Pith. sign in

REVIEW 5 major objections 5 minor 5 cited by

Compression of 3D Gaussian Splatting with Optimized Feature Planes and Standard Video Codecs

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A tri-plane feature field trained on DCT entropy lets standard HEVC compress 3D Gaussian Splatting to under 10 MB, at up to 76x compression and only 0.19 dB PSNR loss.

desk verdict A genuinely useful tri-plane plus video-codec pipeline for 3DGS compression with plausible sub-10MB results, but the entropy model in Eq. (2) is never defined and some per-scene claims are over-stated. read the letter →

arxiv 2501.03399 v1 pith:NWI7YB6C submitted 2025-01-06 cs.CV cs.MM

Soonbin Lee , Fangwen Shu , Yago Sanchez , Thomas Schierl , Cornelius Hellge This is my paper · ORCID
classification cs.CVcs.MM
keywords 3DGaussianSplattingcompressiontri-planefeatureplanesfrequency-domainentropymodelingDCTtransformcodingvideocodecchannel-wisebitallocationnovelviewsynthesisHEVC
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

This paper tries to show that the storage bottleneck of 3D Gaussian Splatting can be broken by re-encoding the scene as video. Instead of compressing millions of gaussian attribute vectors, the authors train a tri-plane feature field that predicts every attribute, then feed those planes to a standard HEVC encoder after shaping them with a DCT-domain entropy loss. On Mip-NeRF360 this reaches up to 76x compression over the original 3DGS with only 0.19 dB PSNR loss, and under 10 MB total storage on every dataset tested. If the claim holds, 3DGS scenes could be shipped as compact standard-codec bitstreams and decoded on devices with hardware video decoding, making high-quality 3D scenes practical for mobile and head-mounted use.

What carries the argument

The load-bearing object is the optimized tri-plane feature field: three axis-aligned 2D grids $P_{XY}, P_{XZ}, P_{YZ}$ with eight channels each, queried at each gaussian position and decoded by a small MLP to produce all splat attributes. What makes the compression work is a frequency-domain entropy loss applied to each plane channel, $L_{\mathrm{ent}} = \sum_c w_c I(F(P_c))$, where $F$ is a $4\times4$ block-wise DCT and the entropy is estimated with uniform noise at Qstep $2^8$; this steers the planes toward signals that a DCT-based video codec can code cheaply. Channel importance scores $w_c = \mathrm{CI}_1(P)/\mathrm{CI}_c(P)$ allocate fewer bits to less visible channels, progressive channel masking stabilizes training, and a piecewise-projective contraction maps unbounded scenes onto the finite plane so neighboring blocks carry real spatial correlation. At deployment, the 32 channels are concatenated into 32 16-bit frames and encoded as one HEVC sequence.

What would settle it

Run the released pipeline on a scene with repetitive high-frequency textures, sweep $\lambda_{\mathrm{ent}}$ while holding the video QP fixed, and compare the ordering of the DCT-entropy loss with the actual HEVC bitstream sizes; any inversion, a plane that scores lower in $L_{\mathrm{ent}}$ but encodes to more bytes, would show the entropy proxy fails. Repeating the same sweep across Qstep values and scenes would test whether the fixed $2^8$ quantizer step remains optimal.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that the millions of per-gaussian attribute vectors in 3DGS can be replaced by a compact continuous field: a tri-plane of 32 16-bit channels (three axis-aligned planes with eight levels each) predicts color, scale, rotation, and opacity through a small decoder multi-layer perceptron, so the scene is compressed by coding those planes rather than the gaussians. The decisive step is to optimize the planes for the codec's frequency domain: instead of sparsifying with L1, the training loss minimizes the entropy of block-wise 4x4 DCT coefficients, and channel importance scores re-weight that loss to spend bits where they affect rendering most. The trained planes are normalized to 16-bit integer frames and fed to an unmodified HEVC encoder at QP=1; point positions are Morton-sorted, packed, and losslessly coded. The reported result is under 10 MB per scene across Mip-NeRF360, DeepBlending, and Tank&Temples, with up to 76x compression versus 3DGS and only 0.19 dB PSNR loss on Mip-NeRF360.

Load-bearing premise

The load-bearing assumption is that minimizing the entropy of the planes' $4\times4$ block frequency coefficients, with one fixed quantization strength, predicts how many bytes the video codec really spends; if that match fails on other scenes, channel counts, or bitrates, the reported size gains shrink.

