pith. sign in

arxiv: 2506.05480 · v4 · submitted 2025-06-05 · 💻 cs.GR · cs.CV· cs.LG

ODE-GS: Latent ODEs for Dynamic Scene Extrapolation with 3D Gaussian Splatting

Daniel Wang , Patrick Rim , Tian Tian , Dong Lao , Alex Wong , Ganesh Sundaramoorthi This is my paper

Pith reviewed 2026-05-19 11:03 UTC · model grok-4.3

classification 💻 cs.GR cs.CVcs.LG
keywords 3D Gaussian Splattingneural ODEdynamic scene extrapolationlatent dynamicsTransformer encodercontinuous-time modelingscene reconstruction
0
0 comments X

The pith

Modeling Gaussian parameter trajectories as continuous latent dynamics with a neural ODE enables extrapolation of dynamic 3D scenes beyond observed times.

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

The paper sets out to show that 3D scenes captured with Gaussian splatting can be predicted at future times by representing the parameters of each splat as evolving along a continuous path in latent space. Instead of conditioning on explicit timestamps as prior deformation networks do, the method encodes observed trajectories with a Transformer and lets a neural ODE carry the latent state forward. A sympathetic reader would care because this removes the hard limit on prediction range that comes from fixed training windows and supports rendering at any chosen future moment. The workflow first fits accurate trajectories inside the training interval, then integrates the ODE to produce new states for novel times.

Core claim

ODE-GS first learns an interpolation model to generate accurate Gaussian trajectories within the observed window, then trains a Transformer encoder to aggregate past trajectories into a latent state evolved via a neural ODE. Numerical integration of this ODE produces smooth future Gaussian trajectories that can be rendered at arbitrary timestamps.

What carries the argument

A neural ODE that evolves a Transformer-encoded latent state to generate future Gaussian parameter trajectories.

If this is right

  • Dynamic scenes can be rendered at future timestamps outside the training interval without retraining.
  • Extrapolation metrics improve by 19.8 percent over leading baselines on D-NeRF, NVFi, and HyperNeRF.
  • Generated trajectories remain continuous and smooth because they follow the learned latent dynamics.
  • Scene prediction no longer requires a fixed time window or direct time inputs for new frames.

Where Pith is reading between the lines

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

  • The same latent-dynamics approach could be tested on other scene representations such as explicit meshes or point clouds for longer-range prediction.
  • Numerical integration errors might accumulate over very long horizons, suggesting experiments that measure drift as a function of prediction length.
  • Adding physics-based regularizers to the ODE could further constrain predicted motions to obey conservation laws not present in the training data.

Load-bearing premise

A Transformer-encoded latent state evolved by a neural ODE will produce accurate and stable Gaussian trajectories at future times without any explicit timestamp conditioning or additional regularization beyond the training procedure described.

What would settle it

If rendered frames at times well beyond the training window on a held-out sequence with accelerating or colliding objects show drifting shapes or sudden discontinuities while time-conditioned baselines remain coherent, the central claim would be falsified.

Figures

Figures reproduced from arXiv: 2506.05480 by Alex Wong, Daniel Wang, Dong Lao, Ganesh Sundaramoorthi, Patrick Rim, Tian Tian.

Figure 1
Figure 1. Figure 1: Overview of ODE-GS. 1: Trajectory Initialization learns a canonical set of 3D Gaussians [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Qualitative visualization on the mutant scene from DNeRF dataset, from left to right are [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
read the original abstract

