Pith. sign in

REVIEW 3 major objections 5 minor 53 references

FreeTimeGS: Free Gaussian Primitives at Anytime and Anywhere for Dynamic Scene Reconstruction

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read FreeTimeGS claims that freeing Gaussian primitives from a canonical time and giving each a straight-line motion and a single-peaked temporal window makes fast, complex dynamic scenes easier to reconstruct, and reports the best rendering…

desk verdict FreeTimeGS is a genuinely different 4D Gaussian representation with strong results on fast-motion scenes, but the 'anytime' claim is not actually tested at unobserved times and the paper would benefit from code, data, and variance reporting. read the letter →

arxiv 2506.05348 v2 pith:OT4A7RPK submitted 2025-06-05 cs.CV

Yifan Wang , Peishan Yang , Zhen Xu , Jiaming Sun , Zhanhua Zhang , Yong Chen , Hujun Bao , Sida Peng
show 1 more author
Xiaowei Zhou
This is my paper
classification cs.CV
keywords dynamicscenereconstruction4DGaussianprimitivesnovelviewsynthesisspace-timerepresentationtemporalopacityregularizationreal-timerenderingcomplexmotion
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

FreeTimeGS is a 4D scene representation for reconstructing dynamic scenes from multi-view video. Instead of anchoring Gaussian primitives in a canonical space and deforming them to each frame, it lets every Gaussian primitive be born at any position and time, move with a simple constant-velocity straight-line motion, and fade in and out through a single-peaked temporal opacity window. The paper argues this removes the need for long-range canonical-to-observation correspondences, which is the step that deformation-based methods struggle with on fast and complex motion, and that the extra flexibility pays off in quality and speed. On its own collected SelfCap dataset of dancing, pets, and bike repairs, the method reports PSNR gains of 2.4 dB and 1.4 dB over 4DGS and STGS, and 4.1 dB and 2.6 dB in dynamic regions, while rendering at 467 FPS. The practical stake is that live free-viewpoint replay of fast scenes, for film, games, or VR, becomes possible without a long offline deformation optimization.

What carries the argument

The machine is the free Gaussian primitive: a 3D Gaussian that exists only inside a temporal window. Equation (1), $\mu_x(t)=\mu_x+v\cdot(t-\mu_t)$, is a constant-velocity motion function that moves the primitive to neighboring regions; Equation (4), $\sigma(t)=\exp(-\frac{1}{2}((t-\mu_t)/s)^2)$, is a single-peaked temporal opacity that lets the primitive's birth time and duration be optimized by gradient descent and controls how much it contributes at each time. Together they turn dynamic reconstruction into a search over short-lived primitives instead of a search over long-range deformations. The argument is carried by the combination of this representation with a regularization loss $L_{\rm reg}=\frac{1}{N}\sum_i \sigma_i\cdot\mathrm{sg}[\sigma_i(t)]$ that suppresses high opacity early in training, a periodic relocation of underused primitives to high-score areas, and a four-dimensional initialization that seeds positions, times, and velocities from tracked 3D points.

What would settle it

Reconstruct a turntable scene with a rapidly rotating textured object using FreeTimeGS and a deformation-based baseline under identical cameras. If, as rotation speed increases, FreeTimeGS's PSNR advantage shrinks or reverses and its primitive count grows much faster than the baseline's, then the constant-velocity straight-line motion assumption is the limiting factor, and the reported gains on fast motion would not generalize to rotation-dominated motion.

Watch

Extended reading notes

Core claim

The central claim is that a 4D Gaussian representation in which primitives are not tied to a canonical time is enough, by itself, to model dynamic scenes with strong motion, and that the remaining optimization difficulties are solved by three accessories. Each primitive carries position $\mu_x$, birth time $\mu_t$, duration $s$, velocity $v$, scale, orientation, opacity, and color; at query time it is moved to $\mu_x + v(t-\mu_t)$ and its opacity is multiplied by $\exp(-\frac{1}{2}((t-\mu_t)/s)^2)$. Because every primitive only needs to explain a short temporal neighborhood, the paper replaces deformation fields with a linear motion function, and claims this makes optimization better-posed. On top of this, a time-weighted opacity regularization term penalizes primitives whose opacity is near 1 during early training, preventing them from blocking gradients; a periodic relocation step moves low-opacity primitives toward regions with high sampling score; and positions, times, and velocities are initialized from ROMA-based multi-view matches and nearest-neighbor correspondences. The paper reports best results on Neural3DV, ENeRF-Outdoor, and SelfCap, including 33.19 PSNR on Neural3DV and 27.41/29.38 PSNR on SelfCap for entire/dynamic regions at 467 FPS.