Editorial extensions

If this is right

  • Scenes from the tested datasets can be stored in under 10 MB, with Mip-NeRF360 compressed up to 76x against the original 3DGS at a cost of 0.19 dB PSNR.
  • After decoding the feature plane frames and predicting attributes once, rendering uses the original 3DGS rasterizer with no extra per-frame overhead, so render speed stays close to uncompressed 3DGS.
  • Because the bitstream is produced by unmodified HEVC encoders, decoding can ride on hardware video decoders, which is the paper's route to mobile and head-mounted deployment.
  • Rate-distortion control is best done during training through $\lambda_{\mathrm{ent}}$, not by raising the codec QP at encode time; the appendix shows QP-based control yields worse rate-distortion performance.
  • The compression transfers to a second HEVC implementation (libx265) with similar savings, so the method is not tied to one codec software.

Reading between the lines

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

  • Because the paper leaves point positions losslessly coded and roughly constant in size, coupling this plane codec with point pruning would likely push total size well below the reported values; the paper notes this possibility but does not implement it.
  • The same DCT-entropy design should transfer to newer transform-based codecs such as VVC, which the paper names as motivation but does not test; larger transform blocks and better intra prediction may improve the rate-distortion curve further, or may require retuning Qstep.
  • The progressive channel structure suggests a natural streaming scheme: decode low-level channels first for a coarse scene, then add higher channels for detail; the paper does not build this, but the training order and channel importance scores make it feasible.
  • The channel-importance weights are computed once at a fixed training iteration from a sensitivity snapshot; recomputing them for the target bitrate or for a user's expected viewing directions could yield a better bit allocation than the single static weight vector.
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 a compression method for 3D Gaussian Splatting (3DGS) that replaces per-Gaussian attributes with a compact tri-plane feature representation, trains the feature planes with a DCT-domain entropy loss and channel-wise bit allocation, and then compresses the resulting 32-channel 16-bit feature planes using standard video codecs (HM, libx265). Point positions are Morton-sorted, packed into images, and losslessly coded. The authors report storage sizes below 10 MB on dataset averages for Mip-NeRF360, DeepBlending, and Tank&Temples, with quality close to the original 3DGS, and claim compression ratios up to 76x. The paper includes per-scene results, ablation studies, and an analysis of quantization step size.

Significance. If the results are reproducible, the paper makes a valuable contribution by showing that standard, hardware-decodable video codecs can serve as the compression engine for 3DGS attributes, potentially enabling practical deployment on mobile devices. The DCT-domain entropy modeling and channel-wise bit allocation are sensible ideas, and the paper reports detailed per-scene metrics and ablation studies. However, the current manuscript omits a definition of the entropy model p used in the training loss, which makes the central optimization objective underspecified and the results non-reproducible. The claim that all datasets are under 10 MB is also contradicted by the paper's own per-scene tables. These issues are load-bearing and must be fixed before the results can be trusted.

