Pith. sign in

REVIEW 4 major objections 6 minor 44 references

RelayGS: Reconstructing Dynamic Scenes with Large-Scale and Complex Motions via Relay Gaussians

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read RelayGS claims that breaking long object trajectories into shorter per-segment Gaussian nodes, trained against pseudo-views, is what lets 3D Gaussian Splatting reconstruct large-scale complex motions, and reports average PSNR gains of…

desk verdict A solid empirical 3DGS paper whose relay-Gaussian idea is genuinely useful, but the mask-first design is fragile at exactly the point the paper does not test. read the letter →

arxiv 2412.02493 v1 pith:GOLPT2QR submitted 2024-12-03 cs.CV

Qiankun Gao , Yanmin Wu , Chengxiang Wen , Jiarui Meng , Luyang Tang , Jie Chen , Ronggang Wang , Jian Zhang This is my paper · ORCID
classification cs.CV
keywords dynamicscenereconstruction3DGaussianSplatting4Drepresentationmotionfieldforeground-backgrounddecouplingtemporaldensificationmulti-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

RelayGS is a 3D Gaussian Splatting method for reconstructing dynamic scenes in which objects move far and in complex ways. The paper's central claim is that a long motion trajectory is easier to reconstruct once broken into shorter pieces, each represented by its own set of foreground Gaussians called Relay Gaussians. The method first learns a static scene from all frames and uses a learnable binary mask to separate fast-moving foreground Gaussians from a nearly static background. It then copies the foreground Gaussians once per temporal segment and trains the copies against pseudo-views formed by blending frames inside that segment, so the copies settle at different points along the true motion path. A final stage learns a compact motion field over these nodes, and the paper reports an average gain of 1.02 dB in PSNR over prior best methods on PanopticSports plus more complete reconstructions of real basketball game scenes.

What carries the argument

The central object is the Relay Gaussian, a per-segment copy of a foreground Gaussian that serves as an explicit node along the object's motion trajectory. Stage 1 learns a static 3DGS from all frames and uses a straight-through-estimator binary mask to decouple high-dynamic foreground and low-dynamic background; Stage 2 replicates foreground Gaussians once per $k$-frame segment and optimizes each copy against pseudo-views, which are uniform blends of three frames from that segment, an operation the paper analyzes as temporal densification; Stage 3 fits a HexPlane-plus-MLP motion field with separate foreground and background MLPs and a learnable scale factor on foreground position offsets. Together these parts turn one long trajectory into several short ones, which is what the paper claims makes large motion learnable.

What would settle it

A decisive test would be to render a synthetic multi-view scene with known ground-truth foreground masks and large motions, run RelayGS, and compare the learned Stage 1 mask with the ground truth; if forcing the mask to the ground-truth labels changes the PSNR gain, then the decoupling step, not the relay decomposition, is carrying the reported improvement.

Watch

Extended reading notes

Core claim

The authors' claim, on their own terms, is that explicit temporal decomposition of foreground motion is what allows a canonical-Gaussian-plus-motion-field representation to handle large-scale, complex dynamics. Their three-stage pipeline yields canonical Gaussians that already contain separated foreground and background, plus per-segment Relay Gaussians that approximate an ideal continuous trajectory by discrete, optimized nodes. With this foundation, the compact motion field only needs to predict short displacements between relay nodes rather than one giant leap, and the reported result is an average 1.02 dB PSNR improvement over the previous best on the six PanopticSports scenes, a 0.45 dB gain over the prior best on the VRU basketball scenes, and visibly more complete and temporally coherent players in rendered video.

Load-bearing premise

The load-bearing premise is that Stage 1's learnable binary mask can tell moving foreground Gaussians from static background Gaussians using only how well each Gaussian explains the first frame versus all other frames; if the mask mislabels a dynamic object as background, those Gaussians are never copied into Relay Gaussians and the later stages inherit the error.