Load-bearing premise

The load-bearing premise is that fast and complex motion can be represented by a large enough set of short-lived Gaussian blobs, each moving in a straight line at constant speed and appearing and disappearing through one smooth rise-and-fall window.

Editorial extensions

If this is right

  • If the central claim is right, scenes with fast and complex motion no longer require learning long-range deformation correspondences; reconstruction reduces to optimizing short-lived primitives, and the failure modes of canonical-space deformation on large motions are bypassed.
  • The reported 467 FPS at 1080p on a single RTX 4090 means photorealistic free-viewpoint replay of dynamic scenes is real-time, so interactive applications such as VR, live sports replay, and virtual production can use the output directly.
  • Storage can be traded against quality smoothly: capping primitives at 500k or 70k lowers PSNR by only about 0.2–0.8 dB on Neural3DV while cutting model size from 125 MB to 41 MB or 8.3 MB, pointing to a compact streaming format.
  • Removing any of the four proposed components, the motion function, the 4D regularization, periodic relocation, or velocity initialization, degrades fast-motion PSNR by up to 3.8 dB in the fastest frames, so the recipe is load-bearing and not just a minor tweak.
  • On the SelfCap fast-motion dataset the gain is largest in dynamic regions, 4.1 dB and 2.6 dB over 4DGS and STGS, indicating the representation helps precisely where prior 4D Gaussian methods overfit or stall.

Reading between the lines

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

  • An implication the paper does not test: curved motion such as rotation or acceleration must be approximated by many short straight-line segments, so the required primitive count should grow with angular speed and curvature; a scene with pure rotation would reveal whether the reported 96 MB model size holds.
  • A testable extension the paper leaves open: replacing ROMA-based multi-view initialization with monocular or sparse correspondences would separate how much of the gain comes from the representation itself and how much from a good starting point.
  • The opacity-blocking-gradient mechanism is stated for dynamic scenes, but it is a general optimization claim: if correct, the same early-training opacity penalty should stabilize static 3D Gaussian Splatting as well.
  • The paper notes its representation is orthogonal to long-video hierarchies; a natural next step is coupling FreeTimeGS with a temporal hierarchy to stream arbitrarily long sequences, which the paper does not attempt.
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

3 major / 5 minor

Summary. The paper proposes FreeTimeGS, a dynamic scene representation in which each 3D Gaussian primitive is assigned a birth time, duration, constant-velocity motion, and a unimodal temporal opacity function. The method avoids canonical-space deformation fields, instead allowing primitives to appear at arbitrary positions and times, and it introduces an opacity regularization loss, periodic relocation of low-opacity primitives, and ROMA-based 4D initialization. Experiments on Neural3DV, ENeRF-Outdoor, and a self-collected SelfCap dataset report state-of-the-art PSNR/DSSIM/LPIPS values and real-time rendering at 450+ FPS.

Significance. If the claims hold, FreeTimeGS provides a simple and efficient alternative to deformation-based dynamic Gaussian methods, with particularly strong results on fast and complex motions. The supplementary per-scene breakdowns and the primitive-count ablations are concrete and useful evidence for the representation's flexibility and rendering-speed trade-off. The opacity regularization and periodic relocation are clearly motivated by the reported local-minimum problem. However, the central 'anytime' component of the claim is not validated because all evaluations are performed at training timestamps, the largest improvements are reported on a self-collected dataset, and one public benchmark table uses different scene subsets for different metrics. The representation is plausible, but the temporal-generalization claim needs explicit experimental support before the headline conclusions can be accepted.

