2, 6

    2023 arXiv

  3. [11]

    Ea- gles: Efficient accelerated 3d gaussians with lightweight en- codings

    Sharath Girish, Kamal Gupta, and Abhinav Shrivastava. Ea- gles: Efficient accelerated 3d gaussians with lightweight en- codings. arXiv preprint arXiv:2312.04564, 2023. 6

    2023 arXiv

  4. [12]

    Calculation of average psnr differences between rd curves

    Bjontegaard Gisle. Calculation of average psnr differences between rd curves. In VCEG-M33, page 7, 2001. 6

    2001

  5. [13]

    Cra5: Extreme compression of era5 for portable global climate and weather research via an efficient varia- tional transformer

    Tao Han, Zhenghao Chen, Song Guo, Wanghan Xu, and Lei Bai. Cra5: Extreme compression of era5 for portable global climate and weather research via an efficient varia- tional transformer. arXiv preprint arXiv:2405.03376, 2024. 3

    2024 arXiv

  6. [14]

    Checkerboard context model for efficient learned image compression

    Dailan He, Yaoyan Zheng, Baocheng Sun, Yan Wang, and Hongwei Qin. Checkerboard context model for efficient learned image compression. In CVPR, pages 14771–14780,

  7. [15]

    Deep blending for free-viewpoint image-based rendering

    Peter Hedman, Julien Philip, True Price, Jan-Michael Frahm, George Drettakis, and Gabriel Brostow. Deep blending for free-viewpoint image-based rendering. ACM Trans. Graph., 37(6):1–15, 2018. 2, 5, 6, 7

    2018

  8. [16]

    Improving deep video compres- sion by resolution-adaptive flow coding

    Zhihao Hu, Zhenghao Chen, Dong Xu, Guo Lu, Wanli Ouyang, and Shuhang Gu. Improving deep video compres- sion by resolution-adaptive flow coding. In ECCV, pages 193–209. Springer, 2020. 3

    2020

  9. [17]

    Fvc: A new framework towards deep video compression in feature space

    Zhihao Hu, Guo Lu, and Dong Xu. Fvc: A new framework towards deep video compression in feature space. In CVPR, pages 1502–1511, 2021. 1, 2, 3

    2021

  10. [18]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4):139–1,

  11. [19]

    Adam: A method for stochastic opti- mization

    Diederik P Kingma. Adam: A method for stochastic opti- mization. ICLR, 2015. 6

    2015

  12. [20]

    Tanks and temples: Benchmarking large-scale scene reconstruction

    Arno Knapitsch, Jaesik Park, Qian-Yi Zhou, and Vladlen Koltun. Tanks and temples: Benchmarking large-scale scene reconstruction. ACM Trans. Graph., 36(4):1–13, 2017. 2, 5, 6

    2017

  13. [21]

    Compact 3d gaussian representation for radiance field

    Joo Chan Lee, Daniel Rho, Xiangyu Sun, Jong Hwan Ko, and Eunbyung Park. Compact 3d gaussian representation for radiance field. In CVPR, pages 21719–21728, 2024. 1, 2, 3, 6

    2024

  14. [22]

    Deep contextual video com- pression

    Jiahao Li, Bin Li, and Yan Lu. Deep contextual video com- pression. Advances in Neural Information Processing Sys- tems, 34, 2021. 1, 2, 3, 5

    2021

  15. [23]

    Icmh- net: Neural image compression towards both machine vision and human vision

    Lei Liu, Zhihao Hu, Zhenghao Chen, and Dong Xu. Icmh- net: Neural image compression towards both machine vision and human vision. In ACM MM, pages 8047–8056, 2023

    2023

  16. [24]

    Pchm-net: A new point cloud compression framework for both human vision and machine vision

    Lei Liu, Zhihao Hu, and Jing Zhang. Pchm-net: A new point cloud compression framework for both human vision and machine vision. In ICME, pages 1997–2002, 2023

    1997

  17. [25]

    Towards point cloud compression for machine perception: A simple and strong baseline by learning the octree depth level predictor

    Lei Liu, Zhihao Hu, and Zhenghao Chen. Towards point cloud compression for machine perception: A simple and strong baseline by learning the octree depth level predictor. In International Joint Conference on Artificial Intelligence WorkShop, pages 3–17. Springer, 2024

    2024

  18. [26]

    An efficient adaptive compression method for human perception and machine vision tasks

    Lei Liu, Zhenghao Chen, Zhihao Hu, and Dong Xu. An efficient adaptive compression method for human perception and machine vision tasks. arXiv preprint arXiv:2501.04329,

    arXiv

  19. [27]