Editorial extensions

If this is right

  • If the central claim holds, dynamic-scene reconstruction methods can address large motions by temporally densifying the trajectory instead of only enlarging the deformation network.
  • The Stage 1 mask result implies that foreground and background Gaussians can be separated without optical flow or semantic segmentation, which would let the pipeline run on datasets where such priors are unavailable.
  • The per-scene PSNR gains imply that most of the improvement is concentrated in the moving foreground, so metrics dominated by static background understate the method's effect.
  • The $k=16$ segment-length ablation implies that segment length is a real hyperparameter: too coarse a decomposition shortens the advantage, while too fine a decomposition increases training cost without further quality gains.

Reading between the lines

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

  • A direct extension the paper leaves implicit is adaptive temporal segmentation: instead of a fixed 16-frame window, segment boundaries could be placed where motion speed changes, which the $k$-ablation suggests would improve on the fixed choice.
  • Because the mask is trained without motion priors, the method should be tested on a synthetic scene with known foreground identities and foreground/background colors made nearly identical; if the separation then fails, the reported gain would be partly attributable to the datasets' mask learnability rather than the relay mechanism itself.
  • The authors' stated limitation about small fast objects such as the ball indicates that temporal densification alone does not solve spatial sparsity; combining relay copies with targeted spatial densification along the ball's path is a natural next experiment.
  • Pseudo-views are currently simple uniform blends of three frames; learned per-pixel blending or frame selection guided by motion magnitude could strengthen the supervisory signal, since the paper's own ablation shows pseudo-views contribute about 0.20 to 0.26 dB.
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

4 major / 6 minor

Summary. The paper proposes RelayGS, a three-stage 3D Gaussian Splatting method for reconstructing dynamic scenes with large and complex motions. In Stage 1, a static 3DGS is trained on all frames together with a learnable per-Gaussian binary mask that separates static background (mask=1) from dynamic foreground (mask=0); all Gaussians render the first frame, while only background Gaussians render later frames. In Stage 2, copies of the foreground Gaussians are created for each temporal segment and optimized against pseudo-views formed by pixel-wise averaging of three frames per segment; these copies are called Relay Gaussians. In Stage 3, the canonical Gaussians and a HexPlane-plus-MLP motion field are jointly optimized, with separate MLPs for foreground and background and a learnable per-Gaussian scaling factor for foreground position deformation. Experiments are reported on the PanopticSports and VRU Basketball Games datasets, with quantitative tables, ablations, and qualitative comparisons.

Significance. If the central claims hold, the idea of temporal densification through Relay Gaussians is a useful and novel contribution to dynamic 3DGS reconstruction. The method avoids pre-trained segmentation or optical-flow priors, and the paper includes ablations for the mask, relay copies, pseudo-views, foreground-background MLP separation, and the scaling factor. The evaluation on held-out test cameras and public datasets is a positive feature. However, the headline PSNR improvements rest on partially estimated baseline numbers, the mask decoupling is not quantitatively validated, and the pseudo-view supervision is theoretically questionable. These issues currently prevent the paper from fully supporting its state-of-the-art claim.

