Pith. sign in

REVIEW 4 major objections 5 minor 36 references

4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction

T0 review · 4 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 The paper claims that compressing 4D Gaussian scenes into a sparse grid of anchors, rather than pruning Gaussians, preserves dynamic-region fidelity while cutting storage to roughly one fortieth of a full 4D Gaussian model.

desk verdict Genuinely new 4D anchor-based dynamic Gaussian method with a real storage-quality win, but the headline 'large margin' rests on numbers that don't reconcile internally—worth referee time, not acceptance yet. read the letter →

arxiv 2411.17044 v2 pith:Z2ZB5GOH submitted 2024-11-26 cs.CV cs.GR

Woong Oh Cho , In Cho , Seoha Kim , Jeongmin Bae , Youngjung Uh , Seon Joo Kim This is my paper · ORCID
classification cs.CVcs.GR
keywords dynamicscenereconstruction4DGaussiansplattinganchor-basedrepresentationdynamic-awaredensificationgeneralizedtemporalopacityneuralrenderingmulti-viewvideonovelviewsynthesis
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

Dynamic scene reconstruction with 4D Gaussians renders fast and looks sharp but stores gigabytes per short video. Previous fixes cut storage by removing Gaussians, which destroys moving content. This paper instead keeps the Gaussians and compresses their descriptions into a sparse, grid-aligned set of 4D anchor features, each anchor spawning a small set of neural 4D Gaussians via shared MLPs. To keep short-lived moving objects from being starved of representation, it reweights each Gaussian's accumulated gradient by how briefly it is visible, so new anchors grow where dynamic content is under-reconstructed. On the N3DV and Technicolor datasets the method reports the best dynamic-region quality among the compared baselines at roughly 149 MB of storage, compared with about 6.2 GB for a full 4D Gaussian model.

What carries the argument

The load-bearing object is the 4D anchor: a voxelized position $p = (x_v, y_v, z_v, t_0)$ with a 32-dimensional feature vector and $K$ learnable 4D offsets. Shared MLPs decode the feature into $K$ neural 4D Gaussians, each with base opacity $\rho$, quaternion $q$, scale $s$, color $c$, temporal scale $\sigma$, and velocity $u$. Temporal behavior uses linear motion $\mu_k = x_k^{xyz} + (t - x_k^t)u$ and the generalized Gaussian opacity $g(t, x_k^t, \sigma_k) = \exp(-(|t - x_k^t|/\sigma_k)^\beta)$. The anchor growing rule accumulates gradients with weight $w(\alpha', \sigma) = \alpha'(1/\sigma)^\gamma$, amplifying gradients from short-lived Gaussians and placing new anchors where dynamic content is under-reconstructed. These pieces together let a sparse anchor grid stand in for a dense 4D Gaussian volume.

What would settle it

Run a controlled comparison on a scene with a fast, short-lived object far from the initial reconstruction points, keeping all settings identical except the growing rule: use the proposed temporal weighting versus the same training with no temporal weighting, and check whether the weighted version places more anchors along the object's trajectory and improves dynamic-region PSNR; if it does not, the central growing mechanism is not doing the claimed work.

Watch

Extended reading notes

Core claim

The paper's central claim is that the conflict between Gaussian count and storage cost is not fundamental when per-Gaussian parameters are compressed into a sparse anchor grid. Each local spatiotemporal neighborhood is represented by a grid-aligned anchor with a 32-dimensional feature, and small shared MLPs decode that feature into $K$ Gaussians whose positions move in straight line segments and whose temporal opacity is a generalized Gaussian function. This parametrization lets a single Gaussian cover a sudden appearance change that would previously require a mixture of univariate Gaussians, so a sufficient number of Gaussians can be kept without a proportional storage bill. The dynamic-aware growing rule $w(\alpha', \sigma) = \alpha'(1/\sigma)^\gamma$ then directs new anchors toward regions whose Gaussians are active for only a few frames. On its own terms, the result is that the anchor-based renderer achieves the best dynamic-region quality among the compared baselines on both N3DV and Technicolor while storing hundreds of megabytes instead of gigabytes.