We introduce ODE-GS, a novel approach that integrates 3D Gaussian Splatting with latent neural ordinary differential equations (ODEs) to enable future extrapolation of dynamic 3D scenes. Unlike existing dynamic scene reconstruction methods, which rely on time-conditioned deformation networks and are limited to interpolation within a fixed time window, ODE-GS eliminates timestamp dependency by modeling Gaussian parameter trajectories as continuous-time latent dynamics. Our approach first learns an interpolation model to generate accurate Gaussian trajectories within the observed window, then trains a Transformer encoder to aggregate past trajectories into a latent state evolved via a neural ODE. Finally, numerical integration produces smooth, physically plausible future Gaussian trajectories, enabling rendering at arbitrary future timestamps. On the D-NeRF, NVFi, and HyperNeRF benchmarks, ODE-GS achieves state-of-the-art extrapolation performance, improving metrics by 19.8% compared to leading baselines, demonstrating its ability to accurately represent and predict 3D scene dynamics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces ODE-GS, a method integrating 3D Gaussian Splatting with latent neural ODEs for extrapolating dynamic 3D scenes beyond observed time windows. It first trains an interpolation model on Gaussian trajectories within the observed window, then employs a Transformer encoder to aggregate past trajectories into a latent state that is evolved via a neural ODE; numerical integration of this state yields future Gaussian parameters for rendering at arbitrary future timestamps. The paper claims state-of-the-art extrapolation performance on the D-NeRF, NVFi, and HyperNeRF benchmarks, with a 19.8% metric improvement over leading baselines.

Significance. If the empirical claims hold under rigorous verification, the work would represent a meaningful advance in dynamic scene modeling by removing explicit timestamp conditioning and enabling continuous-time extrapolation of Gaussian parameters. The latent-ODE formulation offers a principled way to capture scene dynamics that could generalize better than discrete-time or time-conditioned deformation networks, with potential applications in video prediction and immersive graphics.

major comments (2)
  1. [Abstract] Abstract: the central claim that numerical integration of the Transformer-encoded latent state produces accurate and stable Gaussian trajectories at arbitrary future times is load-bearing, yet the manuscript provides no description of regularization (Lipschitz bounds, energy penalties, or divergence penalties) on the learned vector field; without such constraints the method risks drift or divergence for non-periodic motions, directly undermining the extrapolation results.
  2. [Method] Method section (ODE training and decoder stage): the decoder that maps the integrated latent state back to per-Gaussian parameters (position, scale, opacity, etc.) at a queried future time is not shown to be independent of the training window; if the decoder implicitly relies on patterns learned only from observed times, the reported gains on future timestamps may not generalize.
minor comments (2)
  1. Add error bars and statistical significance tests to all quantitative tables reporting the 19.8% improvement.
  2. Clarify the exact numerical integration scheme (e.g., Euler, RK4) and step-size schedule used for ODE solving in the extrapolation experiments.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and detailed comments on our manuscript. These observations highlight important aspects of stability and generalization in our latent ODE formulation. We address each major comment below and have revised the manuscript accordingly to improve clarity and rigor.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that numerical integration of the Transformer-encoded latent state produces accurate and stable Gaussian trajectories at arbitrary future times is load-bearing, yet the manuscript provides no description of regularization (Lipschitz bounds, energy penalties, or divergence penalties) on the learned vector field; without such constraints the method risks drift or divergence for non-periodic motions, directly undermining the extrapolation results.

    Authors: We agree that the absence of an explicit discussion on regularization of the vector field is a gap in the current manuscript. Our neural ODE is trained end-to-end to reconstruct observed Gaussian trajectories via the interpolation model and Transformer encoder, which empirically yields stable integration on the evaluated benchmarks. However, to strengthen the presentation, we will add a new paragraph in the Method section detailing the training loss, the use of standard ODE solvers with adaptive step sizing, and empirical evidence of trajectory stability over extended horizons. We will also report additional quantitative results on longer extrapolation intervals to demonstrate that drift remains limited in practice for the tested dynamic scenes. revision: yes

  2. Referee: [Method] Method section (ODE training and decoder stage): the decoder that maps the integrated latent state back to per-Gaussian parameters (position, scale, opacity, etc.) at a queried future time is not shown to be independent of the training window; if the decoder implicitly relies on patterns learned only from observed times, the reported gains on future timestamps may not generalize.