major comments (4)
  1. [Sec. 4.1, Eqs. (7)-(8)] The binary mask is the keystone of the pipeline: Stage 2 replicates only the mask=0 foreground Gaussians, so any error in mask assignment propagates to the Relay Gaussians. The mask is trained only through the first-frame render plus the static-objective gradients on other frames, and the paper provides no quantitative check of mask quality (e.g., IoU against a known foreground segmentation) and does not report the threshold epsilon. Moreover, because the point cloud is initialized from the first frame, any dynamic object absent or heavily occluded in frame 1 has no Gaussians to be replicated in Stage 2. Please add a mask-accuracy measurement, a sensitivity study around epsilon, and at least one experiment or explicit discussion of the first-frame dependence.
  2. [Sec. 4.2, Eq. (9)] The pseudo-view used as supervision for Relay Gaussians is a pixel-wise average of three frames at different instants. No real camera observes this image, and a static set of Gaussians cannot simultaneously reproduce three different object poses; the optimum of the pseudo-view loss may be a ghosted or averaged position rather than a point on the actual motion trajectory. The paper does not justify why this composite target is appropriate for decomposing large motions, and the ablation in Table 3 only removes pseudo-views entirely. Please provide a principled justification or an ablation comparing Eq. (9) with alternative supervision, such as separate per-frame supervision for each Relay Gaussian copy.
  3. [Table 2 and Sec. 5.2] The average 1.02 dB gain on PanopticSports is the paper's headline quantitative claim, but for two of the four competing methods (Dynamic3DGS and D-MiSo) the table reports numbers 'partially taken directly from their original papers or estimated based on the paper and available code.' It is not demonstrated that these numbers were produced under the same resolution, test cameras, frame range, and initialization as RelayGS. The comparison against those baselines is therefore not apples-to-apples and cannot fully support the state-of-the-art claim. Please re-run all baselines under the same protocol, or explicitly restrict the SOTA claim to the baselines that were re-run.
  4. [Sec. 5.2, Tables 1-2] No variance or multiple-seed results are reported, although several per-scene gains are small (0.27 dB on Juggle, 0.45 dB average on VRU) and the method involves random initialization, stochastic densification, and several hand-set hyperparameters. Without error bars or a paired significance test over the held-out test cameras, the reader cannot assess whether the improvements are robust. Please report at least mean plus/minus standard deviation over three runs for the main tables.
minor comments (6)
  1. [Sec. 4.2] The word 'strightforward' should be 'straightforward'.
  2. [Fig. 1 caption] The word 'Gaaussians' should be 'Gaussians'.
  3. [Sec. 5.2] The phrase 'reduces the rendering speed in some tend' appears to be a typo for 'in some extent'; please rephrase.
  4. [Sec. 4.3 and Sec. 5.1] The final training loss of RelayGS is never written explicitly; Eq. (3) and Eq. (6) are the losses of 3DGS and 4D-GS, and it is unclear whether the third stage uses L1+D-SSIM, L1+Ltv, or a combination. Please state the exact loss.
  5. [Table 2] D-MiSo has no storage or training time entries; please complete these entries or state why they are unavailable.
  6. [Sec. 5.2] The statement that ST-GS is not applicable to PanopticSports because per-frame point clouds cannot be obtained would benefit from a brief explanation, since ST-GS is otherwise a strong baseline.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the central PSNR claims are evaluated on held-out test cameras; the only self-citation (HiCoM) is descriptive related work, not load-bearing.

full rationale

The paper's load-bearing claims are tested against held-out cameras on public datasets, so the reported gains are not defined by any fitted parameter of the model. The Stage-1 learnable mask (Eqs. 7-8) is a self-supervised decoupling mechanism trained by reconstruction loss; whether it reliably separates foreground from background is a robustness concern, not a circularity. Stage-2 pseudo-views (Eq. 9) are training supervision for intermediate Relay Gaussians, and Stage-3 motion prediction is evaluated on novel test views, not on the pseudo-views used for fitting. The only self-citation is to HiCoM [8] in the related-work survey ('HiCoM [8] leverages the non-uniform distribution and local consistency to enable fast and accurate motion learning across frames'), which is descriptive and not used to justify any design choice or to forbid alternatives. The method openly builds on 4D-GS's HexPlane and MLPs rather than importing a contested uniqueness result. The limitations section acknowledges known failure modes (small fast-moving objects, simple temporal segmentation, stationary-camera assumption), which further indicates the evaluation is not constructed to be tautologically successful. Overall, no prediction reduces by definition to its inputs, and no load-bearing argument rests on a self-citation chain.

Assumptions & free parameters 6 free parameters · 6 assumptions · 1 invented entities

RelayGS rests on standard 3DGS rendering and the HexPlane motion field from prior work, adds several hand-chosen hyperparameters (k, p, beta, epsilon, step counts, thresholds) and one per-Gaussian learned scaling factor, and introduces Relay Gaussians as an internal modeling construct rather than a new physical entity. The main ad-hoc elements are the pseudo-view supervision and the scaling factor; neither has a theoretical guarantee.