major comments (3)
  1. [§4, Eq. (1), Eq. (4)] The central 'anytime' claim is never tested at timestamps that were not used in training. All experiments hold out camera views only; every test frame is also a training frame. Because each primitive has free position, time, duration, and velocity, and because temporal opacity is modeled as a unimodal Gaussian, the optimizer can place short-lived primitives near observed timestamps, so the reported large margins (e.g., 2.4 dB and 1.4 dB on SelfCap) may reflect per-frame fitting rather than generalization across time. I request an experiment that holds out entire timestamps, such as training on a subset of frames and evaluating on interleaved or withheld frames, with PSNR/DSSIM/LPIPS reported on novel times. In addition, to justify the constant-velocity local-linearity assumption, please report how the required primitive count or reconstruction error grows with motion complexity (e.g., acceleration or rotation), since Eq. (1) can otherwise only compensate by using more short-duration primitives.
  2. [Table 1] The Neural3DV comparison uses different scene subsets for different metrics: DSSIM1 is reported only on Flame Salmon, while DSSIM2 excludes Coffee Martini. As a result, the claim of outperforming all baselines in all metrics is not supported by a consistent evaluation. For example, DSSIM1 for Ours equals STGS at 0.026, and the DSSIM2 advantage is computed on a subset that excludes a scene where several baselines perform poorly. Please report all metrics on the same set of scenes for all methods, or state explicitly that no consistent full-scene comparison is available. If practical, add multiple seeds and error bars to the main tables.
  3. [Table 3, §4.1] The SelfCap dataset is the primary evidence for the large-margin improvement on fast motion, but it is self-collected and not released, and the evaluation protocol for the two strongest baselines includes traversing camera near-plane settings during testing to maximize floater removal. This protocol may advantage a method with fewer floaters and makes the comparison less standardized. Please report results for 4DGS and STGS under their default near-plane settings as well, describe exactly how the dynamic-region masks are applied when computing PSNR on the cropped masked images, and provide per-run variance or at least clarify how many random seeds were used.
minor comments (5)
  1. [§1, §3.1] The text contains grammatical and typographical errors: 'an motion' in the abstract, 'Experiments results' in §1, and a duplicated sentence in §3.1 that begins 'In addition, our approach assigns a motion function...' twice. The Figure 2 caption also contains 'to to model'.
  2. [Table 4, Table 5] The caption 'entire/fastest).' has an unmatched parenthesis and should be rewritten as 'entire/fastest' or 'entire and fastest subsequence'.
  3. [Eq. (6)] The notation in Eq. (6) uses both sigma and sigma(t) without a clear distinction between the base opacity parameter and the temporal opacity function; please define them consistently and state explicitly that the stop-gradient prevents the regularization from shrinking the temporal opacity.
  4. [§3.1] The statement that each Gaussian primitive consists of 'eight learnable parameters' is imprecise because scale, orientation, and spherical harmonics coefficients are multi-dimensional; rephrase as 'eight groups of learnable parameters'.
  5. [§5] The limitations paragraph acknowledges training time and relighting but does not mention the lack of validation on novel timestamps; given the paper's 'anytime' claim, this should be explicitly discussed as a limitation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper reports empirical benchmark comparisons of an optimized representation, and no fitted quantity is relabeled as a prediction.

full rationale

FreeTimeGS makes no first-principles prediction; its central claim is that a 4D Gaussian representation with per-primitive velocity, temporal opacity, opacity regularization, periodic relocation, and ROMA-based initialization outperforms prior methods on held-out camera views. Equations (1)-(6) define the representation and training losses, and the reported PSNR/DSSIM/LPIPS numbers are computed against external ground-truth frames under a standard benchmark protocol. No parameter fitted to a subset of data is renamed as a prediction: the per-primitive parameters (position, time, duration, velocity, opacity) are optimized end-to-end against the same training views used by all baselines, and the opacity regularizer L_reg uses a stop-gradient so it cannot be trivially minimized by its own weight. The self-citations (ENeRF, 4K4D, EasyVolcap, LongVolCap, Neural Body) are used as baselines, dataset sources, codebase acknowledgments, or related work; none is invoked to justify the central capability of FreeTimeGS itself. The 'anytime' claim is only evaluated at training timestamps, and the constant-velocity/unimodal-opacity assumptions are not stress-tested at unseen times, but unsupported generalization is a correctness risk, not a circularity. I find no step in which a claimed result reduces by construction to its own inputs.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim depends on several hand-set hyperparameters (regularization weight, relocation interval, sampling weights) and on domain assumptions about linear motion, unimodal temporal opacity, and the opacity-local-minima mechanism. The representation itself introduces no new physical entities; Gaussian primitives are an existing construct from 3DGS.

free parameters (5)
  • lambda_reg (opacity regularization weight) = 1e-2
    Chosen via ablation on the dance1 self-captured sequence; 1e-1 degrades PSNR from 29.74 to 26.43, showing sensitivity.
  • lambda_img, lambda_ssim, lambda_perc = 0.8, 0.2, 0.01
    Hand-set loss weights inherited from 3DGS-style training; no ablation shown.
  • lambda_g, lambda_o (sampling score weights) = 0.5, 0.5
    Hand-set for periodic relocation score in Eq. 7.
  • relocation interval N = 100 iterations
    Hand-set; no ablation on interval length.
  • velocity annealing schedule = linear from lambda_0 to lambda_1 over training
    Chosen by hand to model fast motion early and fine motion later; exact endpoints not specified in the paper.
assumptions (4)
  • ad hoc to paper Temporal opacity is modeled as a Gaussian (unimodal) function of time (Eq. 4).
    The paper asserts a unimodal function with a scaling parameter is needed but provides no derivation; it is a design choice that limits each primitive to a single appearance window.
  • domain assumption Motion is linear: position evolves as mu_x + v * (t - mu_t) (Eq. 1).
    Assumes locally constant velocity is sufficient to model complex motions because primitives can be placed freely in time.
  • ad hoc to paper High opacity of some primitives blocks gradient backpropagation and causes local minima.
    Stated as an observed motivation for the regularization loss (Sec. 3.2), not proven analytically; the optimal penalty weight is narrowly tuned.
  • domain assumption ROMA feature matches plus triangulation provide reliable 3D initialization.
    The 4D initialization uses ROMA matches and KNN velocities; if these fail (textureless or very fast regions), the method loses about 1.4 dB PSNR (Table 4).

