Pith. sign in

REVIEW 4 major objections 5 minor 58 references

Learning an Implicit Physics Model for Image-based Fluid Simulation

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A physics-informed neural network predicts 3D velocity fields from a single fluid image, and animating feature-based 3D Gaussians with those velocities produces more realistic videos than prior methods, from both input and novel views.

desk verdict A solid single-image fluid animation system with real gains, but the 'physics-bounded' claim is considerably weaker than advertised. read the letter →

arxiv 2508.08254 v1 pith:S43IVKUJ submitted 2025-08-11 cs.CV

Emily Yue-Ting Jia , Jiageng Mao , Zhiyuan Gao , Yajie Zhao , Yue Wang This is my paper · ORCID
classification cs.CV
keywords single-imageanimationfluiddynamicsphysics-informedneuralnetworkNavier-Stokesequations3DGaussiansplattingvelocityfieldpredictionnovelviewsynthesisimageediting
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

The paper tries to establish that a single photograph of natural water can be turned into a short 3D video by learning a per-point velocity field, provided the learning is guided by the equations of fluid motion. Its move is to supervise a neural network with both optical-flow-derived scene flows and a loss derived from the Navier-Stokes equations, simplified by dropping pressure and viscosity so that the material derivative of velocity equals a single learned external force. Appearance is carried by feature-based 3D Gaussians—splatting kernels lifted from the image and its estimated depth—and the predicted velocities simply displace those kernels before rendering from any camera pose. On the validation split of an existing natural-fluid dataset, the paper reports a PSNR of 24.98 versus 22.81 for the leading prior method on input views, and about 22 percent lower velocity error on a synthetic river scene with a rock obstruction. If correct, the work shows that even a simplified physics equation can act as a practical regularizer that keeps learned fluid motion plausible and aware of boundaries.

What carries the argument

The load-bearing object is the conditional physics-informed neural network, an image-conditioned MLP that maps 4D coordinates $(x,y,z,t)$ to a 3D velocity and is trained with both scene-flow supervision and a physics loss. The physics loss is the squared residual of the simplified Navier-Stokes equation $D\mathbf{v}/Dt=\mathbf{f}$, where $\mathbf{f}$ is a single global vector predicted by a separate network head; losing the viscosity and pressure terms keeps the optimization tractable but changes what the loss can enforce. A second piece of the machinery is the boundary condition: for points whose ground-truth motion crosses the fluid mask, the method penalizes the velocity norm, a no-through condition that makes learned flow separate around rocks and banks. The animation side is feature-based 3D Gaussian splatting, where Gaussian kernels lifted from layered depth images are displaced by the predicted velocity and rendered from novel poses, giving a representation whose natural blending covers holes that point-cloud warping would leave.

What would settle it

Compute, from real video of a natural flow, the magnitudes of the pressure-gradient and viscosity terms that the method omits and compare them with the material derivative $D\mathbf{v}/Dt$; in any scene where those omitted terms are comparable in size, the simplified physics loss should fail to track actual motions, and the claim that it enforces realistic physics would be undermined.

Watch

Extended reading notes

Core claim

On its own terms, the central claim is that animating a still fluid image can be treated as a velocity prediction problem, and that the velocity predictor needs physics supervision, not just data supervision. The paper shows a physics-informed neural network that takes an RGB image, a depth map, and a user fluid mask, and outputs a 3D velocity for every surface point, together with a single external force vector for the whole image. The physics loss asks that the predicted velocities satisfy the simplified Navier-Stokes equation $D\mathbf{v}/Dt=\mathbf{f}$, and a boundary penalty asks that no velocity points out of the fluid region, so fluids cannot penetrate added obstacles. The authors report that this physical guidance improves generated videos on input and novel views, improves velocity accuracy on synthetic scenes with known ground truth, and makes edited scenes, such as a rock added to a river, flow around the new boundary.