free parameters (6)
  • k (frames per temporal segment) = 16
    Hand-chosen default; ablation in Table 4 on GZ shows PSNR peaks at k=16 (28.06 dB) and declines for larger segments.
  • p and beta (pseudo-view frame count and blending weights) = p=3, beta=(1/3,1/3,1/3)
    Uniformly selected frames t1, t2, t3 per segment and uniform blending in Eq. (9); no ablation on p or beta is reported.
  • epsilon (masking threshold) = not specified
    Threshold in the straight-through estimator (Eq. 7) that binarizes the learnable mask; value is not reported in the paper, yet it controls foreground/background separation in Stage 1.
  • gamma (learnable position deformation scale per Relay Gaussian) = learned, initialized to 0
    Introduced in Eq. (10) to let foreground displacements exceed MLP predictions; ablated in Table 3 (#6).
  • Stage step counts and learning-rate/densification settings = 3k/14k/20k steps; fg position lr 1e-3, bg 2e-4; fg gradient threshold 1e-4, fg scaling threshold 1e-3
    Training schedule chosen by hand and described in Sec. 5.1 and Sec. 10; differences between foreground and background thresholds are motivated by encouraging smaller, faster-splitting foreground Gaussians.
  • Per-camera channel-wise affine color tune parameters = learned
    Applied on PanopticSports to correct multi-view color inconsistencies, following Dynamic3DGS; details are not given.
assumptions (6)
  • domain assumption Gaussian Splatting's differentiable rasterization and loss (Eqs. 1-3) from Kerbl et al. are correct and transfer to the 4D setting.
    The whole pipeline renders and optimizes Gaussians via this external method; the paper does not re-derive it.
  • domain assumption Multi-view cameras are static, synchronized, and have known poses.
    Both datasets provide calibrated fixed cameras (Sec. 5.1, Sec. 9); the method does not handle moving or monocular cameras, as the limitation section states.
  • domain assumption A per-Gaussian binary mask, learned without motion priors, can reliably separate high-dynamic foreground from low-dynamic background.
    Stage 1's decoupling (Sec. 4.1, Eqs. 7-8) depends on mask gradients from frame-to-frame inconsistencies; failure of this split breaks Stages 2 and 3.
  • ad hoc to paper Averaging three uniformly selected frames yields a valid pseudo-view for optimizing Relay Gaussians.
    Eq. (9) assumes ghosted averages of moving content provide useful supervision; no analysis of artifacts is given.
  • domain assumption The HexPlane plus lightweight MLP motion field, inherited from 4D-GS, can represent residual motion once trajectories are decomposed.
    Stage 3 adopts this motion field (Sec. 4.3) and only modifies it with separate MLPs and a scaling factor.
  • ad hoc to paper The learnable scaling factor (1+e^gamma) is sufficient to let foreground Gaussians move farther without destabilizing joint optimization.
    Eq. (10) is introduced to fix the limited range of MLP-predicted displacements; its sufficiency is supported only by the Table 3 ablation.
invented entities (1)
  • Relay Gaussians
    purpose: Per-segment copies of foreground Gaussians that act as explicit nodes along each motion trajectory, decomposing large motions into smaller segments (Sec. 4.2).
    They are internal model constructs (duplicated and optimized Gaussians) with no observable signature outside the rendering pipeline; their learned positions are visualized in Figs. 4 and 12, but those are model inspections, not independent evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RelayGS: Reconstructing Dynamic Scenes with Large-Scale and Complex Motions via Relay Gaussians." pith.science (2026). https://pith.science/paper/GOLPT2QR

@misc{pith2026241202493,
  author       = {Pith},
  title        = {Pith review of: RelayGS: Reconstructing Dynamic Scenes with Large-Scale and Complex Motions via Relay Gaussians},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GOLPT2QR}},
  note         = {Machine review of arXiv:2412.02493}
}
read the original abstract