how reviews work

0 comments
Cite this review

Pith. "Pith review of FreeTimeGS: Free Gaussian Primitives at Anytime and Anywhere for Dynamic Scene Reconstruction." pith.science (2026). https://pith.science/paper/OT4A7RPK

@misc{pith2026250605348,
  author       = {Pith},
  title        = {Pith review of: FreeTimeGS: Free Gaussian Primitives at Anytime and Anywhere for Dynamic Scene Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OT4A7RPK}},
  note         = {Machine review of arXiv:2506.05348}
}
read the original abstract

This paper addresses the challenge of reconstructing dynamic 3D scenes with complex motions. Some recent works define 3D Gaussian primitives in the canonical space and use deformation fields to map canonical primitives to observation spaces, achieving real-time dynamic view synthesis. However, these methods often struggle to handle scenes with complex motions due to the difficulty of optimizing deformation fields. To overcome this problem, we propose FreeTimeGS, a novel 4D representation that allows Gaussian primitives to appear at arbitrary time and locations. In contrast to canonical Gaussian primitives, our representation possesses the strong flexibility, thus improving the ability to model dynamic 3D scenes. In addition, we endow each Gaussian primitive with an motion function, allowing it to move to neighboring regions over time, which reduces the temporal redundancy. Experiments results on several datasets show that the rendering quality of our method outperforms recent methods by a large margin. Project page: https://zju3dv.github.io/freetimegs/ .

Figures

Figures reproduced from arXiv: 2506.05348 by the authors.

Figure 1
Figure 1. Photorealistic and real-time rendering of dynamic 3D scenes. Our method achieves the best performance on challenging dynamic scenes with fast and complex motion. Compared with current state-of-the-art methods 4DGS [49] and STGS [21], our PSNR is improved by 2.4dB and 1.4dB on the SelfCap dataset. For dynamic regions, our PSNR is improved by 4.1dB and 2.6dB. What’s more, our method supports real-time rendering at 108… view at source ↗
Figure 2
Figure 2. Pipeline Overview. We represent a dynamic scene using Gaussian primitives that can appear anytime anywhere. Each Gaussian is assigned with a motion function to to model its movement. And its opacity is modulated by the temporal opacity function which control the impact of the Gaussian primitive over time. With this 4D representation, we further regularize the Gaussians with a 4D regularization loss and optimize the … view at source ↗
Figure 3
Figure 3. Qualitative comparison on the ENeRF-Outdoor Dataset. Our method achieves higher quality for fast-moving objects and regions, such as the swinging arms and dolls in hands, and clearer text details on the clothes. Ground Truth Ours 4DGS STGS Ground Truth Ours 4DGS STGS [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison on the Neural 3D Video Dataset. Our method achieves the best rendering quality compared with baseline methods, especially for distant static regions and fast-moving dynamic regions. 6 [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison on our SelfCap Dataset. Our method achieves significantly higher rendering quality than other methods. For example, in the dance sequence, other methods struggle to handle fast-moving regions, such as fingers, faces, and texture details on clothe…
Figure 6
Figure 6. Figure 6: Ablation study of proposed components on dance1 sequence of our SelfCap Dataset. Removing our proposed components leads to visible artifacts in the rendered results, especially in the dynamic regions with fast motion. Our method produce high-quality results even in the…
Figure 7
Figure 7. Figure 7: Dynamic Region Mask. An illustration of the use of dy￾namic region masking during evaluation. (a) Ground Truth image from the SelfCap dataset. (b) Masked Ground Truth, generated by cropping the Ground Truth image using a dynamic region mask ex￾tracted by Background Mat…
Figure 8
Figure 8. Figure 8: Qualitative comparison on our SelfCap Dataset. 5 [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparison on our SelfCap Dataset. 6 [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Qualitative comparison on the Neural 3D Video Dataset. 1, 3, 5, 2, 4 [22] 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, abs/…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

