Pith. sign in

REVIEW 2 major objections 3 minor 28 references

Future rendering quality does not track future surface accuracy in dynamic-scene reconstruction; a new benchmark measures the gap.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 07:19 UTC pith:JWIJZ4QO

load-bearing objection FutureSurf is a genuinely useful evaluation contribution with a clean protocol, but the abstract overstates two results and the decoupling claim needs cross-method evidence. the 2 major comments →

arxiv 2607.21471 v1 pith:JWIJZ4QO submitted 2026-07-23 cs.CV

Future Rendering neq Future Surface: A Benchmark and Dataset for Dynamic Surface Reconstruction Beyond the Observed Window

Yukun Shi , Minglun Gong This is my paper
classification cs.CV
keywords future-surface reconstructiondynamic scenesbenchmarkextrapolationChamfer distancerendering-surface decouplingnovel view synthesisdiagnostic evaluation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper argues that the standard way of evaluating dynamic-scene reconstruction—checking how well a model renders held-out future frames—misses the geometry that deployment actually needs. It introduces a controlled benchmark with analytically defined motions and exact future ground-truth meshes, and shows that two dynamic-Gaussian backbones reconstruct the observed surface accurately but degrade 2.0–6.6× when scored on the future surface mesh. The rendering metrics (PSNR, LPIPS) are weakly rank-correlated with future surface error (mean |ρ|=0.13), so a future rendering can look plausible while the surface is wrong where it moves. If correct, the field's novel-view-synthesis metrics are not a proxy for future-surface quality, and future-time reconstruction needs separate geometric evaluation and stronger temporal inductive bias.

Core claim

The central discovery is a measured decoupling: for the tested dynamic-surface backbones, future rendering quality and future-surface accuracy do not move together. On six animated asset scenes and a suite of eight controlled motions with exact future ground truth, per-frame future Chamfer distance is 2.0–6.6× the observed-window error, and per-frame |ρ(LPIPS,CD)| averages 0.13, with a linear fit explaining under 6% of variance. The future error is structured, concentrating where the surface moves, and it persists after removing a global Sim(3) gauge, so it is genuine non-rigid shape error. The paper frames this as evidence that the field's standard evaluation protocol measures the wrong qua

What carries the argument

The benchmark's load-bearing instrument is the future/observed gap: the ratio of per-frame bidirectional Chamfer distance between extracted and ground-truth meshes on the held-out future window to the same quantity on the observed window. It converts 'how good is the future surface' into a normalized diagnostic that isolates extrapolation failure from interpolation quality, and it is paired with three falsification controls (a surface-invariant twist, a rigid-rotation gauge control, and a frozen-future stop control) designed to expose a broken metric or alignment. A second instrument, the ground-truth-side recoverability oracle, fits simple and learned per-vertex temporal rules to observed g

Load-bearing premise

The paper frames the measured future-surface gap and rendering–surface decoupling as a property of dynamic-scene reconstruction in general, but both tested backbones share the same time-conditioned deformation-MLP temporal model; if a temporally distinct representation extrapolates accurately, the gap would be an artifact of that architecture family rather than a field-wide finding.

What would settle it

Train a dynamic reconstruction method with a temporally distinct representation (e.g., an explicit 4D grid or physics-based simulation) on the released FutureSurf splits; if its future Chamfer gap drops to near unity (future/observed ≈ 1) on the recoverable controlled motions and its per-frame |ρ(LPIPS,CD)| rises substantially, the paper's claim of a general field-wide decoupling would be refuted.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The standard practice of reporting PSNR/SSIM/LPIPS on future frames does not certify future-surface quality; a separate per-frame mesh metric is needed for deployment claims.
  • Any method evaluated only inside the observed window may overstate its usefulness for future-time tasks such as AR overlays, robot interaction, and anticipatory planning.
  • The gap persists across two deformation-MLP backbones and six varied scenes, suggesting the limitation is not scene-specific but tied to the temporal model family.
  • The error's concentration where the surface moves implies that future-surface failure is predictable in location, and that motion-aware diagnostics (per-vertex maps) should accompany scores.
  • The recoverability oracle shows that some futures are known in principle from observed motion; the gap on those is a representation or extrapolation issue, not an information-theoretic one.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural next test is to train a temporally distinct representation—for example, an explicit 4D grid, a physics-based simulator, or a neural-SDF flow—on the released splits; if its future/observed gap drops to near unity on the recoverable controlled motions, the paper's family-level conclusion would be refined into an architecture-specific one.
  • The decoupling result implies that prior dynamic-scene forecasting papers that report only rendering metrics may be unknowingly releasing methods whose future geometry is poor; re-evaluating them with the released mesh protocol would be a high-value, low-cost extension.
  • The benchmark's synthetic-only design could be extended toward real captures by using a fitted proxy for future ground truth (e.g., a high-fidelity offline reconstruction) or depth sensors, though the paper's point that exact future GT requires analytic motion suggests a hybrid evaluation may be needed.
  • If the gap generalizes to other architecture families, it would motivate treating future-surface accuracy as a first-class benchmark axis alongside rendering, and could drive new training objectives that include temporal-extrapolation regularization or drift penalties on static futures.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 3 minor