Reconstructing dynamic scenes with large-scale and complex motions remains a significant challenge. Recent techniques like Neural Radiance Fields and 3D Gaussian Splatting (3DGS) have shown promise but still struggle with scenes involving substantial movement. This paper proposes RelayGS, a novel method based on 3DGS, specifically designed to represent and reconstruct highly dynamic scenes. Our RelayGS learns a complete 4D representation with canonical 3D Gaussians and a compact motion field, consisting of three stages. First, we learn a fundamental 3DGS from all frames, ignoring temporal scene variations, and use a learnable mask to separate the highly dynamic foreground from the minimally moving background. Second, we replicate multiple copies of the decoupled foreground Gaussians from the first stage, each corresponding to a temporal segment, and optimize them using pseudo-views constructed from multiple frames within each segment. These Gaussians, termed Relay Gaussians, act as explicit relay nodes, simplifying and breaking down large-scale motion trajectories into smaller, manageable segments. Finally, we jointly learn the scene's temporal motion and refine the canonical Gaussians learned from the first two stages. We conduct thorough experiments on two dynamic scene datasets featuring large and complex motions, where our RelayGS outperforms state-of-the-arts by more than 1 dB in PSNR, and successfully reconstructs real-world basketball game scenes in a much more complete and coherent manner, whereas previous methods usually struggle to capture the complex motion of players. Code will be publicly available at https://github.com/gqk/RelayGS

Figures

Figures reproduced from arXiv: 2412.02493 by the authors.

Figure 1
Figure 1. Framework of the proposed RelayGS. (a) Initialize the scene with all images and separate the relatively static background and [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Qualitative comparisons on GZ scene of VRU Basketball Games dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Qualitative comparisons on Football scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: The visualization of canonical 3D Gaussians. (a) Reference image of the scene. (b) Initialization by 4D-GS, with the foreground [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Illustrative depiction of two types of densification. In 3DGS for static scene reconstruction, spatial densification is employed to better fit 3D structures. Prior 4D methods, as shown in (a), perform densification within a canonical 3D space, relying on deformation fi…
Figure 6
Figure 6. Figure 6: Qualitative comparisons on DG4 scene of VRU Basketball Games dataset. [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparisons on Juggle scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Qualitative comparisons on Boxes scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparisons on Softball scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]
Figure 10
Figure 10. Figure 10: Qualitative comparisons on Tennis scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparisons on Basketball scene of PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Visualizations of the second-stage dynamic foreground Relay Gaussians (red points) in 6 scenes of the PanopticSports dataset. [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 39 canonical work pages

  1. [1]

    Per-gaussian embedding-based deformation for deformable 3d gaussian splatting

    Jeongmin Bae, Seoha Kim, Youngsik Yun, Hahyun Lee, Gun Bang, and Youngjung Uh. Per-gaussian embedding-based deformation for deformable 3d gaussian splatting. In Pro- ceedings of the European Conference on Computer Vision (ECCV), 2024. 5, 6, 2

    work page 2024

  2. [2]

    Es- timating or propagating gradients through stochastic neurons

    Yoshua Bengio, Nicholas L´eonard, and Aaron Courville. Es- timating or propagating gradients through stochastic neurons. arXiv preprint arXiv:1308.3432, 2013. 3

    arXiv 2013

  3. [3]

    Hexplane: A fast representation for dynamic scenes

    Ang Cao and Justin Johnson. Hexplane: A fast representation for dynamic scenes. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR),

    work page

  4. [4]

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

    Yihang Chen, Qianyi Wu, Weiyao Lin, Mehrtash Harandi, and Jianfei Cai. Hac: Hash-grid assisted context for 3d gaus- sian splatting compression. In Proceedings of the European Conference on Computer Vision (ECCV), 2024. 3

    work page 2024

  5. [5]

    Superpoint gaussian splat- ting for real-time high-fidelity dynamic scene reconstruction

    Gang Zeng Diwen Wan, Ruijie Lu. Superpoint gaussian splat- ting for real-time high-fidelity dynamic scene reconstruction. In Proceedings of the International Conference on Machine Learning (ICML), 2024. 1

    work page 2024

  6. [6]

    4d-rotor gaussian splatting: Towards efficient novel view synthesis for dynamic scenes

    Yuanxing Duan, Fangyin Wei, Qiyu Dai, Yuhang He, Wen- zheng Chen, and Baoquan Chen. 4d-rotor gaussian splatting: Towards efficient novel view synthesis for dynamic scenes. In ACM SIGGRAPH 2024 Conference Papers, 2024. 2

    work page 2024

  7. [7]

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

    Sara Fridovich-Keil, Giacomo Meanti, Frederik Rahbæk War- burg, 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 (CVPR), 2023. 2

    work page 2023

  8. [8]