53 extracted references · 41 canonical work pages

  1. [1]

    Green and yellow cell colors indicate the best and the second best results, respectively

    Benjamin Attal, Jia-Bin Huang, Christian Richardt, Michael Zollhoefer, Johannes Kopf, Matthew O’Toole, and Changil 1 Table 9.Quantitative comparison of view synthesis results on ENeRF-Outdoor [23] Dataset.We report PSNR, SSIM 2, and LPIPSV GG to evaluate the rendering quality. Green and yellow cell colors indicate the best and the second best results, res...

    work page 2023

  2. [2]

    Per-gaussian embedding-based deformation for deformable 3d gaussian splatting.ArXiv, abs/2404.03613, 2024

    Jeongmin Bae, Seoha Kim, Youngsik Yun, Hahyun Lee, Gun Bang, and Youngjung Uh. Per-gaussian embedding-based deformation for deformable 3d gaussian splatting.ArXiv, abs/2404.03613, 2024. 1, 3

    arXiv 2024

  3. [3]

    Deepdeform: Learning non-rigid rgb-d reconstruction with semi-supervised data

    Alja ˇz Bo ˇziˇc, Michael Zollh ¨ofer, Christian Theobalt, and Matthias Nießner. Deepdeform: Learning non-rigid rgb-d reconstruction with semi-supervised data. 2020. 2

    work page 2020

  4. [4]

    Hexplane: A fast representa- tion for dynamic scenes.arXiv, 2023

    Ang Cao and Justin Johnson. Hexplane: A fast representa- tion for dynamic scenes.arXiv, 2023. 2, 5, 1, 4

    work page 2023

  5. [5]

    4d video textures for interactive character appear- ance

    Dan Casas, Marco V olino, John Collomosse, and Adrian Hilton. 4d video textures for interactive character appear- ance. InComputer Graphics Forum, pages 371–380. Wiley Online Library, 2014. 1

    work page 2014

  6. [6]

    Generalizable neural performer: Learning robust radiance fields for human novel view synthesis.arXiv preprint arXiv:2204.11798, 2022

    Wei Cheng, Su Xu, Jingtan Piao, Chen Qian, Wayne Wu, Kwan-Yee Lin, and Hongsheng Li. Generalizable neural performer: Learning robust radiance fields for human novel view synthesis.arXiv preprint arXiv:2204.11798, 2022. 2

    arXiv 2022

  7. [7]

    High-quality streamable free-viewpoint video.ACM Transactions on Graphics (ToG), 34(4):1–13,

    Alvaro Collet, Ming Chuang, Pat Sweeney, Don Gillett, Den- nis Evseev, David Calabrese, Hugues Hoppe, Adam Kirk, and Steve Sullivan. High-quality streamable free-viewpoint video.ACM Transactions on Graphics (ToG), 34(4):1–13,

  8. [8]

    Fusion4d: Real-time performance capture of challeng- ing scenes.ACM TOG, 2016

    Mingsong Dou, Sameh Khamis, Yury Degtyarev, Philip Davidson, Sean Ryan Fanello, Adarsh Kowdle, Sergio Orts Escolano, Christoph Rhemann, David Kim, Jonathan Taylor, et al. Fusion4d: Real-time performance capture of challeng- ing scenes.ACM TOG, 2016. 1, 2

    work page 2016

Show all 53 references
  1. [9]

    RoMa: Robust Dense Feature Matching.IEEE Conference on Computer Vision and Pattern Recognition, 2024

    Johan Edstedt, Qiyu Sun, Georg B ¨okman, M ˚arten Wadenb¨ack, and Michael Felsberg. RoMa: Robust Dense Feature Matching.IEEE Conference on Computer Vision and Pattern Recognition, 2024. 4

    2024

  2. [10]

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

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

    2023

  3. [11]

    Motion-aware 3d gaussian splatting for effi- cient dynamic scene reconstruction.ArXiv, abs/2403.11447,

    Zhiyang Guo, Wen gang Zhou, Li Li, Min Wang, and Houqiang Li. Motion-aware 3d gaussian splatting for effi- cient dynamic scene reconstruction.ArXiv, abs/2403.11447,

    arXiv

  4. [12]

    Going beyond free viewpoint: creating animatable volumetric video of human performances.IET Computer Vi- sion, pages 350–358, 2020

    Anna Hilsmann, Philipp Fechteler, Wieland Morgenstern, Wolfgang Paier, Ingo Feldmann, Oliver Schreer, and Peter Eisert. Going beyond free viewpoint: creating animatable volumetric video of human performances.IET Computer Vi- sion, pages 350–358, 2020. 1