Summary. FutureSurf proposes a standardized benchmark and dataset for evaluating dynamic surface reconstruction at held-out future times. It contributes eight analytically defined controlled motions (five surface-changing, three falsification controls) with exact per-frame ground-truth meshes, a 75/25 train/future split, a method-agnostic per-frame Chamfer protocol, a ground-truth-side recoverability oracle, benchmark card, and Croissant metadata. Using DG-Mesh and Deformable-3DGS as backbones, the paper reports a 2.0–6.6× future/observed Chamfer gap on asset scenes and 2.7–4.1× on the controlled motions, with falsification controls behaving as designed; a per-motion oracle recovers four of five constructed futures using simple matched rules; and a within-method analysis finds weak rank correlation between future rendering quality (LPIPS) and future-surface CD (mean |ρ|=0.13), which the paper interprets as a rendering–surface decoupling. The authors release splits, scoring code, and metadata. The paper is explicitly framed as a diagnostic benchmark rather than a new reconstruction method.

Significance. If taken as scoped, this is a valuable contribution: it makes an unmeasured quantity—future-surface mesh accuracy—measurable, and provides exact ground truth, falsification controls, and reproducible scripts. The recoverability oracle is honestly scoped as an optimistic, ground-truth-side reference, and the paper carefully separates removable gauge error from non-rigid error. The main limitations are the breadth of tested temporal models (both backbones share a deformation-MLP temporal family) and the within-method nature of the decoupling evidence; these limit the generality of the headline 'Future Rendering ≠ Future Surface' but do not undermine the benchmark itself. The release of code, splits, and a CPU-run oracle is a concrete strength for reproducibility.

major comments (2)
  1. [Abstract and §1 (Contribution 2) vs §6] The abstract and Contribution 2 state that 'four of five recoverable from observed motion by a fixed rule.' This is contradicted by §6: Table 6 shows the best rule varies per motion (harmonic K=2 for wave/compound, cubic for stretch, quadratic for accel, velocity for bulge) and the text explicitly says 'No single per-vertex extrapolation rule covers the suite.' The oracle establishes recoverability only under per-motion matched rule families, not a fixed rule. This phrasing is load-bearing because it is used to argue that the 2.7–4.1× gap reflects a backbone limitation rather than intrinsic unknowability. Please rephrase to 'per-motion matched rules' or 'simple rules' and align the abstract/contributions with §6.
  2. [§7, Table 4] The decoupling conclusion is based on per-frame Spearman correlations computed within each DG-Mesh scene over future frames (mean |ρ(LPIPS,CD)| = 0.13). This supports only the within-method statement that, for DG-Mesh, rendering quality does not track surface error over the prediction horizon. It does not support the broader claim that 'the NVS metrics the field reports do not track future geometry,' which is a cross-method claim. No cross-method ranking analysis is provided; the second backbone is not included in the decoupling analysis, and §1 itself states the decoupling is reported on DG-Mesh. Please either add a cross-method analysis (e.g., rank-correlating method-level future PSNR/LPIPS with future CD across methods/scenes) or restrict the claim to the tested DG-Mesh backbone.
minor comments (3)
  1. [§3, §5] Spacing typos throughout: e.g., 'a monocular orbit camera,200frames, the first75%' (§3) and 'gap:2.7–4.1×by mean' (§5). These are likely LaTeX artifacts but should be cleaned before publication.
  2. [Table 4] The column header 'Non-rigid' should explicitly say it is the Sim(3)-gauge-removed future gap; otherwise a reader may confuse it with a non-rigid motion class. The caption explains it, but the header itself is ambiguous.
  3. [§6, Table 6] The oracle CD is bounding-box-normalized linear Chamfer, which the text says is on a different scale from Table 5's absolute mesh CD. It would help to state in the table caption that the ratios vs. the freeze reference are the primary comparison, not the absolute CD values.