    Hicom: Hierarchical coherent motion for dy- namic streamable scenes with 3d gaussian splatting

    Qiankun Gao, Jiarui Meng, Chengxiang Wen, Jie Chen, and Jian Zhang. Hicom: Hierarchical coherent motion for dy- namic streamable scenes with 3d gaussian splatting. In Ad- vances in Neural Information Processing Systems (NeurIPS),

    work page

Show all 44 references
  1. [9]

    Motion-aware 3d gaussian splatting for efficient dynamic scene reconstruction

    Zhiyang Guo, Wengang Zhou, Li Li, Min Wang, and Houqiang Li. Motion-aware 3d gaussian splatting for efficient dynamic scene reconstruction. arXiv preprint arXiv:2403.11447, 2024. 1

    2024 arXiv

  2. [10]

    S4d: Streaming 4d real-world reconstruction with gaussians and 3d control points

    Bing He, Yunuo Chen, Guo Lu, Li Song, and Wenjun Zhang. S4d: Streaming 4d real-world reconstruction with gaussians and 3d control points. arXiv preprint arXiv:2408.13036, 2024. 2

    2024

  3. [11]

    Sc-gs: Sparse-controlled gaus- sian splatting for editable dynamic scenes

    Yi-Hua Huang, Yang-Tian Sun, Ziyi Yang, Xiaoyang Lyu, Yan-Pei Cao, and Xiaojuan Qi. Sc-gs: Sparse-controlled gaus- sian splatting for editable dynamic scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 1, 2

    2024

  4. [12]

    Panoptic studio: A massively multiview system for social motion capture

    Hanbyul Joo, Hao Liu, Lei Tan, Lin Gui, Bart Nabbe, Iain Matthews, Takeo Kanade, Shohei Nobuhara, and Yaser Sheikh. Panoptic studio: A massively multiview system for social motion capture. In Proceedings of the IEEE interna- tional conference on computer vision, 2015. 2, 5

    2015

  5. [13]

    A compact dynamic 3d gaussian representation for real-time dynamic view synthesis

    Kai Katsumata, Duc Minh V o, and Hideki Nakayama. A compact dynamic 3d gaussian representation for real-time dynamic view synthesis. In Proceedings of the European Conference on Computer Vision (ECCV), 2024. 2

    2024

  6. [14]

    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), 2023. 1, 2, 5

    2023

  7. [15]

    Dynmf: Neural motion factorization for real-time dynamic view syn- thesis with 3d gaussian splatting

    Agelos Kratimenos, Jiahui Lei, and Kostas Daniilidis. Dynmf: Neural motion factorization for real-time dynamic view syn- thesis with 3d gaussian splatting. In Proceedings of the Euro- pean Conference on Computer Vision (ECCV), 2024. 1

    2024

  8. [16]

    Compact 3d gaussian representation for radiance field

    Joo Chan Lee, Daniel Rho, Xiangyu Sun, Jong Hwan Ko, and Eunbyung Park. Compact 3d gaussian representation for radiance field. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 3

    2024

  9. [17]

    Self-calibrating 4d novel view synthesis from monocular videos using gaussian splatting

    Fang Li, Hao Zhang, and Narendra Ahuja. Self-calibrating 4d novel view synthesis from monocular videos using gaussian splatting. arXiv preprint arXiv:2406.01042, 2024. 2

    2024 arXiv

  10. [18]

    Neu- ral 3d video synthesis from multi-view video

    Tianye Li, Mira Slavcheva, Michael Zollhoefer, Simon Green, Christoph Lassner, Changil Kim, Tanner Schmidt, Steven Lovegrove, Michael Goesele, Richard Newcombe, et al. Neu- ral 3d video synthesis from multi-view video. In Proceedings of the IEEE/CVF Conference on Computer Visi...