    2020

  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. InEuropean Conference on Com- puter Vision, 2023. 5, 1

    2023

  6. [14]

    3d gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics (TOG), 42(4):1–14, 2023

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics (TOG), 42(4):1–14, 2023. 3, 4, 8, 1

    2023

  7. [15]

    Neural human performer: Learning generalizable ra- diance fields for human performance rendering.Advances in Neural Information Processing Systems, 34, 2021

    Youngjoong Kwon, Dahun Kim, Duygu Ceylan, and Henry Fuchs. Neural human performer: Learning generalizable ra- diance fields for human performance rendering.Advances in Neural Information Processing Systems, 34, 2021. 2

    2021

  8. [16]

    Dgd: Dynamic 3d gaussians distillation.ArXiv, abs/2405.19321, 2024

    Isaac Labe, Noam Issachar, Itai Lang, and Sagie Be- naim. Dgd: Dynamic 3d gaussians distillation.ArXiv, abs/2405.19321, 2024. 1, 3

    2024 arXiv

  9. [17]

    Fully explicit dynamic gaussian splat- ting

    Junoh Lee, Chang-Yeon Won, Hyunjun Jung, Inhwan Bae, and Hae-Gon Jeon. Fully explicit dynamic gaussian splat- ting. 2024. 5, 1, 4

    2024

  10. [18]

    Neu- ral 3d video synthesis.CVPR, 2022

    Tianye Li, Mira Slavcheva, Michael Zollhoefer, Simon Green, Christoph Lassner, Changil Kim, Tanner Schmidt, Steven Lovegrove, Michael Goesele, and Zhaoyang Lv. Neu- ral 3d video synthesis.CVPR, 2022. 2 2 Table 10.Quantitative comparison of view synthesis results onSelfCapDatas...

    2022

  11. [19]

    Neural 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. Neural 3d video synthesis from multi-view video. In Proceedings of the IEEE/CVF Conference on Computer Vi- si...

    2022

  12. [20]

    Neural scene flow fields for space-time view synthesis of dy- namic scenes

    Zhengqi Li, Simon Niklaus, Noah Snavely, and Oliver Wang. Neural scene flow fields for space-time view synthesis of dy- namic scenes. InCVPR, 2021. 2

    2021

  13. [21]

    Space- time gaussian feature splatting for real-time dynamic view synthesis.2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 8508–8520, 2023

    Zhan Li, Zhang Chen, Zhong Li, and Yinghao Xu. Space- time gaussian feature splatting for real-time dynamic view synthesis.2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 8508–8520, 2023. 3 Table 11.Quantitative comparison of view synthesis re...

    2024

  14. [22]

    Gaufre: Gaussian deformation fields for real-time dynamic novel view synthesis.ArXiv, abs/2312.11458, 2023

    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, abs/2312.11458, 2023. 1, 3

    2023 arXiv

  15. [23]

    Efficient neural radiance fields for interactive free-viewpoint video

    Haotong Lin, Sida Peng, Zhen Xu, Yunzhi Yan, Qing Shuai, Hujun Bao, and Xiaowei Zhou. Efficient neural radiance fields for interactive free-viewpoint video. InSIGGRAPH Asia Conference Proceedings, 2022. 2, 5

    2022

  16. [24]

    Seitz, and Ira Kemelmacher- Shlizerman

    Shanchuan Lin, Andrey Ryabtsev, Soumyadip Sengupta, Brian Curless, Steven M. Seitz, and Ira Kemelmacher- Shlizerman. Real-time high-resolution background matting. 2021 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 8758–8767, 2020. 1, 2

    2021

  17. [25]

    Modgs: Dy- namic gaussian splatting from casually-captured monocular videos

    Qingming Liu, Yuan Liu, Jie-Chao Wang, Xianqiang Lyv, Peng Wang, Wenping Wang, and Junhui Hou. Modgs: Dy- namic gaussian splatting from casually-captured monocular videos. 2024. 1, 3

    2024

  18. [26]

    Neural vol- umes: Learning dynamic renderable volumes from images

    Stephen Lombardi, Tomas Simon, Jason Saragih, Gabriel Schwartz, Andreas Lehrmann, and Yaser Sheikh. Neural vol- umes: Learning dynamic renderable volumes from images. ACM Trans. Graph., 38(4):65:1–65:14, 2019. 5, 1, 4

    2019

  19. [27]

    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. 2024 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 8900–8910, 2024. 1, 3

    2024

  20. [28]

    Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis.arXiv preprint arXiv:2308.09713, 2023