major comments (5)
  1. [Section 3.3, Eq. (2)] The entropy loss I(P) = E[-log p(P̃)] requires a probability model p of the quantized (or noised) plane coefficients. The paper never defines p: no functional form, architecture, or training procedure is given. The citation to Ballé et al. suggests a factorized density, but neither the main text nor the supplementary material specifies how p is parameterized or learned. Since the total loss in Eq. (8) includes λ_ent L_ent, where L_ent is defined by this p, the entire training objective is undefined as written. This is not an incidental omission; the ablation results in Fig. 7 and Table 2 cannot be reproduced without knowing p. The authors must either explicitly define p (e.g., a factorized logistic model with a small neural network, as in variational image compression) or use a simple fixed density (e.g., a histogram or Laplacian) and say so.
  2. [Section 4.2, Table 6] The text claims that the method "achieved sizes under 10MB for all datasets." This is contradicted by the per-scene results in Table 6: the `garden` scene requires 15.22 MB and `stump` requires 14.30 MB on Mip-NeRF360. Only the dataset averages (9.78 MB, 8.62 MB, 7.46 MB) are under 10 MB. The claim as stated is factually incorrect and should be rephrased to "dataset averages under 10 MB" or the method should be re-evaluated. The abstract and introduction also use "typically <10MB," which is ambiguous; the discrepancy between the headline claim and the actual per-scene data must be resolved.
  3. [Section 3.3 and Appendix Table 5] The DCT entropy loss operates on 4×4 block DCT coefficients with a quantization step Qstep=2^8, chosen as "the best" from an analysis on a single scene (`bonsai`). No sensitivity analysis across scenes or bitrate ranges is provided. More fundamentally, the proxy is not validated against actual HEVC bitrate: the paper shows in Fig. 7 that adding L_ent improves RD on four scenes, but it does not demonstrate that the entropy estimate correlates with the HM output size beyond those cases. Since the method's entire rate-control mechanism depends on this proxy, the authors should provide quantitative evidence (e.g., a scatter plot of L_ent vs. encoded size across scenes and λ_ent values) or at least a cross-scene validation of the Qstep choice.
  4. [Section 4.1 and Section 3.1] The feature-plane architecture is described inconsistently. The paper first defines a tri-plane P (XY, XZ, YZ), then says "Each feature plane has 8 channels and a 512×512 resolution with a decoder MLP to predict each attribute (color, scale, rotation, opacity) separately. Therefore, a total of 32 channels are used to predict all attributes." A tri-plane with three planes of 8 channels would give 24 channels, not 32. The ambiguity makes it impossible to understand whether the method uses one tri-plane per attribute, four separate planes, or some other layout. Please clarify the number of planes, the channel count per plane, and how the 32 channels are derived.
  5. [Section 3.3 (DCT entropy modeling)] The proposed entropy loss is applied separately to each channel's 4×4 DCT coefficients, but the actual HEVC encoding operates on the 32 concatenated frames with inter-frame prediction, adaptive transform sizes, and rate-distortion optimization. The paper provides no analysis of how well the per-channel DCT entropy approximates the output bitrate under these conditions. While the empirical RD gains in Fig. 7 are encouraging, the generalization of the proxy to other scenes, other codec configurations, or other bitrate ranges is uncertain. The authors should either add a direct comparison of predicted vs. actual bitrate or acknowledge this limitation explicitly in the paper.
minor comments (5)
  1. [Section 3.3] The sentence "we found that a Qstep of 28 yields the best results" is ambiguous; the appendix (Table 5) uses exponents (2^0, 2^2, ...), so it should read "Qstep of 2^8".
  2. [Supplementary Material, Table 5] The Qstep values in Table 5 are listed as "20, 22, ..." without superscripts, making it unclear whether these are powers of two; please format as 2^0, 2^2, etc.
  3. [General] The paper contains numerous garbled text fragments and encoding artifacts (e.g., sequences like "/uni00000014/uni00000019" in Section 4.1 and corrupted reference entries). The manuscript should be thoroughly proofread and regenerated.
  4. [Figure 5] The x-axis is labeled as a log2 scale, but the axis ticks are not explained; please add clear axis labels or captions that describe the unit and scale.
  5. [Section 4.1] The exact hyperparameters λ_ent and λ_1 are not stated in the main text or supplementary material, aside from "small amount" for L1. Please provide the numeric values or a range used in the experiments.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: reported sizes are measured HEVC bitstreams, and the DCT-entropy loss is an empirically validated proxy rather than an identity.

full rationale

The paper's central quantitative claims are not derived from its training objective; they are measured by feeding the trained, 16-bit feature planes and Morton-packed positions to an external HEVC encoder (HM 16.0 / libx265) and summing the resulting bitstream sizes. Eq. (6)'s Lent is a differentiable proxy whose minimization is shown in Fig. 7 to improve actual codec output, so the relation between Lent and the reported sizes is empirical validation, not identity or construction. The Qstep=2^8 choice is an ablation over final size/PSNR (Table 5), not a parameter that directly sets the reported bitrate. The only self-citation ([21], ECRF) appears in a related-work enumeration of Fourier-based transforms and is not load-bearing for any equation or result. The underspecification of the density p in Eq. (2) is a reproducibility/correctness concern, not a circularity: it does not make the reported HEVC sizes equal to Lent by construction. No step reduces a predicted quantity to a fitted input or imports a uniqueness/ansatz via self-citation.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several modeling assumptions about how well plane features and standard video codecs interact. No new physical or conceptual entities are introduced. The most important engineering choices, Qstep, QP, lambda_ent, lambda_1, and the training schedule, are hand-set and not all values are reported, which limits independent verification.

