Pith. sign in

REVIEW 4 major objections 6 minor 27 references

GSVC: Efficient Video Representation and Compression Through 2D Gaussian Splatting

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper tries to establish that 2D Gaussian splats can serve as a practical primitive for video compression, with rate-distortion performance comparable to AV1 and HEVC and decoding at over 1500 fps.

desk verdict GSVC is a real first step for 2D Gaussian video compression, but the rate accounting in Figure 6 must be pinned down before the AV1/HEVC parity claim can be trusted. read the letter →

arxiv 2501.12060 v2 pith:SQGT2XF5 submitted 2025-01-21 cs.CV cs.MM

Longan Wang , Yuang Shi , Wei Tsang Ooi This is my paper · ORCID
classification cs.CVcs.MM
keywords 2DGaussiansplattingvideocompressionrepresentationrate-distortionoptimizationtemporalpredictionkey-frameselectionsplatpruningaugmentation
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 establish that 2D Gaussian splats, already successful as an explicit primitive for static scenes and images, can also serve as an efficient primitive for video representation and compression. The proposed codec, GSVC, treats selected frames as key-frames learned from scratch and predicts the rest from the previous frame, then prunes low-contribution splats, injects new splats to absorb motion, and inserts a key-frame when a scene change is detected. On UVG clips, the paper reports rate-distortion curves comparable to AV1 and HEVC in perceptual metrics and decoding at roughly 1583 frames per second, orders of magnitude faster than neural video codecs. If the bitrate accounting is complete, this would give a practical, pretraining-free alternative to both block-based standards and neural codecs, with explicit control over rate, quality, and encoding time.

What carries the argument

The carrying object is the 2D Gaussian splat, an explicit primitive with eight parameters: position $\mu\in\mathbb{R}^2$, weighted color $c'\in\mathbb{R}^3$ (opacity multiplied into the color), and Cholesky vector $\ell\in\mathbb{R}^3$ defining the covariance $\Sigma=LL^T$. Rendering accumulates $w_n c'_n \exp(-\sigma_n)$ over splats covering a pixel, where $\sigma_n=\frac{1}{2}d_n^T\Sigma^{-1}d_n$ and $w_n$ is a learned importance weight. The mechanism has four moving parts: frame prediction reuses the previous frame's splats as initialization and stores only deltas; Gaussian Splat Pruning removes splats with small $|w_n|^2$; Gaussian Splat Augmentation inserts new splats with $w\approx 0$ that can be activated by sharp weight changes; and the Dynamic Key-frame Selector flags a frame as a key-frame when its P-frame pretraining loss exceeds the I-frame loss by the local mean plus three local standard deviations. Encoding quantizes positions to 16-bit floats, Cholesky coefficients through $b$-bit asymmetric quantization, and colors through $M$-stage residual vector quantization.

What would settle it

Re-encode Beauty, HoneyBee, and Jockey with GSVC; serialize the quantized splat parameters, codebooks, I-frame data, and P-frame deltas into an actual file, apply any entropy coder the implementation uses, and plot the resulting bits-per-pixel against the published curves. If the true file size puts GSVC's PSNR/MS-SSIM/VMAF points to the right of the AV1 or HEVC curves at equal quality, the claimed rate-distortion parity fails.

Watch

Extended reading notes

Core claim

The central claim is that a video can be represented by a set of 2D Gaussian splats per frame, and that with temporal prediction this representation compresses competitively. Each splat carries eight parameters: a 2D position, a weighted RGB color with opacity folded in, and a three-coefficient Cholesky factor for its covariance, rendered by accumulated summation of weighted Gaussian contributions. GSVC encodes an I-frame from scratch and a P-frame as quantized deltas relative to the previous frame's splats, so temporal redundancy is stored once. A learnable importance weight per splat drives pruning; randomly inserted splats with near-zero initial weights capture newly appearing objects or fast motion; and frames whose P-frame loss exceeds the local I-frame loss by more than three local standard deviations become key-frames. The paper reports that this scheme reaches AV1/HEVC-comparable MS-SSIM and VMAF, with VVC remaining stronger, and decodes 1920x1080 content at 1583.72 fps, while neural codecs in the same comparison decode at around 2.4 fps.