    2022

  11. [19]

    Spacetime gaus- sian feature splatting for real-time dynamic view synthesis

    Zhan Li, Zhang Chen, Zhong Li, and Yi Xu. Spacetime gaus- sian feature splatting for real-time dynamic view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 5, 6, 2, 3

    2024

  12. [20]

    Gaufre: Gaussian deformation fields for real-time dynamic novel view synthesis

    Yiqing Liang, Numair Khan, Zhengqin Li, Thu Nguyen- Phuoc, Douglas Lanman, James Tompkin, and Lei Xiao. Gaufre: Gaussian deformation fields for real-time dynamic novel view synthesis. arXiv preprint arXiv:2312.11458, 2023. 2

    2023 arXiv

  13. [21]

    Gaussian- flow: 4d reconstruction with dynamic 3d gaussian particle

    Youtian Lin, Zuozhuo Dai, Siyu Zhu, and Yao Yao. Gaussian- flow: 4d reconstruction with dynamic 3d gaussian particle. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 1

    2024

  14. [22]

    Modgs: Dynamic gaussian splatting from causually-captured monocular videos

    Qingming Liu, Yuan Liu, Jiepeng Wang, Xianqiang Lv, Peng Wang, Wenping Wang, and Junhui Hou. Modgs: Dynamic gaussian splatting from causually-captured monocular videos. arXiv preprint arXiv:2406.00434, 2024

    2024 arXiv

  15. [23]

    3d geometry-aware deformable gaussian splatting for dynamic view synthesis

    Zhicheng Lu, Xiang Guo, Le Hui, Tianrui Chen, Min Yang, Xiao Tang, Feng Zhu, and Yuchao Dai. 3d geometry-aware deformable gaussian splatting for dynamic view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 1

    2024

  16. [24]

    Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis

    Jonathon Luiten, Georgios Kopanas, Bastian Leibe, and Deva Ramanan. Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis. In International Conference on 3D Vision (3DV), 2024. 2, 5, 6

    2024

  17. [25]

    Splatfields: Neural gaussian splats for sparse 3d and 4d reconstruction

    Marko Mihajlovic, Sergey Prokudin, Siyu Tang, Robert Maier, Federica Bogo, Tony Tung, and Edmond Boyer. Splatfields: Neural gaussian splats for sparse 3d and 4d reconstruction. In Proceedings of the European Conference on Computer Vision (ECCV), 2024. 1

    2024

  18. [26]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view synthe- sis. In Proceedings of the European Conference on Computer Vision (ECCV), 2020. 1 9

    2020

  19. [27]

    Instant neural graphics primitives with a multires- olution hash encoding

    Thomas M¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a multires- olution hash encoding. ACM transactions on graphics (TOG),

  20. [28]

    Nerfies: Deformable neural radiance fields

    Keunhong Park, Utkarsh Sinha, Jonathan T Barron, Sofien Bouaziz, Dan B Goldman, Steven M Seitz, and Ricardo Martin-Brualla. Nerfies: Deformable neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021. 1, 2

    2021

  21. [29]

    Barron, Sofien Bouaziz, Dan B Goldman, Ricardo Martin- Brualla, and Steven M

    Keunhong Park, Utkarsh Sinha, Peter Hedman, Jonathan T. Barron, Sofien Bouaziz, Dan B Goldman, Ricardo Martin- Brualla, and Steven M. Seitz. Hypernerf: A higher- dimensional representation for topologically varying neural radiance fields. ACM Transactions on Graphics (TOG), 2021

    2021

  22. [30]

    D-nerf: Neural radiance fields for dynamic scenes

    Albert Pumarola, Enric Corona, Gerard Pons-Moll, and Francesc Moreno-Noguer. D-nerf: Neural radiance fields for dynamic scenes. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR),

  23. [31]

    Structure- from-motion revisited

    Johannes L Schonberger and Jan-Michael Frahm. Structure- from-motion revisited. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR),

  24. [32]

    Tensor4d: Efficient neural 4d decomposition for high-fidelity dynamic reconstruction and rendering

    Ruizhi Shao, Zerong Zheng, Hanzhang Tu, Boning Liu, Hong- wen Zhang, and Yebin Liu. Tensor4d: Efficient neural 4d decomposition for high-fidelity dynamic reconstruction and rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  25. [33]

    3dgstream: On-the-fly training of 3d gaus- sians for efficient streaming of photo-realistic free-viewpoint videos

    Jiakai Sun, Han Jiao, Guangyuan Li, Zhanjie Zhang, Lei Zhao, and Wei Xing. 3dgstream: On-the-fly training of 3d gaus- sians for efficient streaming of photo-realistic free-viewpoint videos. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition ...