free parameters (6)
  • DCT entropy Qstep = 2^8 (256)
    Empirically selected on the Bonsai scene with fixed lambda_ent in Appendix Table 5; used for all scenes. It controls how the entropy loss maps to final HEVC bitrate.
  • Video codec QP = 1
    Feature planes are encoded with QP=1 and all QP offsets set to 0; chosen based on Appendix Figure 9, where adjusting QP gives worse rate-distortion than adjusting lambda_ent.
  • Entropy loss weight lambda_ent = Not stated numerically
    Balances bitrate and rendering quality; varied per experiment (Tables 3 and Figure 9), but exact values are not reported in the main text.
  • L1 sparsity weight lambda_1 = Not stated numerically
    Small L1 loss added to reduce noisy feature plane regions; exact value is not given in the paper.
  • Progressive training schedule = Ti={0,5000,10000,15000}, Li={2,4,6,8}
    Hand-chosen curriculum for channel masking during feature plane training; no sensitivity study is reported.
  • Plane resolution and channel count = 512x512 resolution, 8 channels per plane
    Architectural choice determining the size of the feature plane video frames; chosen without an ablation.
assumptions (6)
  • domain assumption Gaussian attributes can be adequately predicted by decoding tri-plane features with small MLPs using Hadamard product factorization (Eq. 1).
    Section 3.1 adopts k-planes; the whole compression scheme depends on plane features capturing the attribute variability of millions of gaussians.
  • domain assumption A block-wise 4x4 DCT entropy loss on feature planes is a sufficient proxy for HEVC bitrate; minimizing I(F(P)) improves actual coded size.
    Section 3.3 and Figure 7; this is the core codec-compatibility premise, validated only empirically.
  • domain assumption Piecewise-projective contraction maps unbounded scenes onto finite planes without significant quality loss and improves spatial correlation for DCT.
    Section 4.1, Eq. 9, adopted from Merf [31]; used for all unbounded scenes.
  • domain assumption Morton-order sorting and lossless HEVC coding of quantized 16-bit positions reproduce point coordinates exactly.
    Section 4.1 states point positions are critical and a lossless setting is used; this assumes no information loss in packing or coding.
  • domain assumption Two-phase training with a fixed point set after 15k 3DGS iterations does not prevent the feature planes from learning high-quality attributes.
    Section 3.1 keeps the number of points constant during plane training; if this conflicts with densification, rendering quality would drop.
  • standard math Standard DCT and entropy modeling with uniform noise and a straight-through estimator are valid for optimizing quantized transform coefficients.
    Section 3.3, Eqs. 2 and 3; background from Ballé et al. [1,2,3] and standard transform coding.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Compression of 3D Gaussian Splatting with Optimized Feature Planes and Standard Video Codecs." pith.science (2026). https://pith.science/paper/NWI7YB6C

@misc{pith2026250103399,
  author       = {Pith},
  title        = {Pith review of: Compression of 3D Gaussian Splatting with Optimized Feature Planes and Standard Video Codecs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NWI7YB6C}},
  note         = {Machine review of arXiv:2501.03399}
}
read the original abstract

3D Gaussian Splatting is a recognized method for 3D scene representation, known for its high rendering quality and speed. However, its substantial data requirements present challenges for practical applications. In this paper, we introduce an efficient compression technique that significantly reduces storage overhead by using compact representation. We propose a unified architecture that combines point cloud data and feature planes through a progressive tri-plane structure. Our method utilizes 2D feature planes, enabling continuous spatial representation. To further optimize these representations, we incorporate entropy modeling in the frequency domain, specifically designed for standard video codecs. We also propose channel-wise bit allocation to achieve a better trade-off between bitrate consumption and feature plane representation. Consequently, our model effectively leverages spatial correlations within the feature planes to enhance rate-distortion performance using standard, non-differentiable video codecs. Experimental results demonstrate that our method outperforms existing methods in data compactness while maintaining high rendering quality. Our project page is available at https://fraunhoferhhi.github.io/CodecGS

Figures

Figures reproduced from arXiv: 2501.03399 by the authors.