Load-bearing premise

The load-bearing premise is that the bits-per-pixel values plotted for GSVC in Figure 6 include every bit needed to reconstruct the video, including the quantized splat positions, Cholesky coefficients, color codebooks and their indices, I-frame data, and P-frame deltas, because the paper never states the quantization bit-width $b$, the number of RVQ stages $M$, the codebook size $B$, or whether a final entropy coder is applied.

Editorial extensions

If this is right

  • Because splats are explicit and countable, the number of splats N becomes a direct bitrate knob, the loss convergence threshold a quality knob, and the training iteration count an encoding-time knob.
  • Decoding at 1583 fps makes real-time playback and interactive editing feasible with an ordinary GPU rasterizer, without a specialized neural decoder.
  • GSVC requires no pretraining on a dataset, so it can in principle generalize to any input video, unlike the neural baselines compared in the paper.
  • The ablation ordering (GSP contributes most, then GSA, then DKS) indicates that where splats are placed matters more than the raw splat count.
  • Fine-grain progressive coding follows directly: pruned splats can be added back to refine quality without retraining from scratch.

Reading between the lines

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

  • The delta encoding used for P-frames is symmetric enough that the same machinery could support B-frames predicted from both past and future frames, which the paper names as future work; this should improve rate-distortion at the cost of latency.
  • Because splats are explicit objects, a region-of-interest mode is possible: allocate more splats where saliency or viewer attention is high, a knob block-based codecs do not expose.
  • The pruning mechanism implies a variable-rate mode: let N_t vary per frame under a quality threshold instead of holding N constant, which would mimic constant-quality coding and likely save bits on static stretches.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper introduces GSVC, a video representation and compression framework based on 2D Gaussian splats. Each frame is represented by a set of 2D Gaussians with position, color, and Cholesky parameters. An I/P-frame structure is used: key frames are trained from scratch, while P-frames are initialized from the previous frame and fine-tuned. The method combines Gaussian Splat Pruning (GSP), Gaussian Splat Augmentation (GSA), and a Dynamic Key-frame Selector (DKS) to control rate, handle motion, and detect scene changes. Encoding quantizes parameters using 16-bit float positions, b-bit asymmetric quantization for Cholesky coefficients, and M-stage RVQ codebooks of size B for colors. Experiments on three UVG sequences report rate-distortion curves against AVC, HEVC, VVC, AV1, DCVC-DC, and DCVC-HEM, plus decoding speed of 1583.72 fps. The paper claims comparable rate-distortion performance to standard codecs in MS-SSIM and VMAF, except for VVC, while offering very fast decoding.

Significance. If the rate-distortion comparison is correct, GSVC is a promising explicit, editable representation for video with a useful speed advantage over neural codecs. The method's parameter controls (number of splats, pruning, convergence) are a practical strength, and the I/P-frame formulation is a natural extension of 2D Gaussian splatting. The paper also honestly discusses limitations and positions its results as preliminary. However, the central comparison rests on a rate axis whose bit accounting is not fully specified, and the evaluation is limited to three sequences without variance estimates. These issues currently prevent the parity claim from being verified, so the significance is conditional on a corrected and fully specified bitrate computation.