    Authors: The decoder is implemented as a time-independent MLP that receives solely the evolved latent state (after ODE integration) and directly regresses the full set of Gaussian attributes. No explicit time stamp or training-window-specific features are provided as input. This architectural choice is intended to ensure that extrapolation relies only on the learned continuous dynamics. We acknowledge that the current manuscript does not sufficiently emphasize this independence. In the revision we will expand the decoder description, include a clear statement that the decoder contains no time conditioning, and add a supplementary figure illustrating the data flow from integrated latent state to Gaussian parameters. revision: yes

Circularity Check

0 steps flagged

No circularity: two-stage separation keeps extrapolation independent of fitted interpolation parameters

full rationale

The paper's derivation proceeds in distinct stages: an interpolation model is first trained on the observed time window to produce Gaussian trajectories, after which a Transformer encoder aggregates those trajectories into a latent state that is evolved by a separately trained neural ODE. Numerical integration of the learned vector field then generates trajectories at future times. This structure does not reduce the reported extrapolation metrics to quantities defined by the same fitted parameters, nor does it rely on self-citations, imported uniqueness theorems, or ansatzes that collapse the central claim back to its inputs. The method is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on standard assumptions from neural ODE literature and 3D Gaussian Splatting; no new entities are introduced and free parameters are typical of deep learning training.

axioms (1)
  • domain assumption Gaussian parameters evolve according to continuous latent dynamics that can be captured by a neural ODE
    Invoked when the paper states that trajectories are modeled as continuous-time latent dynamics.

pith-pipeline@v0.9.0 · 5716 in / 1221 out tokens · 38973 ms · 2026-05-19T11:03:48.329468+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Space-Time Forecasting of Dynamic Scenes with Motion-aware Gaussian Grouping

    cs.CV 2026-02 unverdicted novelty 7.0

    MoGaF groups Gaussians by motion in 4D splatting representations to enable stable long-term forecasting of dynamic scenes.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Hyperreel: High-fidelity 6-dof video with ray-conditioned sampling

    Benjamin Attal, Jia-Bin Huang, Christian Richardt, Michael Zollhoefer, Johannes Kopf, Matthew O’Toole, and Changil Kim. Hyperreel: High-fidelity 6-dof video with ray-conditioned sampling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16610–16620, 2023

    work page 2023

  2. [2]

    Gaussianvideo: Ef- ficient video representation via hierarchical gaussian splatting.arXiv preprint arXiv:2501.04782, 2025

    Andrew Bond, Jui-Hsien Wang, Long Mai, Erkut Erdem, and Aykut Erdem. Gaussianvideo: Ef- ficient video representation via hierarchical gaussian splatting.arXiv preprint arXiv:2501.04782, 2025

    work page arXiv 2025

  3. [3]

    Immersive light field video with a layered mesh representation

    Michael Broxton, John Flynn, Ryan Overbeck, Daniel Erickson, Peter Hedman, Matthew Duvall, Jason Dourgarian, Jay Busch, Matt Whalen, and Paul Debevec. Immersive light field video with a layered mesh representation. ACM Transactions on Graphics (TOG), 39(4):86–1, 2020

    work page 2020

  4. [4]

    Hexplane: A fast representation for dynamic scenes

    Ang Cao and Justin Johnson. Hexplane: A fast representation for dynamic scenes. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 130–141, 2023

    work page 2023

  5. [5]

    Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, and David K. Duvenaud. Neural ordinary differential equations. In Advances in Neural Information Processing Systems (NeurIPS) , volume 31, pages 6571–6583, 2018

    work page 2018

  6. [6]

    GRU-ODE-Bayes: Con- tinuous modeling of sporadically-observed time series

    Edward De Brouwer, Jaak Simm, Adam Arany, and Yves Moreau. GRU-ODE-Bayes: Con- tinuous modeling of sporadically-observed time series. In Advances in Neural Information Processing Systems (NeurIPS), volume 32, pages 7366–7376, 2019

    work page 2019

  7. [7]

    Fu- sion4d: Real-time performance capture of challenging scenes

    Mingsong Dou, Sameh Khamis, Yury Degtyarev, Philip Davidson, Sean Ryan Fanello, Adarsh Kowdle, Sergio Orts Escolano, Christoph Rhemann, David Kim, Jonathan Taylor, et al. Fu- sion4d: Real-time performance capture of challenging scenes. ACM Transactions on Graphics (ToG), 35(4):1–13, 2016

    work page 2016

  8. [8]