    Compgs: Efficient 3d scene representa- tion via compressed gaussian splatting

    Xiangrui Liu, Xinju Wu, Pingping Zhang, Shiqi Wang, Zhu Li, and Sam Kwong. Compgs: Efficient 3d scene representa- tion via compressed gaussian splatting. In ACM MM, 2024. 6

    2024

  20. [28]

    Dvc: An end-to-end deep video com- pression framework

    Guo Lu, Wanli Ouyang, Dong Xu, Xiaoyun Zhang, Chunlei Cai, and Zhiyong Gao. Dvc: An end-to-end deep video com- pression framework. In CVPR, pages 11006–11015, 2019. 1, 2, 3

    2019

  21. [29]

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

    Tao Lu, Mulin Yu, Linning Xu, Yuanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In CVPR, pages 20654–20664, 2024. 1, 2, 3, 4, 5, 6

    2024

  22. [30]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. Communications of the ACM, 65(1):99–106, 2021. 1, 2, 5, 6

    2021

  23. [31]

    Joint autoregressive and hierarchical priors for learned im- age compression

    David Minnen, Johannes Ball ´e, and George D Toderici. Joint autoregressive and hierarchical priors for learned im- age compression. NeurIPS, 31, 2018. 1, 2, 3, 5

    2018

  24. [32]

    Compact 3d scene representation via self- organizing gaussian grids

    Wieland Morgenstern, Florian Barthel, Anna Hilsmann, and Peter Eisert. Compact 3d scene representation via self- organizing gaussian grids. ECCV, 2025. 2, 6

    2025

  25. [33]

    Compact3d: Com- pressing gaussian splat radiance field models with vector quantization

    KL Navaneet, Kossar Pourahmadi Meibodi, Soroush Abbasi Koohpayegani, and Hamed Pirsiavash. Compact3d: Com- pressing gaussian splat radiance field models with vector quantization. arXiv preprint arXiv:2311.18159, 2023. 1, 2, 6

    2023 arXiv

  26. [34]

    Compressed 3d gaussian splatting for accelerated novel view synthesis

    Simon Niedermayr, Josef Stumpfegger, and R ¨udiger West- ermann. Compressed 3d gaussian splatting for accelerated novel view synthesis. In CVPR, pages 10349–10358, 2024. 1, 2, 6

    2024

  27. [35]

    Reducing the memory footprint of 3d gaussian splatting

    Panagiotis Papantonakis, Georgios Kopanas, Bernhard Kerbl, Alexandre Lanvin, and George Drettakis. Reducing the memory footprint of 3d gaussian splatting. Proceedings of the ACM on Computer Graphics and Interactive Tech- niques, 7(1):1–17, 2024. 3

    2024

  28. [36]

    Pointnet++: Deep hierarchical feature learning on point sets in a metric space

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. NeurIPS, 30, 2017. 4

    2017

  29. [37]

    Learning accurate entropy model with global reference for image compression

    Yichen Qian, Zhiyu Tan, Xiuyu Sun, Ming Lin, Dongyang Li, Zhenhong Sun, Li Hao, and Rong Jin. Learning accurate entropy model with global reference for image compression. In ICLR, 2021. 2, 3, 5

    2021

  30. [38]

    Entroformer: A transformer-based entropy model for learned image compression

    Yichen Qian, Xiuyu Sun, Ming Lin, Zhiyu Tan, and Rong Jin. Entroformer: A transformer-based entropy model for learned image compression. In ICLR, 2022. 2, 3, 5

    2022

  31. [39]

    ContextGS : Compact 3d gaussian splatting with anchor level context model

    Yufei Wang, Zhihao Li, Lanqing Guo, Wenhan Yang, Alex Kot, and Bihan Wen. ContextGS : Compact 3d gaussian splatting with anchor level context model. In The Thirty- eighth Annual Conference on Neural Information Processing Systems, 2024. 1, 2, 3, 4, 5, 6, 7, 8

    2024

  32. [40]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE TIP, 13(4):600–612, 2004. 6

    2004

  33. [41]

    Differentiable point-based radiance fields for efficient view synthesis

    Qiang Zhang, Seung-Hwan Baek, Szymon Rusinkiewicz, and Felix Heide. Differentiable point-based radiance fields for efficient view synthesis. In SIGGRAPH Asia 2022 Con- ference Papers, pages 1–12, 2022. 3

    2022

  34. [42]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, pages 586–595,

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.