Load-bearing premise

The load-bearing premise is that the temporal-coverage weighting in the growing rule sends new anchors exactly to under-reconstructed moving regions without draining resources from the static background; the paper shows the balance depends on a hyperparameter, so this allocation heuristic is not guaranteed to transfer to other scene types.

Editorial extensions

If this is right

  • Dynamic-region PSNR on N3DV reaches 28.86 dB at 149 MB, while the full 4DGS baseline achieves 27.65 dB at 6194 MB, so the storage-quality trade-off can be shifted without pruning Gaussians.
  • On Technicolor the method reports the best full-region metrics among the compared baselines, with a 34.11 dB PSNR at 278 MB.
  • The Ours-light variant shows that relaxing to a larger voxel size still gives a compact model whose dynamic-region quality is competitive with the efficient baselines.
  • The Scaff-naive ablation shows that extending static scaffolding to 4D without temporal-coverage-aware growing fails in moving regions, making the dynamic-aware rule load-bearing for the result.
  • Storage is decoupled from Gaussian count: the model keeps more Gaussians than storage-efficient baselines while remaining smaller on disk.

Reading between the lines

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

  • Because storage scales with the number of anchors rather than the number of Gaussians, the same construction could plausibly extend to longer or higher-resolution videos by adding temporal anchors instead of multiplying Gaussians; the paper does not test this.
  • The generalized-Gaussian temporal opacity is a transferable ingredient that could be dropped into deformation-based dynamic splatting models to cover abrupt appearance changes with fewer Gaussians, but that transfer is not demonstrated here.
  • The same gradient-reweighting idea may apply to other compressed 4D representations such as feature grids or hash encodings whenever brief appearances are systematically under-densified; this is an extrapolation from the paper's growing rule.
  • The acknowledged weakness on objects that appear for only one or two frames suggests a practical ceiling set by the temporal grid resolution, and closing that gap would need a mechanism beyond anchor growing.
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 / 5 minor

Summary. The paper proposes a 4D anchor-based Gaussian representation for dynamic scene reconstruction. Anchors are placed on a sparse 4D grid and shared MLPs generate K neural 4D Gaussians per anchor, with a compact parametrization consisting of linear motion and a generalized-Gaussian temporal opacity. A dynamic-aware anchor growing strategy, which weights accumulated gradients by temporal opacity and inverse temporal scale, is introduced to allocate anchors to under-reconstructed dynamic regions. The method is evaluated on N3DV and Technicolor against deformation-based and 4D-Gaussian baselines, and the authors report state-of-the-art dynamic-region quality with substantially reduced storage (149 MB vs 6.2 GB for 4DGS on N3DV).

Significance. If the reported numbers are correct, the central idea that storage can be reduced by compressing a large number of Gaussians into 4D anchor features rather than pruning them is a valuable contribution, and the dynamic-aware growing rule is a plausible mechanism for preserving dynamic-region fidelity. The paper includes broad benchmark comparisons, ablations of K, beta, and gamma, a quality-storage trade-off analysis, and qualitative evidence for anchor growing. However, the headline N3DV dynamic-region numbers are internally inconsistent across Table 1, Table 3, and Table S4, no error bars or multiple runs are reported, and the code has not been released. These issues currently prevent verification of the claimed 'large margin' over baselines.