Figure 1
Figure 1. Our method achieves a 146× compression with negli￾gible loss in image quality, a significant improvement over 3DGS [17] the ‘bicycle’ scene. Our method seamlessly integrates with standard video codecs and utilizes the original 3DGS rendering pipeline, achieving comparable rendering speeds with minimal overhead. as a collection of 3D points, each characterized by learn￾able gaussian attributes. These parameters are o… view at source ↗
Figure 2
Figure 2. Overview of the proposed model. Following the original 3DGS densification, we train the model to predict all gaussian attributes using the feature plane architecture. The feature plane achieves a more compact representation through our proposed DCT entropy modeling and channel bit allocation techniques, which leads to performance improvements. g(P(x)) = {c, ˜ s, ˜ q, ˜ σ˜} (1) Despite its high expressiveness, learni… view at source ↗
Figure 4
Figure 4. Visualization of the 1st, 3rd, and 5th channels of XZ plane with progressive training for the ‘Flowers’ scene. The iteration stages Ti are {0, 5000, 10000} with corresponding Li values of {2, 4, 6} for 30k iteration training. With the progressive training, parameter energy is mainly concentrated in the lower￾level channels, while the higher-level channels show sparser rep￾resentations. to the range [0, 1] and then s… view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: RD curves for quantitative comparisons. Rate-distortion (RD) plots are provided for each dataset using 3DGS compression models. For the x-axis, a log2 scale is used for better visualization [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Qualitative results for visual comparison. Each subfigure displays the storage size along with the PSNR, SSIM, and LPIPS metrics. access (RA) configuration. The default RA configuration includes predefined QP offsets between frames to improve general video compression …
Figure 7
Figure 7. Figure 7: RD curves from the ablation study demonstrate the effectiveness of our proposed method. Based on L1, adding Lent as a loss function provides a clear benefit. Moreover, in￾troducing channel-wise bit allocation wc results in improved rate￾distortion performance. Datasets…
Figure 8
Figure 8. Figure 8: Visualization of channel levels in XZ plane for the ‘Bonsai’ scene. With dynamic wc, the lower-level channels preserve more information, whereas the higher-level channels are largely minimized due to the higher λent weight assignment. Analysis of Bit Allocation [PITH_…
Figure 10
Figure 10. Figure 10: Visualization of each contraction method for the ‘Bonsai’ scene. (a) sphere contraction and (b) piecewise￾projective contraction. Contraction methods [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 9
Figure 9. Figure 9: RD curves with video QP adjustment. Controlling rate-distortion with video QP is worse than using λent with QP=1. Performance analysis of QP adjustment. Beyond ad￾justing the parameter λent, the rate-distortion trade-off can also be controlled by modifying the quantiza…
Figure 11
Figure 11. Figure 11: Qualitative results for visual comparison for Mip-NeRF360 dataset. Each subfigure displays the storage size along with the PSNR, SSIM, and LPIPS metrics. Detailed observation is encouraged by zooming in. 13 [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Qualitative results for visual comparison for DeepBlending and T&T dataset. Each subfigure displays the storage size along with the PSNR, SSIM, and LPIPS metrics. Detailed observation is encouraged by zooming in. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

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

  1. TinySplat: Feedforward Approach for Generating Compact 3D Scene Representation

    cs.CV 2025-06 conditional novelty 7.0 of 10

    TinySplat compresses feedforward 3D Gaussian scenes by 105-199x on two-view benchmarks (about 50x on DL3DV) while keeping rendered quality close to the uncompressed model.

  2. CGHair: Compact Gaussian Hair Reconstruction with Card Clustering

    cs.CV 2026-04 conditional novelty 6.0 of 10

    Hierarchical card clustering plus shared Gaussian texture codebooks reconstructs multi-view hair with 200x lower memory and 4x faster strand generation while matching prior 3DGS visual quality.

  3. CF3: Compact and Fast 3D Feature Fields

    cs.CV 2025-08 conditional novelty 6.0 of 10

    CF3 builds a compact 3D feature field from a pre-trained 3DGS by feature lifting, per-Gaussian autoencoding, and adaptive sparsification, matching baseline segmentation quality with roughly 5% of the Gaussians.

  4. Hybrid 3D-4D Gaussian Splatting for Fast Dynamic Scene Representation

    cs.CV 2025-05 conditional novelty 6.0 of 10

    A dynamic scene rendering method that tags Gaussians as static or dynamic by their temporal scale, converting static ones to 3D to cut training time.