major comments (4)
  1. [§3.8, Eq. (14)–(16) and Figure 6] The bitrate computation underlying Figure 6 is underspecified. The paper defines quantization operators but never states the bit-depth b for Cholesky coefficients, the number of RVQ stages M, the codebook size B, or whether a final entropy coder is applied. It also does not explain how the reported bpp counts the 16-bit float positions of I-frames, the I-frame cost relative to GOP length, the learned codebooks (which are shared and not frame-local), the P-frame deltas for all three attributes, or any side information such as pruning masks and key-frame decisions. Because the claim of parity with AV1 and HEVC depends directly on the bpp values in Figure 6, the rate axis must be specified exactly enough for a reader to reproduce the cost. Please provide the missing parameters and a precise formula for bpp, including how codebooks and I-frame overhead are amortized.
  2. [§4.1, Datasets and Implementation Details] The evaluation uses only three sequences (Beauty, HoneyBee, Jockey) and reports single-run results without error bars or multiple trials. The training procedure is stochastic (random initialization, random GSA insertion, learned weights), so the reported curves may not be stable. Please report at least three independent runs per configuration with mean and variance, or justify why single runs are sufficient for this type of comparison.
  3. [§4.2, Rate-Distortion Trade-off] The central claim that GSVC achieves 'better or comparable performance in terms of MS-SSIM and VMAF... except VVC' is stated without quantitative backing. The text also admits lower PSNR in certain scenarios. Please specify at which bitrates and for which sequences the method is better, comparable, or worse, and report the actual metric differences (e.g., BD-rate or BD-PSNR relative to each codec). Without these numbers, the qualitative statement cannot be assessed, especially given the underspecified rate axis.
  4. [§5, Discussion and Conclusion] The paper itself describes the results as 'preliminary support of evidence.' This is appropriate, but it should be reconciled with the stronger claims in the Abstract and Section 4.2. Please either temper the abstract and comparison statements to match the preliminary nature of the evidence, or provide the additional experiments and rate accounting needed to support the stronger claims.
minor comments (6)
  1. [§2, Related Work] The paragraph beginning 'Recent efforts have explored Gaussian-based representations...' appears twice verbatim; please remove the duplicate.
  2. [§4.1, Baseline] The abbreviation for the GaussianImage baseline is written 'Outs-GI' in the baseline paragraph and 'Ours-GI' in the caption of Figure 6 and elsewhere; please use one consistent spelling.
  3. [Figure 5 caption] The caption uses 'DSK' while the method is called 'DKS' in the text; clarify the intended acronym.
  4. [§4.2, Computational Cost] The text mentions 'GuassianVideos' in 'the neural-based approach achieves significant improvements over GuassianVideos'; this appears to be a typo and should be corrected (e.g., to 'GaussianImage' or 'the Gaussian-based baselines' as appropriate).
  5. [§3.8 and §4.1] The term 'bpp' is used without a formal definition in the context of GSVC; please define bits per pixel explicitly, including how the total bit count is computed from the encoded representation.
  6. [§4.3, Ablation Study] The ablation text states that 'DSK has small improvements' whereas Figure 7 appears to show the curves for variants (1), (2), (3), and Ours; please clarify which variant includes DKS and report the corresponding numerical differences, since the figure alone does not clearly separate the DKS contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: GSVC's central derivation is self-contained and its comparative evaluation rests on external codecs and metrics.

full rationale

The paper's derivation chain does not reduce to its inputs. The central claim is that a video can be represented as a set of 2D Gaussian splats with I/P-frame prediction, pruning, augmentation, and dynamic key-frame selection, and that this representation is competitive with standard codecs. The method is optimized per video against ground-truth frames using an L2 loss (Equation 13), and the rate-distortion comparison in Figure 6 is evaluated with external metrics (PSNR, MS-SSIM, VMAF) against external baselines (AVC, HEVC, VVC, AV1, DCVC-DC, DCVC-HEM). The paper does not fit a parameter and then rename that fit as a prediction: the bitrate is controlled by the splat count N, and quality is measured directly on reconstructed frames. There is no load-bearing self-citation: the cited prior work, notably GaussianImage [Zhang et al. 2025], is by other authors and supplies the 2D Gaussian parameterization and quantization-aware fine-tuning primitives, not the video-level claims of this paper. The DKS key-frame selection is not circular: it computes loss differences from pre-training all frames as scratch key-frames and as P-frames, and then selects key-frames using the 3-sigma rule before final training; the final key-frame set is not used to compute the losses that select it. The paper's own limitations, such as Section 5's statement that the results "serve as preliminary support of evidence" and Section 4.2's admission of lower PSNR in certain scenarios, are honest caveats rather than circular reasoning. The main weakness of the paper is evidential, not circular: Section 3.8 specifies quantization operators but does not state the bit-width b, the number of RVQ stages M, the codebook size B, or whether a final entropy coder is applied, leaving the bitrate accounting in Figure 6 underspecified. That is a transparency or correctness risk for the comparative claim, but it is not a case where the derivation is equivalent to its inputs by construction. Therefore the circularity score is 0.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central empirical claims rest on several hand-set hyperparameters, an unverified bitrate accounting assumption, and the transfer of image-level Gaussian rendering assumptions to video. No new physical entities are introduced.