    Jonathon Luiten, Georgios Kopanas, Bastian Leibe, and Deva Ramanan. Dynamic 3d gaussians: Tracking by persistent dynamic view synthesis.arXiv preprint arXiv:2308.09713, 2023. 3

    2023 arXiv

  21. [29]

    Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar

    Ben Mildenhall, Pratul P. Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar. Local light field fusion: Practical view syn- thesis with prescriptive sampling guidelines.ACM Transac- tions on Graphics (TOG), 2019. 5, 1, 4

    2019

  22. [30]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis.Communications of the ACM, 65(1):99–106, 2021

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis.Communications of the ACM, 65(1):99–106, 2021. 1, 2 7

    2021

  23. [31]

    Dynamicfusion: Reconstruction and tracking of non-rigid scenes in real-time

    Richard A Newcombe, Dieter Fox, and Steven M Seitz. Dynamicfusion: Reconstruction and tracking of non-rigid scenes in real-time. InCVPR, 2015. 1, 2

    2015

  24. [32]

    Holoportation: Virtual 3d teleportation in real-time

    Sergio Orts-Escolano, Christoph Rhemann, Sean Fanello, Wayne Chang, Adarsh Kowdle, Yury Degtyarev, David Kim, Philip L Davidson, Sameh Khamis, Mingsong Dou, et al. Holoportation: Virtual 3d teleportation in real-time. InUIST,

  25. [33]

    Barron, Sofien Bouaziz, Dan B Goldman, Steven M

    Keunhong Park, Utkarsh Sinha, Jonathan T. Barron, Sofien Bouaziz, Dan B Goldman, Steven M. Seitz, and Ricardo Martin-Brualla. Nerfies: Deformable neural radiance fields. InICCV, 2021. 2

    2021

  26. [34]

    Hypernerf: A higher- dimensional representation for topologically varying neural radiance fields.arXiv preprint arXiv:2106.13228, 2021

    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.arXiv preprint arXiv:2106.13228, 2021. 2

    2021 arXiv

  27. [35]

    Neural body: Implicit neural representations with structured latent codes for novel view synthesis of dynamic humans

    Sida Peng, Yuanqing Zhang, Yinghao Xu, Qianqian Wang, Qing Shuai, Hujun Bao, and Xiaowei Zhou. Neural body: Implicit neural representations with structured latent codes for novel view synthesis of dynamic humans. InProceed- ings of the IEEE/CVF Conference on Computer Vision an...

    2021

  28. [36]

    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. InICCV, 2021. 2

    2021

  29. [37]

    Swings: Sliding windows for dynamic 3d gaussian splatting

    Richard Shaw, Jifei Song, Arthur Moreau, Michal Nazarczuk, Sibi Catley-Chandar, Helisa Dhamo, and Ed- uardo P´erez-Pellitero. Swings: Sliding windows for dynamic 3d gaussian splatting. 2023. 1, 3, 5

    2023

  30. [38]

    Novel view synthesis of human interactions from sparse multi-view videos

    Qing Shuai, Chen Geng, Qi Fang, Sida Peng, Wenhao Shen, Xiaowei Zhou, and Hujun Bao. Novel view synthesis of human interactions from sparse multi-view videos. InSIG- GRAPH Conference Proceedings, 2022. 2

    2022

  31. [39]

    Nerf- player: A streamable dynamic scene representation with de- composed neural radiance fields.IEEE Transactions on Visu- alization and Computer Graphics, 29(5):2732–2742, 2023

    Liangchen Song, Anpei Chen, Zhong Li, Zhang Chen, Lele Chen, Junsong Yuan, Yi Xu, and Andreas Geiger. Nerf- player: A streamable dynamic scene representation with de- composed neural radiance fields.IEEE Transactions on Visu- alization and Computer Graphics, 29(5):2732–2742, 2...

    2023

  32. [40]

    Variable bitrate neural fields.ACM SIGGRAPH 2022 Con- ference Proceedings, 2022

    Towaki Takikawa, Alex Evans, Jonathan Tremblay, Thomas M¨uller, Morgan McGuire, Alec Jacobson, and Sanja Fidler. Variable bitrate neural fields.ACM SIGGRAPH 2022 Con- ference Proceedings, 2022. 2

    2022

  33. [41]

    Compressible-composable nerf via rank-residual de- composition.arXiv preprint arXiv:2205.14870, 2022

    Jiaxiang Tang, Xiaokang Chen, Jingbo Wang, and Gang Zeng. Compressible-composable nerf via rank-residual de- composition.arXiv preprint arXiv:2205.14870, 2022. 2

    2022 arXiv

  34. [42]