  5. Confident Splatting: Confidence-Based Compression of 3D Gaussian Splatting via Learnable Beta Distributions

    cs.GR 2025-06 conditional novelty 5.0 of 10

    Learned per-splat Beta-distributed confidence scores enable test-time pruning of 3D Gaussian splats with minor quality loss, and average confidence is proposed as a scene quality metric.

Reference graph

Works this paper leans on

43 extracted references · 39 canonical work pages · cited by 5 Pith papers

  1. [1]

    In International Con- ference on Learning Representations (ICLR) , 2017

    Johannes Ball´ e, V alero Laparra, and Eero P Simoncelli.End- to-end optimized image compression. In International Con- ference on Learning Representations (ICLR) , 2017. 3

    work page 2017

  2. [2]

    V ariational image compres- sion with a scale hyperprior

    Johannes Ball´ e, David Minnen, Saurabh Singh, Sung Jin Hwang, and Nick Johnston. V ariational image compres- sion with a scale hyperprior. In International Conference on Learning Representations (ICLR) , 2018

    work page 2018

  3. [3]

    Nonlinear transform coding

    Johannes Ball´ e, Philip A Chou, David Minnen, Saurabh Singh, Nick Johnston, Eirikur Agustsson, Sung Jin Hwang, and George Toderici. Nonlinear transform coding. IEEE Journal of Selected Topics in Signal Processing , 15(2):339– 353, 2020. 3

    work page 2020

  4. [4]

    Mip-nerf: A multiscale representation for anti-aliasing n eu- ral radiance fields

    Jonathan T Barron, Ben Mildenhall, Matthew Tancik, Pete r Hedman, Ricardo Martin-Brualla, and Pratul P Srinivasan. Mip-nerf: A multiscale representation for anti-aliasing n eu- ral radiance fields. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision (ICCV) , pages 5855– 5864, 2021. 1

    work page 2021

  5. [5]

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

    Jonathan T Barron, Ben Mildenhall, Dor V erbin, Pratul P Srinivasan, and Peter Hedman. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5470–5479, 2022. 1, 5, 7

    work page 2022

  6. [6]

    Overview of the versatile video coding (vvc) standard and its applications

    Benjamin Bross, Ye-Kui Wang, Yan Ye, Shan Liu, Jianle Chen, Gary J Sullivan, and Jens-Rainer Ohm. Overview of the versatile video coding (vvc) standard and its applications. IEEE Transactions on Circuits and Systems for Video tech- nology (TCSVT), 31(10):3736–3764, 2021. 2, 4

    work page 2021

  7. [7]

    Tensorf: Tensorial radiance fields

    Anpei Chen, Zexiang Xu, Andreas Geiger, Jingyi Y u, and Hao Su. Tensorf: Tensorial radiance fields. In European Conference on Computer Vision (ECCV) , pages 333–350. Springer, 2022. 2

    work page 2022

  8. [8]

    How Far can we Compress Instant-NGP-Based NeRF?

    Yihang Chen, Qianyi Wu, Mehrtash Harandi, and Jianfei Cai. How Far can we Compress Instant-NGP-Based NeRF? . In 2024 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR) , pages 20321–20330, Los Alami- tos, CA, USA, 2024. IEEE Computer Society. 2

    work page 2024

Show all 43 references
  1. [9]

    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 European Conference on Computer Vision, 2024. 2, 4, 5, 7

    2024

  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, pages 1–16, 2023. 1, 2, 5

    2023 arXiv

  3. [11]

    Accessed: 2024-09-12

    http://ffmpeg.org/ FFmpeg Developers. Accessed: 2024-09-12. 8

    2024

  4. [12]

    K-planes: Explicit radiance fields in space, time, and appearance

    Sara Fridovich-Keil, Giacomo Meanti, Frederik Rahbæk Warburg, Benjamin Recht, and Angjoo Kanazawa. K-planes: Explicit radiance fields in space, time, and appearance. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 12479–12488, 2023. 2

    2023

  5. [13]

    Ea- gles: Efficient accelerated 3d gaussians with lightweight encodings

    Sharath Girish, Kamal Gupta, and Abhinav Shrivastava. Ea- gles: Efficient accelerated 3d gaussians with lightweight encodings. arXiv preprint arXiv:2312.04564 , pages 1–10, 2023. 2, 5