    2024

  26. [34]

    Vru-sequence, 2024

    VRU. Vru-sequence, 2024. https://anonymous. 4open.science/r/VRU-Sequence/. 2, 5

    2024

  27. [35]

    D-miso: Editing dynamic 3d scenes using multi-gaussians soup

    Joanna Waczy´nska, Piotr Borycki, Joanna Kaleta, Sławomir Tadeja, and Przemysław Spurek. D-miso: Editing dynamic 3d scenes using multi-gaussians soup. In Proceedings of the Ad- vances in Neural Information Processing Systems (NeurIPS),

  28. [36]

    4d gaussian splatting for real-time dynamic scene rendering

    Guanjun Wu, Taoran Yi, Jiemin Fang, Lingxi Xie, Xiaopeng Zhang, Wei Wei, Wenyu Liu, Qi Tian, and Xinggang Wang. 4d gaussian splatting for real-time dynamic scene rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 2, 3, 5, 6, 1

    2024

  29. [37]

    Deformable 3d gaussians for high- fidelity monocular dynamic scene reconstruction

    Ziyi Yang, Xinyu Gao, Wen Zhou, Shaohui Jiao, Yuqing Zhang, and Xiaogang Jin. Deformable 3d gaussians for high- fidelity monocular dynamic scene reconstruction. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 1, 2

    2024

  30. [38]

    Real-time photorealistic dynamic scene representation and rendering with 4d gaussian splatting

    Zeyu Yang, Hongye Yang, Zijie Pan, and Li Zhang. Real-time photorealistic dynamic scene representation and rendering with 4d gaussian splatting. In International Conference on Learning Representations (ICLR), 2024. 1, 2

    2024

  31. [39]

    Egogaussian: Dynamic scene understanding from egocentric video with 3d gaussian splatting

    Daiwei Zhang, Gengyan Li, Jiajie Li, Micka ¨el Bressieux, Otmar Hilliges, Marc Pollefeys, Luc Van Gool, and Xi Wang. Egogaussian: Dynamic scene understanding from egocentric video with 3d gaussian splatting. In International Conference on 3D Vision (3DV), 2025. 2 10 RelayGS: R...

    2025

  32. [40]

    In standard 3DGS, regions with in- sufficient spatial representation are typically addressed by add Gaussians in those areas, an operation we refer to as spatial densification

    Relay Gaussians from Densification Perspec- tive Spatial Densification. In standard 3DGS, regions with in- sufficient spatial representation are typically addressed by add Gaussians in those areas, an operation we refer to as spatial densification. Most prior 4D reconstruction...

  33. [41]

    Unified vs. Segment-Based Reconstruction Despite the explicit use of temporal segments in the sec- ond stage of our method for learning Relay Gaussians, our approach is fundamentally different from segment-based re- construction methods such as Deformable3DGS [ 37] and ST-GS [...

  34. [42]

    The cameras are temporally aligned with accurate intrinsic and extrinsic parameters

    Dataset Details PanopticSports Dataset. The cameras are temporally aligned with accurate intrinsic and extrinsic parameters. Po- sitioned in a roughly hemispherical arrangement around the area of interest in the middle of the capture studio, the cam- eras provide comprehensive...

    1920

  35. [43]

    The background learning rates are similar to those used in previous methods, with the initial learning rate for position set to 2e-4 and the minimum learn- ing rate to 1e-5

    More Implementation Details Our method employs slightly different settings for learning rates and densification thresholds between the foreground and background Gaussians. The background learning rates are similar to those used in previous methods, with the initial learning ra...

  36. [44]

    Using low-resolution views during these stages produces comparable results while significantly reducing training time

    Additional Experimental Results The goal of the first two stages of our method is to learn a more robust base Gaussian representation, simplifying complex motion patterns in the scene and preparing for full learning in the final stage. Using low-resolution views during these s...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.