Circularity Check

0 steps flagged

No significant circularity: the paper's central claims are empirical benchmark measurements, not derivations from fitted inputs or self-citations.

full rationale

FutureSurf is an evaluation contribution. Its headline results—future CD gaps of 2.0–6.6× and the weak per-frame rendering–surface rank correlation (|ρ(LPIPS,CD)|=0.13)—are direct measurements on trained backbones scored against exact analytic ground-truth meshes with released CPU scoring code. No parameter is fitted to a target quantity and then renamed as a prediction; the recoverability oracle is explicitly a ground-truth-side diagnostic ('This is deliberately an oracle evaluation... not a baseline method'), and its rule families are matched to the analytic construction, so it does not enter the gap or decoupling measurements. The paper contains no load-bearing self-citations; all cited prior work is external. The acknowledged limitation that both backbones share a deformation-MLP temporal model, and the concern that the decoupling is reported only within one backbone over time, are threats to external validity, not circularity. The falsification controls are sanity checks of the protocol rather than derived predictions. The derivation chain is therefore self-contained and the empirical claims stand on their own measurements.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The paper's conclusions are empirical measurements on a purpose-built dataset; no physical entities are invented. The central load-bearing inputs are exact-ground-truth assumption of the analytic motions, the Chamfer protocol, the enforcement of the temporal split, and the representativeness of the two deformation-MLP backbones. The oracle's rule families are fitted to observed GT trajectories and constitute an optimistic, disclosed reference.

free parameters (3)
  • Controlled motion parameters = e.g., wave amplitude 0.22, bulge amplitude 0.34, bulge width 0.28
    Hand-chosen amplitudes and frequencies define the dataset; they affect absolute Chamfer magnitudes and gap sizes but are not fitted to the conclusions.
  • Split ratio (75/25) = 75% observed / 25% future
    Default split; the paper reports the gap persists at 60/40 and 85/15, so it is a design choice rather than a fitted parameter.
  • Oracle rule families per motion = wave/compound harmonic K=2; stretch cubic; accel quadratic; bulge ≈ freeze
    Rule families and coefficients fitted to observed GT vertex trajectories; used only for the recoverability oracle, which is explicitly optimistic and not a baseline.
axioms (4)
  • domain assumption The analytic motion maps F_m(x,t) define exact per-frame ground-truth surfaces.
    §3; necessary for exact future GT; if generator equations are implemented incorrectly, all GT and oracle results fail.
  • standard math Bidirectional Chamfer on vertex sets in a shared canonical frame is a valid measure of surface accuracy for the primary score.
    §4 Eq. 1; protocol choice; reasonable but not the only possible geometry metric.
  • domain assumption The tested backbones (DG-Mesh, Deformable-3DGS) are representative of the class of dynamic reconstruction methods.
    §5, §8; both share a deformation-MLP temporal model, so this assumption is fragile; explicitly acknowledged as a limitation. The central 'gap' generalization rests on it.
  • domain assumption Training on observed frames only is enforced; backbones do not observe future frames during training.
    §4 task/split; the protocol's validity depends on the split being respected.

pith-pipeline@v1.3.0-alltime-deepseek · 11402 in / 12404 out tokens · 112438 ms · 2026-08-01T07:19:30.172145+00:00 · methodology