free parameters (6)
  • Gaussian splats per frame N = 10,000 to 50,000 in 10,000 increments
    Used as the rate control knob; each rate-distortion point is produced by choosing N, so the claimed trade-off is a scan over a hand-set parameter.
  • GSP pruning fraction = 10% of N
    Section 4.1 states key-frames and P-frames prune 10% of the splats; this ratio is chosen by hand and affects the quality-size trade-off.
  • GSA augmentation fraction = 10% of N
    Section 4.1 states each P-frame starts with 90% of N and adds 10% random splats; this hand-set fraction influences how dynamic content is captured.
  • DKS window size wwin = 10
    Section 4.1 sets the local sliding window for loss-difference outlier detection; the choice affects which frames become key-frames.
  • DKS outlier threshold = 3 standard deviations
    Equation (9) flags a scene transition when the loss difference exceeds the local mean by 3 sigma; this threshold is chosen by hand without calibration.
  • Quantization bit-depth b and RVQ hyperparameters (M, B) = unspecified
    Section 3.8 defines b-bit asymmetric quantization and M-stage RVQ with codebook size B, but the values never appear. The bpp values in Figure 6 depend directly on these missing settings.
assumptions (4)
  • domain assumption The 2D Gaussian splat parameterization, rendering equation, and L2 loss form a valid differentiable image model whose optima represent frames at the reported quality.
    Sections 3.2 and 3.7 inherit this from GaussianImage and gsplat; the paper assumes it generalizes from images to video frames.
  • ad hoc to paper The learned importance weight w faithfully measures each splat's contribution, so removing splats with smallest |w|^2 preserves quality.
    Section 3.4 introduces this assumption without proof or measurement of the correlation between w-magnitude pruning and true rate-distortion loss.
  • ad hoc to paper A loss difference above a local 3-sigma threshold marks a scene transition that requires an I-frame.
    Section 3.6, Equation (9): the threshold and window are chosen by hand; no evidence shows this criterion is calibrated across varied content.
  • domain assumption The bitstream formed by quantized full parameters and deltas in Equation (14) contains all information needed to reconstruct frames, and bpp is measured from this bitstream.
    Section 3.8 never states how bits are accumulated, so this unverified measurement assumption is load-bearing for the rate-distortion claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GSVC: Efficient Video Representation and Compression Through 2D Gaussian Splatting." pith.science (2026). https://pith.science/paper/SQGT2XF5

@misc{pith2026250112060,
  author       = {Pith},
  title        = {Pith review of: GSVC: Efficient Video Representation and Compression Through 2D Gaussian Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SQGT2XF5}},
  note         = {Machine review of arXiv:2501.12060}
}
read the original abstract

3D Gaussian splats have emerged as a revolutionary, effective, learned representation for static 3D scenes. In this work, we explore using 2D Gaussian splats as a new primitive for representing videos. We propose GSVC, an approach to learning a set of 2D Gaussian splats that can effectively represent and compress video frames. GSVC incorporates the following techniques: (i) To exploit temporal redundancy among adjacent frames, which can speed up training and improve the compression efficiency, we predict the Gaussian splats of a frame based on its previous frame; (ii) To control the trade-offs between file size and quality, we remove Gaussian splats with low contribution to the video quality; (iii) To capture dynamics in videos, we randomly add Gaussian splats to fit content with large motion or newly-appeared objects; (iv) To handle significant changes in the scene, we detect key frames based on loss differences during the learning process. Experiment results show that GSVC achieves good rate-distortion trade-offs, comparable to state-of-the-art video codecs such as AV1 and VVC, and a rendering speed of 1500 fps for a 1920x1080 video.

Figures

Figures reproduced from arXiv: 2501.12060 by the authors.

Figure 1
Figure 1. Using Gaussian splats to model an image. Image is taken from UVG [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Image taken from UVG Jockey video. Intermediate training results after t iterations for 10,000 Gaussian splats, illustrating how Gaussian splats parameters are optimized to fit the content of a frame. • We present experiment results, demonstrating that GSVC achieves a rate-distortion performance comparable to state-of-the-art video compression standards while allowing fast decoding speeds of up to 1500 FPS. 2 Relate… view at source ↗
Figure 3
Figure 3. Distribution of Gaussian Splat Centers ( [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Distribution of Gaussian Splats (45k) on f5 in Beauty. 3.5 Dynamic Content During training, the Gaussian splats of P-frames adapt their parameters to changes from the previous frames. This approach, however, can fail to capture highly dynamic content. For example, a ne…
Figure 5
Figure 5. Figure 5: ∆ℓ and PSNR over frames on the concatenated videos [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Rate-Distortion Curves in PSNR, MS-SSIM and VMAF: comparison of our approach with baselines [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Ablation Study of PSNR and MS-SSIM over Gaussian Number on UVG dataset. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 23 canonical work pages