    Neural radiance flow for 4d view synthesis and video processing

    Yilun Du, Yinan Zhang, Hong-Xing Yu, Joshua B Tenenbaum, and Jiajun Wu. Neural radiance flow for 4d view synthesis and video processing. In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 14304–14314. IEEE Computer Society, 2021

    work page 2021

  9. [9]

    Augmented neural ODEs

    Emilien Dupont, Arnaud Doucet, and Yee Whye Teh. Augmented neural ODEs. In Advances in Neural Information Processing Systems (NeurIPS) , volume 32, pages 3134–3144, 2019

    work page 2019

  10. [10]

    Fast dynamic radiance fields with time-aware neural voxels

    Jiemin Fang, Taoran Yi, Xinggang Wang, Lingxi Xie, Xiaopeng Zhang, Wenyu Liu, Matthias Nießner, and Qi Tian. Fast dynamic radiance fields with time-aware neural voxels. In SIG- GRAPH Asia 2022 Conference Papers, pages 1–9, 2022

    work page 2022

  11. [11]

    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 Vision and Pattern Recognition , pages 12479–12488, 2023

    work page 2023

  12. [12]

    Dynamic view synthesis from dynamic monocular video

    Chen Gao, Ayush Saraf, Johannes Kopf, and Jia-Bin Huang. Dynamic view synthesis from dynamic monocular video. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 5712–5721, 2021

    work page 2021

  13. [13]

    Learning neural volumetric representations of dynamic humans in minutes

    Chen Geng, Sida Peng, Zhen Xu, Hujun Bao, and Xiaowei Zhou. Learning neural volumetric representations of dynamic humans in minutes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8759–8770, 2023

    work page 2023

  14. [14]

    arXiv preprint arXiv:2410.11648 , year=

    Joshua Gubbi, Ben Leimkuhler, Charles Matthews, Atul Sharma, and Eng-Yeow Teo. Efficient, accurate and stable gradients for neural ODEs. arXiv preprint arXiv:2410.11648, 2024

    work page arXiv 2024

  15. [15]

    2d gaussian splatting for geometrically accurate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically accurate radiance fields. In ACM SIGGRAPH 2024 conference papers, pages 1–11, 2024. 10

    work page 2024

  16. [16]

    Neural fields in robotics: A survey

    Muhammad Zubair Irshad, Mauro Comi, Yen-Chen Lin, Nick Heppert, Abhinav Valada, Rares Ambrus, Zsolt Kira, and Jonathan Tremblay. Neural fields in robotics: A survey. arXiv preprint arXiv:2410.20220, 2024

    work page arXiv 2024

  17. [17]

    Conerf: Controllable neural radiance fields

    Kacper Kania, Kwang Moo Yi, Marek Kowalski, Tomasz Trzci´nski, and Andrea Tagliasacchi. Conerf: Controllable neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18623–18632, 2022

    work page 2022

  18. [18]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4):139–1, 2023

    work page 2023

  19. [19]

    Patrick Kidger, Ricky T. Q. Chen, and Miles Cranmer. torchode: A parallel ode solver for pytorch. https://github.com/patrick-kidger/torchode, 2021

    work page 2021

  20. [20]

    Nvfi: Neural velocity fields for 3d physics learning from dynamic videos

    Jinxi Li, Ziyang Song, and Bo Yang. Nvfi: Neural velocity fields for 3d physics learning from dynamic videos. Advances in Neural Information Processing Systems , 36:34723–34751, 2023

    work page 2023

  21. [21]

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

    Zhengqi Li, Simon Niklaus, Noah Snavely, and Oliver Wang. Neural scene flow fields for space-time view synthesis of dynamic scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6498–6508, 2021

    work page 2021

  22. [22]

    Neural volumes: Learning dynamic renderable volumes from images,

    Stephen Lombardi, Tomas Simon, Jason Saragih, Gabriel Schwartz, Andreas Lehrmann, and Yaser Sheikh. Neural volumes: Learning dynamic renderable volumes from images. arXiv preprint arXiv:1906.07751, 2019

    work page arXiv 1906

  23. [23]