0 comments
read the original abstract

Dynamic-scene reconstruction is almost always evaluated inside the observed time window, yet deployment settings such as AR overlays, robot interaction, and anticipatory planning need the future surface: the geometry at times beyond those captured. No standard benchmark measures this. We introduce FutureSurf, a controlled diagnostic benchmark and dataset for future-time surface reconstruction that trades scene diversity for exact future ground truth and falsification controls. A method trains on the observed first 75% of a sequence; we score its extracted per-frame surface on the held-out future by Chamfer distance, reporting absolute future CD as the primary score and the future/observed gap as a diagnostic. The dataset contains eight analytically defined controlled motions, including three falsification controls, with exact per-frame ground-truth meshes. We also provide a ground-truth-side recoverability oracle. The release includes split files, scoring code, a benchmark card, and Croissant metadata. On the controlled motions, the DG-Mesh backbone leaves a 2.7-4.1$\times$ gap even for futures predictable in principle (four of five recoverable from observed motion by a fixed rule), while the falsification controls behave as designed (the surface-invariant motion shows no gap). Beyond the contributed dataset, the gap persists across six animated DG-Mesh asset scenes and a second backbone, Deformable-3DGS (2.0-6.6$\times$; both share a deformation-MLP temporal model). The benchmark also shows that future rendering quality and future-surface accuracy are statistically decoupled, so the novel-view-synthesis metrics the field reports do not track future geometry. The future error is structured, concentrating where the surface moves. The dataset, evaluation toolkit, and scoring code are available on Hugging Face and GitHub (https://github.com/Ricky-S/futuresurf).

Figures

Figures reproduced from arXiv: 2607.21471 by Minglun Gong, Yukun Shi.

Figure 1
Figure 1. Figure 1: FUTURESURF: does the reconstructed surface survive beyond the observed window? Three of our eight controlled motions, shown as ground-truth mesh sequences (red arrows: ground-truth surface velocity), isolate periodic motion, localized deformation, and a rigid-motion control. A method trains on the first 75% of timestamps (observed) and is scored on its extracted surface over the held-out future (25%). Beca… view at source ↗
Figure 2
Figure 2. Figure 2: Future rendering ̸= future surface (duck, DG-Mesh, far-horizon future frame 190). Left pair: the predicted rendering remains a recognizable duck (PSNR = 17.9 dB). Middle: the reconstructed surface at the same frame is substantially wrong where the surface moves (per-vertex error peaks on the paddling foot). Right: over the whole future window, per-frame rendering quality stays in a narrow band while per-fr… view at source ↗
Figure 3
Figure 3. Figure 3: The controlled-motion dataset (representative frames): [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: On the analytic surfaces, the second backbone is a point cloud, not a surface mesh. Cross-sections of four con￾structed motions: DG-Mesh’s mesh (green) lies on the ground￾truth surface (black); Deformable-3DGS’s Gaussian centers (red) form a loose cloud 2–7× farther from the surface than DG-Mesh’s across all eight scenes (median over observed frames). meshes, the controlled-motion gap and recoverability nu… view at source ↗
Figure 5
Figure 5. Figure 5: The stop control over the prediction horizon. With the motion frozen from t=0.5, per-frame future CD stays within ∼2× the observed-window level for ∼25 frames, then drifts: the defor￾mation field extrapolates non-constantly even though the observed motion has already stopped. tions all show large future gaps: 2.7–4.1× by mean and 3.2–4.6× by median, across periodic, period+trend, mono￾tonic, localized, and… view at source ↗
Figure 6
Figure 6. Figure 6: Future-surface degradation over the prediction horizon [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 15 linked inside Pith

  1. [1]

    Croissant: A metadata format for ML- ready datasets

    Mubashara Akhtar, Omar Benjelloun, Costanza Conforti, Luca Foschini, Pieter Gijsbers, Joan Giner-Miguelez, Su- jata Goswami, Nitisha Jain, Michalis Karamousadakis, Satyapriya Krishna, Michael Kuchnik, Sylvain Lesage, Quentin Lhoest, Pierre Marcenac, Manil Maskey, Peter Mattson, Luis Oala, Hamidah Oderinwale, Pierre Ruyssen, Tim Santos, Rajat Shinde, Elena...

    2024

  2. [2]

    4D Gaussian splatting as a learned dynamical system.arXiv preprint arXiv:2512.19648, 2025

    Arnold Caleb Asiimwe and Carl V ondrick. 4D Gaussian splatting as a learned dynamical system.arXiv preprint arXiv:2512.19648, 2025. 3

    arXiv 2025

  3. [3]

    GIC: Gaussian-informed continuum for physical property identi- fication and simulation

    Junhao Cai, Yuji Yang, Weihao Yuan, Yisheng He, Zilong Dong, Liefeng Bo, Hui Cheng, and Qifeng Chen. GIC: Gaussian-informed continuum for physical property identi- fication and simulation. InAdvances in Neural Information Processing Systems (NeurIPS), 2024. arXiv:2406.14927. 3

    Pith/arXiv arXiv 2024

  4. [4]