major comments (4)
  1. [Experiments, Table 1 vs Table 3] The same full configuration, 'Ours' with dynamic-aware growing and 149 MB storage, is reported as achieving 28.86 dB dynamic-region PSNR and 0.054 LPIPS in Table 1, but 29.57 dB and 0.050 LPIPS in Table 3. The caption of Table 3 does not state which dataset or scene subset is used, and the ablation section does not explain the discrepancy. Since the abstract's claim of a 'large margin' is quantified from these dynamic-region numbers, the inconsistency is load-bearing; the authors must reconcile the two tables or explicitly state that Table 3 uses a different subset.
  2. [Supplementary, Table S4] The per-scene dynamic-region PSNR values for Ours in Table S4 (27.07, 27.39, 31.80, 24.82, 27.84, 31.44) average to 28.39 dB, which matches neither Table 1's 28.86 dB nor Table 3's 29.57 dB. Similarly, the per-scene LPIPS values average to about 0.057, not 0.054. If Table 1 is a pixel-weighted or otherwise pooled average, the pooling rule must be stated; otherwise the per-scene values imply that one or more of the headline numbers is incorrect. This directly affects the reported margin over 4DGS, which is 0.74, 1.21, or 1.92 dB depending on which number is used.
  3. [Method, Eq. (8)] Equation (8) is not printed unambiguously: the numerator and denominator appear to be identical as rendered, which would make the dynamic-aware gradient weighting undefined. Please write the expression explicitly, e.g., with a sum over N of w(alpha', sigma) times the gradient norm in the numerator and a sum over N of w(alpha', sigma) in the denominator, and clarify whether alpha' is the time-variant opacity component g(...) only or the total opacity alpha_k = rho_k * g(...). This equation is central to the anchor-growing contribution, so it must be checkable.
  4. [Ablations and hyperparameters, Figure S3] The paper's own Figure S3 shows that the dynamic-aware growing parameter gamma is a compromise: larger gamma improves dynamic-region quality but degrades static-region quality, and the authors select gamma = 1 empirically. Because the headline claim of large dynamic-region gains rests on this heuristic, the manuscript should provide a clearer selection principle or sensitivity analysis beyond the two N3DV scenes, and should state whether gamma was chosen using the evaluation scenes themselves. Without this, transferability to other scene types remains an unquantified risk.
minor comments (5)
  1. [Table 2] Grid4D is listed in the baselines paragraph but is absent from the Technicolor results in Table 2; please either add the missing numbers or note why it is omitted.
  2. [Table S2] The caption of Table S2 reports results for 'cook spinach and flame salmon' but does not specify whether the PSNR is dynamic-region or full-region; the values should be labeled consistently with Tables 1 and S4.
  3. [Method, Eq. (6)] The text says beta = 2*beta' with beta' as a hyperparameter and then states 'we set beta = 2'; it would be clearer to state the effective exponent directly and to note that this choice makes Eq. (6) a standard Gaussian except for the parameterization.
  4. [Experiments, metrics] The dynamic-region mask is described as a binarized combination of Global-Median and Temporal-Difference, but the threshold of 50 is not specified in units or in relation to the combined mask; please clarify the mask construction.
  5. [General] The statement 'Our code will be made publicly available' appears without a URL or release plan; given that the central comparison depends on exact numbers, providing the code or a detailed evaluation script would substantially aid verification.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the method is an empirically evaluated anchor-based 4D Gaussian system, and its novel growing heuristic is a tuned design choice, not a fitted restatement of the reported metrics.

full rationale

The paper's central contribution is a 4D anchor-based representation with a dynamic-aware anchor growing rule (Eqs. 8-9). This rule is introduced as a heuristic that reweights accumulated gradients by temporal coverage; it is not constructed from the reported dynamic-region PSNR/LPIPS values, and the final numbers are measured on a held-out test view after training on the remaining views. The initialization from SfM points at t0 and the subsequent anchor growing are mechanisms whose outputs are evaluated empirically against external baselines, not derived from the target metrics by construction. Hyperparameters such as beta, gamma, K, and grid sizes are tuned on the same datasets, which is a standard experimental limitation, but the evaluation metrics are not reconstructed from those hyperparameters in a way that would force the reported outcomes. The paper includes self-citations (e.g., Bae et al. 2024, with overlapping authorship), but these are related-work and metric-convention references, not load-bearing justifications of the central claim. The limitations statement acknowledges poor reconstruction of elements appearing for only 1-2 frames, which is an honest scope restriction rather than a circular step. One notable concern is that the reported N3DV dynamic-region numbers are internally inconsistent (Table 1: 28.86 dB; Table 3: 29.57 dB; Table S4 per-scene values average to about 28.39 dB). This weakens the reliability of the 'large margin' claim, but it is an evidence-consistency problem, not a circularity problem, because no derivation reduces the claim to its own inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 2 invented entities