    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 2024 International Conference on 3D Vision (3DV), pages 800–809. IEEE, 2024

    work page 2024

  24. [24]

    Nerf: Representing scenes as neural radiance fields for view synthesis

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

    work page 2021

  25. [25]

    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. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 343–352, 2015

    work page 2015

  26. [26]

    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. In Proceedings of the 29th annual symposium on user interface software and technology, pages 741–754, 2016

    work page 2016

  27. [27]

    Neural scene graphs for dynamic scenes

    Julian Ost, Fahim Mannan, Nils Thuerey, Julian Knodt, and Felix Heide. Neural scene graphs for dynamic scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2856–2865, 2021

    work page 2021

  28. [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 international conference on computer vision , pages 5865–5874, 2021

    work page 2021

  29. [29]

    Animatable neural radiance fields for modeling dynamic human bodies

    Sida Peng, Junting Dong, Qianqian Wang, Shangzhan Zhang, Qing Shuai, Xiaowei Zhou, and Hujun Bao. Animatable neural radiance fields for modeling dynamic human bodies. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 14314– 14323, 2021

    work page 2021

  30. [30]

    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. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9054–9063, 2021

    work page 2021

  31. [31]

    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 conference on computer vision and pattern recognition , pages 10318–10327, 2021. 11

    work page 2021

  32. [32]

    Yulia Rubanova, Ricky T. Q. Chen, and David K. Duvenaud. Latent ordinary differential equations for irregularly-sampled time series. In Advances in Neural Information Processing Systems (NeurIPS), volume 32, pages 5321–5331, 2019

    work page 2019

  33. [33]

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

    Ruizhi Shao, Zerong Zheng, Hanzhang Tu, Boning Liu, Hongwen 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, pages 16632–16642, 2023

    work page 2023

  34. [34]

    Nerfplayer: A streamable dynamic scene representation with decomposed neural radiance fields

    Liangchen Song, Anpei Chen, Zhong Li, Zhang Chen, Lele Chen, Junsong Yuan, Yi Xu, and Andreas Geiger. Nerfplayer: A streamable dynamic scene representation with decomposed neural radiance fields. IEEE Transactions on Visualization and Computer Graphics, 29(5):2732– 2742, 2023

    work page 2023

  35. [35]

    Non-rigid neural radiance fields: Reconstruction and novel view synthesis of a dynamic scene from monocular video

    Edgar Tretschk, Ayush Tewari, Vladislav Golyanik, Michael Zollhöfer, Christoph Lassner, and Christian Theobalt. Non-rigid neural radiance fields: Reconstruction and novel view synthesis of a dynamic scene from monocular video. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12959–12970, 2021

    work page 2021

  36. [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 , pages 20310–20320, 2024

    work page 2024

  37. [37]

    Nguyen, Andrea L

    Hedi Xia, Vai Suliafu, Hangjie Ji, Tan M. Nguyen, Andrea L. Bertozzi, Stanley J. Osher, and Bao Wang. Heavy ball neural ordinary differential equations. InAdvances in Neural Information Processing Systems (NeurIPS), volume 34, pages 11437–11449, 2021

    work page 2021

  38. [38]

    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 Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 20331–20341, 2024

    work page 2024

  39. [39]

    V olume rendering of neural implicit surfaces

    Lior Yariv, Jiatao Gu, Yoni Kasten, and Yaron Lipman. V olume rendering of neural implicit surfaces. Advances in Neural Information Processing Systems , 34:4805–4815, 2021

    work page 2021

  40. [40]

    ODE2V AE: Deep generative second order ODEs with bayesian neural networks

    Caglar Yildiz, Markus Heinonen, and Harri Lähdesmäki. ODE2V AE: Deep generative second order ODEs with bayesian neural networks. In Advances in Neural Information Processing Systems (NeurIPS), volume 32, pages 10280–10290, 2019

    work page 2019

  41. [41]

    Gaussianprediction: 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. Gaussianprediction: Dynamic 3d gaussian prediction for motion extrapolation and free view synthesis. In ACM SIGGRAPH 2024 Conference Papers, pages 1–12, 2024. 12

    work page 2024