    Mixed neural voxels for fast multi-view video syn- thesis.2023 IEEE/CVF International Conference on Com- puter Vision (ICCV), pages 19649–19659, 2022

    Feng Wang, Sinan Tan, Xinghang Li, Zeyue Tian, and Huap- ing Liu. Mixed neural voxels for fast multi-view video syn- thesis.2023 IEEE/CVF International Conference on Com- puter Vision (ICCV), pages 19649–19659, 2022. 5, 1, 4

    2023

  35. [43]

    Image quality assessment: from error visibility to structural similarity.IEEE TIP, 2004

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity.IEEE TIP, 2004. 4, 5

    2004

  36. [44]

    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 Wang Xinggang. 4d gaussian splatting for real-time dynamic scene rendering. arXiv preprint arXiv:2310.08528, 2023. 1, 2, 3, 5, 4

    2023 arXiv

  37. [45]

    4k4d: Real-time 4d view synthesis at 4k resolution.2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 20029–20040, 2023

    Zhen Xu, Sida Peng, Haotong Lin, Guangzhao He, Ji- aming Sun, Yujun Shen, Hujun Bao, and Xiaowei Zhou. 4k4d: Real-time 4d view synthesis at 4k resolution.2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 20029–20040, 2023. 5, 2

    2024

  38. [46]

    Easyvolcap: Accelerating neural volumetric video research

    Zhen Xu, Tao Xie, Sida Peng, Haotong Lin, Qing Shuai, Zhiyuan Yu, Guangzhao He, Jiaming Sun, Hujun Bao, and Xiaowei Zhou. Easyvolcap: Accelerating neural volumetric video research. 2023. 9

    2023

  39. [47]

    Representing long volumet- ric video with temporal gaussian hierarchy.ACM Transac- tions on Graphics, 43(6), 2024

    Zhen Xu, Yinghao Xu, Zhiyuan Yu, Sida Peng, Jiaming Sun, Hujun Bao, and Xiaowei Zhou. Representing long volumet- ric video with temporal gaussian hierarchy.ACM Transac- tions on Graphics, 43(6), 2024. 3

    2024

  40. [48]

    Deformable 3d gaussians for high-fidelity monocular dynamic scene reconstruction.arXiv preprint arXiv:2309.13101, 2023

    Ziyi Yang, Xinyu Gao, Wen Zhou, Shaohui Jiao, Yuqing Zhang, and Xiaogang Jin. Deformable 3d gaussians for high-fidelity monocular dynamic scene reconstruction.arXiv preprint arXiv:2309.13101, 2023. 1, 2, 3

    2023 arXiv

  41. [49]

    Real-time photorealistic dynamic scene representa- tion and rendering with 4d gaussian splatting.arXiv preprint arXiv 2310.10642, 2023

    Zeyu Yang, Hongye Yang, Zijie Pan, Xiatian Zhu, and Li Zhang. Real-time photorealistic dynamic scene representa- tion and rendering with 4d gaussian splatting.arXiv preprint arXiv 2310.10642, 2023. 1, 3, 5, 8, 2, 4

    2023 arXiv

  42. [50]

    Bodyfu- sion: Real-time capture of human motion and surface geom- etry using a single depth camera

    Tao Yu, Kaiwen Guo, Feng Xu, Yuan Dong, Zhaoqi Su, Jian- hui Zhao, Jianguo Li, Qionghai Dai, and Yebin Liu. Bodyfu- sion: Real-time capture of human motion and surface geom- etry using a single depth camera. InThe IEEE International Conference on Computer Vision (ICCV). IEEE, 2017. 2

    2017

  43. [51]

    Doublefu- sion: Real-time capture of human performances with inner body shapes from a single depth sensor

    Tao Yu, Zerong Zheng, Kaiwen Guo, Jianhui Zhao, Qionghai Dai, Hao Li, Gerard Pons-Moll, and Yebin Liu. Doublefu- sion: Real-time capture of human performances with inner body shapes from a single depth sensor. InCVPR, 2018. 1, 2

    2018

  44. [52]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 4, 5

    2018

  45. [53]

    Motiongs: Exploring explicit motion guidance for deformable 3d gaussian splatting.arXiv preprint arXiv:2410.07707, 2024

    Ruijie Zhu, Yanzhe Liang, Hanzhi Chang, Jiacheng Deng, Jiahao Lu, Wenfei Yang, Tianzhu Zhang, and Yongdong Zhang. Motiongs: Exploring explicit motion guidance for deformable 3d gaussian splatting.arXiv preprint arXiv:2410.07707, 2024. 1, 3 8

    arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.