The central claim rests on modeling assumptions about the expressiveness of anchor-spawned Gaussians and on the heuristic gradient weighting for anchor growth. Free hyperparameters β, γ, K, and grid sizes are tuned on the same benchmarks and materially affect the reported tradeoff.

free parameters (5)
  • β (generalized Gaussian exponent) = 2
    Controls steepness of temporal opacity; tuned on N3DV/Technicolor; higher values reduce anchor counts but above 8 destabilize training (Figure S4).
  • γ (gradient weight exponent) = 1
    Controls influence of temporal scale on densification; ablation Figure S3 shows it trades dynamic vs static quality; empirically set to 1.
  • K (Gaussians per anchor) = 10
    Number of neural Gaussians spawned per anchor; follows Scaffold-GS and is ablated in Table S2; larger K improves quality and reduces anchors.
  • Spatial voxel grid size = 0.001
    Anchor grid spacing in xyz; controls anchor density and storage; adjusted for Ours-light.
  • Temporal grid size = 0.0333 (N3DV), 0.02 (Technicolor)
    Frame interval used as temporal anchor spacing; set per dataset.
assumptions (5)
  • domain assumption A sparse 4D grid of anchors plus shared MLPs can generate Gaussians that accurately represent local spatiotemporal regions.
    Core representational premise; no proof, only empirical validation across two datasets.
  • domain assumption Linear motion segments and generalized-Gaussian temporal opacity are expressive enough to capture real-world dynamics.
    Section 'Compact Parametrization'; ablations show benefit over polynomial and univariate Gaussian on the tested scenes, but no general guarantee.
  • ad hoc to paper Weighting gradients by α'(1/σ)^γ grows anchors in under-reconstructed dynamic regions without harming static regions.
    Eqs. 8-9; this heuristic is specific to the paper and the γ=1 choice is a compromise (Figure S3).
  • domain assumption Initial anchors from SfM points at t0 can be grown into all relevant dynamic regions.
    Eq. 1 and Figure S5; initial anchors come only from the first frame's point cloud; depends on growth reaching later-appearing objects.
  • standard math The 3D Gaussian splatting rasterizer is differentiable and supports the gradient-based densification.
    Inherited from Kerbl et al. 2023 and Scaffold-GS; used as the rendering backbone.
invented entities (2)
  • 4D anchor grid
    purpose: A sparse grid-aligned set of feature vectors in space-time from which neural 4D Gaussians are spawned, compressing storage.
    Introduced in this paper as the central compression structure; its validity is measured only through the method's own rendering, with no external falsifiable handle.
  • Neural 4D Gaussians with linear motion and generalized-Gaussian temporal opacity
    purpose: Compact primitive for representing local spatiotemporal appearance with few parameters.
    A model component, not an observable entity; no independent evidence outside the paper's evaluations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction." pith.science (2026). https://pith.science/paper/Z2ZB5GOH

@misc{pith2026241117044,
  author       = {Pith},
  title        = {Pith review of: 4D Scaffold Gaussian Splatting with Dynamic-Aware Anchor Growing for Efficient and High-Fidelity Dynamic Scene Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z2ZB5GOH}},
  note         = {Machine review of arXiv:2411.17044}
}
read the original abstract

Modeling dynamic scenes through 4D Gaussians offers high visual fidelity and fast rendering speeds, but comes with significant storage overhead. Recent approaches mitigate this cost by aggressively reducing the number of Gaussians. However, this inevitably removes Gaussians essential for high-quality rendering, leading to severe degradation in dynamic regions. In this paper, we introduce a novel 4D anchor-based framework that tackles the storage cost in different perspective. Rather than reducing the number of Gaussians, our method retains a sufficient quantity to accurately model dynamic contents, while compressing them into compact, grid-aligned 4D anchor features. Each anchor is processed by an MLP to spawn a set of neural 4D Gaussians, which represent a local spatiotemporal region. We design these neural 4D Gaussians to capture temporal changes with minimal parameters, making them well-suited for the MLP-based spawning. Moreover, we introduce a dynamic-aware anchor growing strategy to effectively assign additional anchors to under-reconstructed dynamic regions. Our method adjusts the accumulated gradients with Gaussians' temporal coverage, significantly improving reconstruction quality in dynamic regions. Experimental results highlight that our method achieves state-of-the-art visual quality in dynamic regions, outperforming all baselines by a large margin with practical storage costs.