  1. [1]

    [Bhalgat et al.(2020)] Yash Bhalgat, Jinwon Lee, Markus Nagel, Tijmen Blankevoort, and Nojun Kwak

    work page 2020

  2. [2]

    [Wiegand et al.(2003)] Thomas Wiegand, Gary J Sullivan, Gisle Bjontegaard, and Ajay Luthra

    Ieee, 1398–1402. [Wiegand et al.(2003)] Thomas Wiegand, Gary J Sullivan, Gisle Bjontegaard, and Ajay Luthra

    work page 2003

  3. [3]

    APSIPA Transactions on Signal and Information Processing 9 (2020), e6

    An overview of coding tools in A V1: the first video codec from the alliance for open media. APSIPA Transactions on Signal and Information Processing 9 (2020), e6. [Gray(1984)] Robert Gray

    work page 2020

  4. [8]

    InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023, V ancouver , Canada, June 18-22,

    Neural Video Compression with Diverse Contexts. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023, V ancouver , Canada, June 18-22,

    work page 2023

  5. [9]

    [Li et al.(2024)] Jiahao Li, Bin Li, and Yan Lu

    work page 2024

  6. [12]

    In Proceedings of the 11th ACM Multimedia Systems Con- ference

    UVG dataset: 50/120fps 4K se- quences for video codec analysis and development. In Proceedings of the 11th ACM Multimedia Systems Con- ference. 297–302. 10 GSVC: Efficient Video Representation and Compression Through 2D Gaussian Splatting A PREPRINT [Mildenhall et al.(2021)] Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Rama...

    work page 2021

  7. [14]

    https://github.com/Netflix/ vmaf.git

    VMAF - Video Multi-Method Assessment Fusion. https://github.com/Netflix/ vmaf.git. Accessed: 2025-01-06. [Paszke et al.(2019)] Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al

    work page 2019

  8. [15]

    Advances in neural information processing systems 32 (2019)

    PyTorch: An imperative style, high- performance deep learning library. Advances in neural information processing systems 32 (2019). [Smolak-Dy˙zewska et al.(2024)] Weronika Smolak-Dy ˙zewska, Dawid Malarz, Kornel Howil, Jan Kaczmarczyk, Marcin Mazur, and Przemysław Spurek

    work page 2019

Show all 27 references
  1. [16]

    arXiv preprint arXiv:2411.11024 (2024)

    VeGaS: Video Gaussian Splatting. arXiv preprint arXiv:2411.11024 (2024). [Sullivan(2020)] Gary Sullivan

    2024

  2. [17]

    In 2020 IEEE International Conference on Visual Communications and Image Processing (VCIP)

    Versatile video coding (VVC) arrives. In 2020 IEEE International Conference on Visual Communications and Image Processing (VCIP) . IEEE, 1–1. [Sullivan et al.(2012)] Gary J Sullivan, Jens-Rainer Ohm, Woo-Jin Han, and Thomas Wiegand

    2012

  3. [19]

    arXiv preprint arXiv:2406.13870 (2024)

    Splatter a Video: Video Gaussian Representation for Versatile Processing. arXiv preprint arXiv:2406.13870 (2024). [Wang et al.(2003)] Zhou Wang, Eero P Simoncelli, and Alan C Bovik

    2024 arXiv

  4. [22]

    264/A VC video coding standard

    Overview of the H. 264/A VC video coding standard. IEEE Transactions on circuits and systems for video technology 13, 7 (2003), 560–576. [Xie et al.(2024)] Xingyu Xie, Pan Zhou, Huan Li, Zhouchen Lin, and Shuicheng Yan

    2003

  5. [23]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Adan: Adaptive nesterov momentum algorithm for faster optimizing deep models. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024). [Ye et al.(2024)] Vickie Ye, Ruilong Li, Justin Kerr, Matias Turkulainen, Brent Yi, Zhuoyang Pan, Otto Seiskari, Jianbo Ye, Jeff...