    2023 arXiv

  6. [14]

    Deep blending for free-viewpoint image-based rendering

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

    2018

  7. [15]

    Kim, and Jin-Hwa Kim

    Hwan Heo, Taekyung Kim, Jiyoung Lee, Jaewon Lee, Soohyun Kim, Hyunwoo J. Kim, and Jin-Hwa Kim. Robust camera pose refinement for multi-resolution hash encoding. In Proceedings of the 40th International Conference on Ma- chine Learning. JMLR.org, 2023. 3

    2023

  8. [16]

    Accessed: 2024-09-12

    https://vcgit.hhi.fraunhofer.de/jvet/HM ISO/IEC 23008-2 High Efficiency Video Coding (HEVC) Test Model reference software. Accessed: 2024-09-12. 5, 8

    2024

  9. [17]

    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 Transactions on Graphics (ToG), 42(4):1–14, 2023. 1, 5

    2023

  10. [18]

    Synergistic integration of coordinate network and ten - sorial feature for improving nerfs from sparse inputs

    Mingyu Kim, Jun Seong Kim, Se Y oung Y un, and Jin Hwa Kim. Synergistic integration of coordinate network and ten - sorial feature for improving nerfs from sparse inputs. In Pro- ceedings of the 41th International Conference on Machine Learning. PMLR, 2024. 3

    2024

  11. [19]

    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 Transactions on Graphics, 36(4):1–13, 2017. 5, 7

    2017

  12. [20]

    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. arXiv preprint arXiv:2311.13681, 2023. 1, 2, 5

    2023 arXiv

  13. [21]

    Ecrf: Entropy-constrained neural ra - diance fields compression with frequency domain optimiza- tion, 2023

    Soonbin Lee, Fangwen Shu, Yago Sanchez, Thomas Schierl , and Cornelius Hellge. Ecrf: Entropy-constrained neural ra - diance fields compression with frequency domain optimiza- tion, 2023. 2

    2023

  14. [22]

    Compressing volumetric radiance fields to 1 mb

    Lingzhi Li, Zhen Shen, Zhongshu Wang, Li Shen, and Liefeng Bo. Compressing volumetric radiance fields to 1 mb. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 4222–4231, 2023. 2

    2023

  15. [23]

    Compgs: Efficient 3d scene represen- tation via compressed gaussian splatting

    Xiangrui Liu, Xinju Wu, Pingping Zhang, Shiqi Wang, Zhu Li, and Sam Kwong. Compgs: Efficient 3d scene represen- tation via compressed gaussian splatting. In Proceedings of the 32nd ACM International Conference on Multimedia , page 2936–2944, New Y ork, NY , USA, 2024. Associati...

    2024

  16. [24]

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

    Tao Lu, Mulin Y u, Linning Xu, Y uanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20654–20664, 2024. 2, 5, 7

    2024

  17. [25]

    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

    2021

  18. [26]

    Compact 3d scene representation via self- organizing gaussian grids, 2024

    Wieland Morgenstern, Florian Barthel, Anna Hilsmann, and Peter Eisert. Compact 3d scene representation via self- organizing gaussian grids, 2024. 2, 5 9

    2024

  19. [27]

    G. M. Morton. A computer oriented geodetic data base; an d a new technique in file sequencing. Technical report, IBM Ltd., Ottawa, Canada, 1966. 6

    1966

  20. [28]

    Instant neural graphics primitives with a mul- tiresolution hash encoding

    Thomas M¨ uller, Alex Evans, Christoph Schied, and Alex an- der Keller. Instant neural graphics primitives with a mul- tiresolution hash encoding. ACM Transactions on Graphics (ToG), 41(4):1–15, 2022. 2

    2022

  21. [29]

    Compressed 3d gaussian splatting for accelerated novel view synthesis

    Simon Niedermayr, Josef Stumpfegger, and R¨ udiger Wes t- ermann. Compressed 3d gaussian splatting for accelerated novel view synthesis. arXiv preprint arXiv:2401.02436 , pages 1–10, 2023. 1, 2, 4, 5

    2023 arXiv

  22. [30]

    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. Proc. ACM Comput. Graph. Interact. Tech., 7(1), 2024. 2