Figures

Figures reproduced from arXiv: 2411.17044 by the authors.

Figure 1
Figure 1. Dynamic region quality comparison. (left) Comparison of dynamic region quality (y-axis) versus storage cost (x￾axis) on N3DV. (right) Results on the flame salmon scene from N3DV. Scaff-naive refers to the naive anchor-based model without our neural Gaussian design and dynamic-aware anchor growing. Our method significantly outperforms all baselines with efficient storage usage, while other methods either exhibit degr… view at source ↗
Figure 2
Figure 2. Proposed 4D anchor-based framework. We begin with a sparse 4D grid of anchors, each defined by a unique 4D spatiotemporal position p and learnable offsets. Shared MLPs utilize these anchors to generate neural 4D Gaussians, capturing dynamic appearance changes with view direction d. These 4D Gaussians are then projected to 3D Gaussians at specific time tr for rendering, using a 3D Gaussian splatting pipeline to produ… view at source ↗
Figure 3
Figure 3. Illustration of the modified temporal opacity. To reconstruct the sudden and continuous appearance of an object, the univariate Gaussian function requires the sum of multiple Gaussians, while the generalized Gaussian achieves the same expressiveness with only a single Gaussian. Linear motion. As described in Eq. (3), the time-varying 3D position of the 3D Gaussian µk ∈ R 3 is determined by the time-varying function … view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons on the N3DV dataset. Our method achieves high-fidelity in both static and dynamic regions. 20 40 60 80 100 120 140 160180 Storage (MB) 25.5 26.0 26.5 27.0 27.5 28.0 PSNR (dB) Scaffold-naive 200K Ex4DGS 47K STG 30K 31K 58K 54K 101K 182K 231K 357K…
Figure 5
Figure 5. Figure 5: Quality-storage trade-off. The results on the flame steak scene of N3DV, along with the number of Gaus￾sians, are reported. We increase the number of employed Gaussians of our model by adjusting the anchor grid size. to capture complex scene dynamics (see the boxed reg…
Figure 6
Figure 6. Figure 6: Anchor growing analysis. Top-to-bottom: ac￾cumulated gradients at 5000th training iteration, final 4D anchors at 106 frame, and the rendering results. Our an￾chor growing effectively accumulates gradients in under￾reconstructed dynamic regions. sign lead to a substanti…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 20 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Attal, B.; Huang, J.-B.; Richardt, C.; Zollhoefer, M.; Kopf, J.; O’Toole, M.; and Kim, C. 2023. HyperReel: High-fidelity 6-DoF video with ray-conditioned sampling. In CVPR, 16610--16620

    work page 2023

  4. [4]

    Bae, J.; Kim, S.; Yun, Y.; Lee, H.; Bang, G.; and Uh, Y. 2024. Per-Gaussian Embedding-Based Deformation for Deformable 3D Gaussian Splatting. In ECCV

    work page 2024

  5. [5]

    Chen, A.; Xu, Z.; Zhao, F.; Zhang, X.; Xiang, F.; Yu, J.; and Su, H. 2021. Mvsnerf: Fast generalizable radiance field reconstruction from multi-view stereo. In ICCV, 14124--14133

    work page 2021

  6. [6]

    Chen, Y.; Wu, Q.; Lin, W.; Harandi, M.; and Cai, J. 2025. Hac: Hash-grid assisted context for 3d gaussian splatting compression. In ECCV, 422--438. Springer

    work page 2025

  7. [7]