    DynaSurfGS: Dynamic surface reconstruction with planar- based Gaussian splatting.arXiv preprint arXiv:2408.13972,

    Weiwei Cai, Weicai Ye, Peng Ye, Tong He, and Tao Chen. DynaSurfGS: Dynamic surface reconstruction with planar- based Gaussian splatting.arXiv preprint arXiv:2408.13972,

    Pith/arXiv arXiv

  5. [5]

    Monocular dynamic view synthesis: A reality check

    Hang Gao, Ruilong Li, Shubham Tulsiani, Bryan Russell, and Angjoo Kanazawa. Monocular dynamic view synthesis: A reality check. InNeurIPS, 2022. 4

    2022

  6. [6]

    Fluid- Nexus: 3D fluid reconstruction and prediction from a single video

    Yue Gao, Hong-Xing Yu, Bo Zhu, and Jiajun Wu. Fluid- Nexus: 3D fluid reconstruction and prediction from a single video. InCVPR, 2025. arXiv:2503.04720. 3

    Pith/arXiv arXiv 2025

  7. [7]

    Datasheets for datasets.Communications of the ACM, 2021

    Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jen- nifer Wortman Vaughan, Hanna Wallach, Hal Daum´e III, and Kate Crawford. Datasheets for datasets.Communications of the ACM, 2021. 3

    2021

  8. [8]

    Klaus Greff, Francois Belletti, Lucas Beyer, Carl Do- ersch, Yilun Du, Daniel Duckworth, David J. Fleet, Dan Gnanapragasam, Florian Golemo, Charles Herrmann, Thomas Kipf, Abhijit Kundu, Dmitry Lagun, Issam Laradji, Hsueh-Ti Derek Liu, Henning Meyer, Yishu Miao, Derek Nowrouzezahrai, Cengiz Oztireli, Etienne Pot, Noha Rad- wan, Daniel Rebain, Sara Sabour, ...

    2022

  9. [9]

    2D Gaussian splatting for geometrically ac- curate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2D Gaussian splatting for geometrically ac- curate radiance fields. InSIGGRAPH 2024 Conference Pa- pers, 2024. 3

    2024

  10. [10]

    3D Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics,

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3D Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics,

  11. [11]

    GP- 4DGS: Probabilistic 4D Gaussian splatting from monocular video via variational Gaussian processes

    Mijeong Kim, Jungtaek Kim, and Bohyung Han. GP- 4DGS: Probabilistic 4D Gaussian splatting from monocular video via variational Gaussian processes. InCVPR, 2026. arXiv:2604.02915. 3

    Pith/arXiv arXiv 2026

  12. [12]

    DySurface: Consistent 4D surface reconstruction via bridging explicit Gaussians and implicit functions.arXiv preprint arXiv:2605.10360, 2026

    Minje Kim, Younghyun Noh, Jaesoon Kim, and Tae-Kyun Kim. DySurface: Consistent 4D surface reconstruction via bridging explicit Gaussians and implicit functions.arXiv preprint arXiv:2605.10360, 2026. 3

    Pith/arXiv arXiv 2026

  13. [13]

    Space-time forecasting of dynamic scenes with motion-aware Gaussian grouping

    Junmyeong Lee, Hoseung Choi, and Minsu Cho. Space-time forecasting of dynamic scenes with motion-aware Gaussian grouping. InCVPR, 2026. arXiv:2602.21668. 3

    Pith/arXiv arXiv 2026

  14. [14]

    Dynamic Gaus- sians mesh: Consistent mesh reconstruction from dynamic scenes

    Isabella Liu, Hao Su, and Xiaolong Wang. Dynamic Gaus- sians mesh: Consistent mesh reconstruction from dynamic scenes. InInternational Conference on Learning Represen- tations (ICLR), 2025. arXiv:2404.12379. 3, 4

    Pith/arXiv arXiv 2025

  15. [15]

    Model cards for model reporting

    Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, In- ioluwa Deborah Raji, and Timnit Gebru. Model cards for model reporting. InProceedings of the Conference on Fair- ness, Accountability, and Transparency (FAT* ’19), 2019. 3

    2019

  16. [16]

    Barron, Sofien Bouaziz, Dan B

    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, 2021. Proc. SIGGRAPH Asia 2021. 3

    2021

  17. [17]

    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. InCVPR, 2021. 3

    2021

  18. [18]