    2024

  23. [31]

    Merf: Memory-efficient radiance fields for real-time view synthesis in unbounded scenes

    Christian Reiser, Rick Szeliski, Dor V erbin, Pratul Sr ini- vasan, Ben Mildenhall, Andreas Geiger, Jon Barron, and Peter Hedman. Merf: Memory-efficient radiance fields for real-time view synthesis in unbounded scenes. ACM Trans. Graph., 42(4), 2023. 6

    2023

  24. [32]

    Masked wavelet representation for compact neural radiance fields

    Daniel Rho, Byeonghyeon Lee, Seungtae Nam, Joo Chan Lee, Jong Hwan Ko, and Eunbyung Park. Masked wavelet representation for compact neural radiance fields. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 20680–20690, 2023. 2

    2023

  25. [33]

    Binary radiance fields

    Seungjoo Shin and Jaesik Park. Binary radiance fields. I n Advances in Neural Information Processing Systems , 2023. 2

    2023

  26. [34]

    Overview of the high efficiency video coding (hevc) standard

    Gary J Sullivan, Jens-Rainer Ohm, Woo-Jin Han, and Thomas Wiegand. Overview of the high efficiency video coding (hevc) standard. IEEE Transactions on circuits and systems for video technology (TCSVT) , 22(12):1649–1668, 2012. 2, 4

    2012

  27. [35]

    Direct voxel grid optimization: Super-fast convergence for radiance fie lds reconstruction

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fie lds reconstruction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 5459–5469, 2022. 2

    2022

  28. [36]

    V ariable bitrate neural fields

    Towaki Takikawa, Alex Evans, Jonathan Tremblay, Thoma s M¨ uller, Morgan McGuire, Alec Jacobson, and Sanja Fidler. V ariable bitrate neural fields. In ACM SIGGRAPH 2022 Con- ference Proceedings, pages 1–9, 2022. 2

    2022

  29. [37]

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

    Henan Wang, Hanxin Zhu, Tianyu He, Runsen Feng, Jia- jun Deng, Jiang Bian, and Zhibo Chen. End-to-end rate- distortion optimized 3d gaussian representation. In European Conference on Computer Vision, 2024. 2, 4, 5

    2024

  30. [38]

    Neural residual radiance fields for streamably free-viewpoint videos

    Liao Wang, Qiang Hu, Qihan He, Ziyu Wang, Jingyi Y u, Tinne Tuytelaars, Lan Xu, and Minye Wu. Neural residual radiance fields for streamably free-viewpoint videos. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 76–87, 2023. 2

    2023

  31. [39]

    VideoRF: Rendering Dynamic Radiance Fields as 2D Feature Video Streams

    Liao Wang, Kaixin Yao, Chengcheng Guo, Zhirui Zhang, Qiang Hu, Jingyi Y u, Lan Xu, and Minye Wu. VideoRF: Rendering Dynamic Radiance Fields as 2D Feature Video Streams . In 2024 IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR) , pages 470–481, Los Alamit...

    2024

  32. [40]

    TeTriRF: Temporal Tri-Plane Radiance Fields for Effi - cient Free-Viewpoint Video

    Minye Wu, Zehao Wang, Georgios Kouros, and Tinne Tuyte- laars. TeTriRF: Temporal Tri-Plane Radiance Fields for Effi - cient Free-Viewpoint Video . In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 6487–6496, Los Alamitos, CA, USA, 2024. IEEE ...

    2024

  33. [41]

    0 7 . 5 9 . 0 10 . 5 Total Size (MB)

  34. [42]

    RD curves with video QP adjustment

    00 PSNR (dB) QP=8 QP=16 QP=8 QP=16 λ ent = 1e− 10 λ ent = 1e− 9 λ ent = 1e− 8 Room Figure 9. RD curves with video QP adjustment. Controlling rate-distortion with video QP is worse than using λ ent with QP=1. Performance analysis of QP adjustment. Beyond ad- justing the paramet...

  35. [43]

    (a) sphere contraction (b) piecewise contraction Figure 10

    Our results suggest this may be because video codecs focus exclusively on feature plane restoration rather than the rendering view quality. (a) sphere contraction (b) piecewise contraction Figure 10. Visualization of each contraction method for the ‘Bonsai’ scene. (a) sphere c...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.