    Deng, T.; Chen, Y.; Zhang, L.; Yang, J.; Yuan, S.; Liu, J.; Wang, D.; Wang, H.; and Chen, W. 2024. Compact 3d gaussian splatting for dense visual slam. arXiv preprint arXiv:2403.11247

    arXiv 2024

  8. [8]

    Fan, Z.; Wang, K.; Wen, K.; Zhu, Z.; Xu, D.; and Wang, Z. 2023. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps. arXiv preprint arXiv:2311.17245

    arXiv 2023

Show all 36 references
  1. [9]

    Girish, S.; Gupta, K.; and Shrivastava, A. 2023. Eagles: Efficient accelerated 3d gaussians with lightweight encodings. arXiv preprint arXiv:2312.04564

    2023 arXiv

  2. [10]

    Jiawei, X.; Zexin, F.; Jian, Y.; and Jin, X. 2024. Grid4D : 4D Decomposed Hash Encoding for High-Fidelity Dynamic Scene Rendering. The Thirty-eighth Annual Conference on Neural Information Processing Systems

    2024

  3. [11]

    Kerbl, B.; Kopanas, G.; Leimk \"u hler, T.; and Drettakis, G. 2023. 3D Gaussian Splatting for Real-Time Radiance Field Rendering. ACM TOG, 42(4)

    2023

  4. [12]

    Krizhevsky, A.; Sutskever, I.; and Hinton, G. E. 2012. Imagenet classification with deep convolutional neural networks. NeurIPS, 25

    2012

  5. [13]

    Y.; Cheong, W.-S.; Oh, J.; and Kim, M

    Kwak, S.; Kim, J.; Jeong, J. Y.; Cheong, W.-S.; Oh, J.; and Kim, M. 2025. MoDec-GS: Global-to-Local Motion Decomposition and Temporal Interval Adjustment for Compact Dynamic 3D Gaussian Splatting. In CVPR

    2025

  6. [14]

    Lee, J.; Won, C.; Jung, H.; Bae, I.; and Jeon, H.-G. 2024 a . Fully Explicit Dynamic Guassian Splatting. In NeurIPS

    2024

  7. [15]

    C.; Rho, D.; Sun, X.; Ko, J

    Lee, J. C.; Rho, D.; Sun, X.; Ko, J. H.; and Park, E. 2024 b . Compact 3D Gaussian Representation for Radiance Field. In CVPR, 21719--21728

    2024

  8. [16]

    C.; Rho, D.; Sun, X.; Ko, J

    Lee, J. C.; Rho, D.; Sun, X.; Ko, J. H.; and Park, E. 2024 c . Compact 3D Gaussian Splatting for Static and Dynamic Radiance Fields. arXiv preprint arXiv:2408.03822

    2024 arXiv

  9. [17]

    Li, T.; Slavcheva, M.; Zollhoefer, M.; Green, S.; Lassner, C.; Kim, C.; Schmidt, T.; Lovegrove, S.; Goesele, M.; Newcombe, R.; et al. 2022. Neural 3d video synthesis from multi-view video. In CVPR

    2022

  10. [18]

    Li, Z.; Chen, Z.; Li, Z.; and Xu, Y. 2023. Spacetime Gaussian Feature Splatting for Real-Time Dynamic View Synthesis. arXiv preprint arXiv:2312.16812

    2023 arXiv

  11. [19]

    Lu, T.; Yu, M.; Xu, L.; Xiangli, Y.; Wang, L.; Lin, D.; and Dai, B. 2024 a . Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In CVPR, 20654--20664

    2024

  12. [20]

    Lu, T.; Yu, M.; Xu, L.; Xiangli, Y.; Wang, L.; Lin, D.; and Dai, B. 2024 b . Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In CVPR, 20654--20664

    2024

  13. [21]

    Niedermayr, S.; Stumpfegger, J.; and Westermann, R. 2024. Compressed 3d gaussian splatting for accelerated novel view synthesis. In CVPR, 10349--10358

    2024

  14. [22]