    Recon- Phys: Reconstruct appearance and physical attributes from single video.arXiv preprint arXiv:2604.07882, 2026

    Boyuan Wang, Xiaofeng Wang, Yongkang Li, Zheng Zhu, Yifan Chang, Angen Ye, Guosheng Zhao, Chaojun Ni, Guan Huang, Yijie Ren, Yueqi Duan, and Xingang Wang. Recon- Phys: Reconstruct appearance and physical attributes from single video.arXiv preprint arXiv:2604.07882, 2026. 3

    Pith/arXiv arXiv 2026

  19. [19]

    ODE-GS: Latent ODEs for dynamic scene extrapolation with 3D Gaussian splat- ting

    Daniel Wang, Patrick Rim, Tian Tian, Dong Lao, Alex Wong, and Ganesh Sundaramoorthi. ODE-GS: Latent ODEs for dynamic scene extrapolation with 3D Gaussian splat- ting. InInternational Conference on Learning Representa- tions (ICLR), 2026. arXiv:2506.05480. 3

    Pith/arXiv arXiv 2026

  20. [20]

    Space-time 2D Gaussian splatting for accurate sur- face reconstruction under complex dynamic scenes.arXiv preprint arXiv:2409.18852, 2024

    Shuo Wang, Binbin Huang, Ruoyu Wang, and Shenghua Gao. Space-time 2D Gaussian splatting for accurate sur- face reconstruction under complex dynamic scenes.arXiv preprint arXiv:2409.18852, 2024. 3

    Pith/arXiv arXiv 2024

  21. [21]

    DUSt3R: Geometric 3D vision made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and J ´erˆome Revaud. DUSt3R: Geometric 3D vision made easy. InCVPR, 2024. arXiv:2312.14132. 3

    Pith/arXiv arXiv 2024

  22. [22]

    4D Gaussian splatting for real-time dynamic scene render- ing

    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 render- ing. InCVPR, 2024. arXiv:2310.08528. 3

    Pith/arXiv arXiv 2024

  23. [23]

    Alvarez, and Miaomiao Liu

    Renjie Wu, Hongdong Li, Jose M. Alvarez, and Miaomiao Liu. 4DSurf: High-fidelity dynamic scene surface recon- struction. InCVPR, 2026. arXiv:2603.28064. 3

    arXiv 2026

  24. [24]

    PhysGaussian: Physics- integrated 3D Gaussians for generative dynamics

    Tianyi Xie, Zeshun Zong, Yuxing Qiu, Xuan Li, Yutao Feng, Yin Yang, and Chenfanfu Jiang. PhysGaussian: Physics- integrated 3D Gaussians for generative dynamics. InCVPR,

  25. [25]

    Deformable 3D Gaussians for 9 high-fidelity monocular dynamic scene reconstruction

    Ziyi Yang, Xinyu Gao, Wen Zhou, Shaohui Jiao, Yuqing Zhang, and Xiaogang Jin. Deformable 3D Gaussians for 9 high-fidelity monocular dynamic scene reconstruction. In CVPR, 2024. arXiv:2309.13101. 3

    Pith/arXiv arXiv 2024

  26. [26]

    Gaus- sianPrediction: Dynamic 3D Gaussian prediction for motion extrapolation and free view synthesis

    Boming Zhao, Yuan Li, Ziyu Sun, Lin Zeng, Yujun Shen, Rui Ma, Yinda Zhang, Hujun Bao, and Zhaopeng Cui. Gaus- sianPrediction: Dynamic 3D Gaussian prediction for motion extrapolation and free view synthesis. InSIGGRAPH 2024 Conference Papers, 2024. 3

    2024

  27. [27]

    GauSTAR: Gaussian surface tracking and reconstruction

    Chengwei Zheng, Lixin Xue, Juan Zarate, and Jie Song. GauSTAR: Gaussian surface tracking and reconstruction. In CVPR, 2025. arXiv:2501.10283. 3

    Pith/arXiv arXiv 2025

  28. [28]

    Reconstruction and simulation of elastic objects with spring- mass 3D Gaussians

    Licheng Zhong, Hong-Xing Yu, Jiajun Wu, and Yunzhu Li. Reconstruction and simulation of elastic objects with spring- mass 3D Gaussians. InEuropean Conference on Computer Vision (ECCV), 2024. arXiv:2403.09434. 3 10

    Pith/arXiv arXiv 2024