    2024

  6. [24]

    arXiv preprint arXiv:2409.06765 (2024)

    gsplat: An Open-Source Library for Gaussian Splatting. arXiv preprint arXiv:2409.06765 (2024). arXiv:2409.06765 [cs.CV] https://arxiv. org/abs/2409.06765 [Zeghidour et al.(2021)] Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi

    2024 arXiv

  7. [25]

    IEEE/ACM Transactions on Audio, Speech, and Lan- guage Processing 30 (2021), 495–507

    SoundStream: An end-to-end neural audio codec. IEEE/ACM Transactions on Audio, Speech, and Lan- guage Processing 30 (2021), 495–507. [Zhang et al.(2025)] Xinjie Zhang, Xingtong Ge, Tongda Xu, Dailan He, Yan Wang, Hongwei Qin, Guo Lu, Jing Geng, and Jun Zhang

    2021

  8. [27]

    arXiv preprint arXiv:2407.01866 (2024)

    Image-GS: Content-Adaptive Image Representation via 2D Gaussians. arXiv preprint arXiv:2407.01866 (2024). 11

    2024 arXiv

  9. [1984]

    IEEE Assp Magazine 1, 2 (1984), 4–29

    Vector quantization. IEEE Assp Magazine 1, 2 (1984), 4–29. [Higham(2009)] Nicholas J Higham

    1984

  10. [2003]

    In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003 , V ol

    Multiscale structural similarity for image quality assessment. In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003 , V ol

    2003

  11. [2009]

    Wiley interdisciplinary reviews: computational statistics 1, 2 (2009), 251–254

    Cholesky factorization. Wiley interdisciplinary reviews: computational statistics 1, 2 (2009), 251–254. [Kerbl et al.(2023)] Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis

    2009

  12. [2012]

    IEEE Transactions on circuits and systems for video technology 22, 12 (2012), 1649–1668

    Overview of the high efficiency video coding (HEVC) standard. IEEE Transactions on circuits and systems for video technology 22, 12 (2012), 1649–1668. [Sun et al.(2024)] Yang-Tian Sun, Yi-Hua Huang, Lin Ma, Xiaoyang Lyu, Yan-Pei Cao, and Xiaojuan Qi

    2012

  13. [2019]

    In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    DVC: An end-to-end deep video compression framework. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11006–11015. [Mercat et al.(2020)] Alexandre Mercat, Marko Viitanen, and Jarno Vanne

    2020

  14. [2020]

    In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops

    LSQ+: Improving low-bit quantization through learnable offsets and better initialization. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops . 696–697. [Chen et al.(2020)] Yue Chen, Debargha Mukherjee, Jingning Han, Adrian Grange, Ya...

    2020

  15. [2021]

    NeRF: Representing scenes as neural radiance fields for view synthesis. Commun. ACM 65, 1 (2021), 99–106. [Netflix(2023)] Netflix

    2021

  16. [2022]

    In Proceedings of the 30th ACM International Conference on Multimedia

    Hybrid Spatial-Temporal Entropy Modelling for Neural Video Compression. In Proceedings of the 30th ACM International Conference on Multimedia . [Li et al.(2023)] Jiahao Li, Bin Li, and Yan Lu

    2023

  17. [2023]

    ACM Trans

    3D Gaus- sian Splatting for Real-Time Radiance Field Rendering. ACM Trans. Graph. 42, 4 (2023), 139–1. [Li et al.(2022)] Jiahao Li, Bin Li, and Yan Lu

    2023

  18. [2024]

    In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Neural video compression with feature modulation. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 26099–26108. [Lu et al.(2019)] Guo Lu, Wanli Ouyang, Dong Xu, Xiaoyun Zhang, Chunlei Cai, and Zhiyong Gao

    2019

  19. [2025]

    In European Conference on Computer Vision

    GaussianImage: 1000 fps image representation and compression by 2d gaussian splatting. In European Conference on Computer Vision. Springer, 327–345. [Zhang et al.(2024)] Yunxiang Zhang, Alexandr Kuznetsov, Akshay Jindal, Kenneth Chen, Anton Sochenov, Anton Kaplanyan, and Qi Sun

    2024

Pith tools

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