    Papantonakis, P.; Kopanas, G.; Kerbl, B.; Lanvin, A.; and Drettakis, G. 2024. Reducing the Memory Footprint of 3D Gaussian Splatting. Proceedings of the ACM on Computer Graphics and Interactive Techniques, 7(1): 1--17

    2024

  15. [23]

    Reizenstein, J.; Shapovalov, R.; Henzler, P.; Sbordone, L.; Labatut, P.; and Novotny, D. 2021. Common Objects in 3D: Large-Scale Learning and Evaluation of Real-life 3D Category Reconstruction. arXiv:2109.00512

    2021 arXiv

  16. [24]

    Sabater, N.; Boisson, G.; Vandame, B.; Kerbiriou, P.; Babon, F.; Hog, M.; Gendrot, R.; Langlois, T.; Bureller, O.; Schubert, A.; et al. 2017. Dataset and pipeline for multi-view light-field video. In Proceedings of the IEEE conference on computer vision and pattern recognition...

    2017

  17. [25]

    L.; and Frahm, J.-M

    Sch\" o nberger, J. L.; and Frahm, J.-M. 2016. Structure-from-Motion Revisited. In CVPR

    2016

  18. [26]

    Shaw, R.; Nazarczuk, M.; Song, J.; Moreau, A.; Catley-Chandar, S.; Dhamo, H.; and P \'e rez-Pellitero, E. 2024. Swings: sliding windows for dynamic 3D gaussian splatting. In ECCV. ECCV

    2024

  19. [27]

    V.; Wang, P.; Chen, X.; Chen, T.; Venugopalan, S.; and Wang, Z

    T, M. V.; Wang, P.; Chen, X.; Chen, T.; Venugopalan, S.; and Wang, Z. 2023. Is Attention All That Ne RF Needs? In The Eleventh International Conference on Learning Representations

    2023

  20. [28]

    Wang, F.; Tan, S.; Li, X.; Tian, Z.; and Liu, H. 2022. Mixed neural voxels for fast multi-view video synthesis. arXiv preprint arXiv:2212.00190

    2022 arXiv

  21. [29]

    T.; Martin-Brualla, R.; Snavely, N.; and Funkhouser, T

    Wang, Q.; Wang, Z.; Genova, K.; Srinivasan, P.; Zhou, H.; Barron, J. T.; Martin-Brualla, R.; Snavely, N.; and Funkhouser, T. 2021. IBRNet: Learning Multi-View Image-Based Rendering. In CVPR

    2021

  22. [30]

    Wu, G.; Yi, T.; Fang, J.; Xie, L.; Zhang, X.; Wei, W.; Liu, W.; Tian, Q.; and Xinggang, W. 2023. 4D Gaussian Splatting for Real-Time Dynamic Scene Rendering. arXiv preprint arXiv:2310.08528

    2023 arXiv

  23. [31]

    Wu, M.; and Tuytelaars, T. 2024. Implicit gaussian splatting with efficient multi-level tri-plane representation. arXiv preprint arXiv:2408.10041

    2024 arXiv

  24. [32]

    Xie, S.; Zhang, W.; Tang, C.; Bai, Y.; Lu, R.; Ge, S.; and Wang, Z. 2025. MesonGS: Post-training Compression of 3D Gaussians via Efficient Attribute Transformation. In ECCV, 434--452. Springer

    2025

  25. [33]

    Yang, Z.; Gao, X.; Zhou, W.; Jiao, S.; Zhang, Y.; and Jin, X. 2023. Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction. arXiv preprint arXiv:2309.13101

    2023 arXiv

  26. [34]

    Yang, Z.; Yang, H.; Pan, Z.; and Zhang, L. 2024. Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting. In ICLR

    2024

  27. [35]

    Yu, A.; Ye, V.; Tancik, M.; and Kanazawa, A. 2021. pixelNeRF : Neural Radiance Fields from One or Few Images. In CVPR

    2021

  28. [36]

    A.; Shechtman, E.; and Wang, O

    Zhang, R.; Isola, P.; Efros, A. A.; Shechtman, E.; and Wang, O. 2018. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 586--595

    2018

Pith tools

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