Load-bearing premise

The method assumes that a heavily simplified version of the Navier-Stokes equations—with pressure and viscosity removed and all external forces collapsed into one learned vector per image—still captures enough physics that enforcing it makes predicted water motion realistic.

Editorial extensions

If this is right

  • Animations from a single fluid image are quantitatively better than leading prior methods: PSNR rises from 22.81 to 24.98 on input views and from 22.46 to 24.34 on novel views, with similar gains in SSIM, LPIPS, and VMAF.
  • Predicted velocity fields are more accurate: on a synthetic river scene with a rock, the L1 velocity error drops by roughly 22 percent relative to the prior method.
  • Boundary editing becomes meaningful: adding a rock to the input image makes the predicted streamlines split around it instead of passing through, because of the no-through penalty.
  • The pipeline needs only an image, a depth estimate, and a user fluid mask, and generates a 60-frame 720p animation in about 13 seconds on an H100 GPU.

Reading between the lines

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

  • An untested but natural extension is to make the external force a spatial field rather than a single vector per image; the paper's own acknowledged failure mode, merging rivers, is exactly where a constant force cannot capture two competing pressure-driven flows.
  • The no-through boundary penalty suggests a cheap way to give any image-animation pipeline obstacle awareness: penalize predicted motion that exits a user-supplied region mask, with no 3D model of the obstacle required.
  • If the simplified physics loss is truly the cause of the accuracy gain, the method's advantage over data-only predictors should be largest when water interacts with boundaries and smallest for free-flowing water with no obstacles; this is a directly testable prediction not made in the paper.
  • Because the force head is global per image, the method is tied to quasi-steady flows; scenes where the driving force changes over time, such as a pulse of water, would require a time- or location-dependent force.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. This paper presents a method for generating 4D scenes with fluid animation from a single natural image. The proposed pipeline has two main components: a physics-informed neural dynamics model that predicts a 3D velocity field on fluid surface points, trained with a combination of scene-flow supervision and a loss derived from a simplified Navier-Stokes equation; and an animation module that reconstructs feature-based 3D Gaussians from the input plus depth and renders animated frames from arbitrary camera trajectories. Experiments on the Holynski et al. validation set report consistent improvements over 3D-Cinemagraphy and Make-it-4D on image metrics, a user study, and a synthetic velocity benchmark with a rock obstruction.

Significance. The empirical results are solid: the method improves PSNR by about 2 dB over 3D-Cinemagraphy on input views and by about 1.9 dB on novel views, the user study strongly prefers the proposed outputs, and the synthetic velocity benchmark is an appropriate external test. The use of 3D Gaussians appears to provide real benefits for hole filling and novel-view rendering, and the paper is generally clear and well organized. If the physics-informed component were genuinely enforcing fluid momentum, this would be a significant contribution to single-image fluid animation. However, as presented, the physics loss is too weak to support the 'physics-grounded' claim, because the learned global external force makes Eq. (6) a spatial-variance regularizer rather than a physical constraint, so the main novelty is not currently demonstrated even though the empirical system works well.

major comments (4)
  1. [Sec. 3.3.1, Eqs. (5)-(6); Sec. 3.3.2] The physics loss in Eq. (6) uses a single learned global external force vector f per image. After optimizing f, minimizing L_physics is equivalent to minimizing the spatial variance of the material derivative Dv/Dt over sampled fluid points, so any velocity field with nearly uniform material derivative can achieve a zero physics loss irrespective of physical correctness. The loss therefore acts as a smoothness regularizer, not as an enforcement of the momentum equation. This is underscored by Sec. 11, which acknowledges that the lack of pressure fields limits the method on interactions such as river merging. Since the paper's central claim is that the predicted dynamics are 'physics-grounded' (Abstract, Sec. 1), this is a load-bearing issue. Please either modify the loss so that the external force is not a free per-image parameter that can absorb the residual (e.g., fix f to a physical prior such as gravity, or predict a spatially varying force field with a physical prior), or soften the claims and provide evidence that the variance regularizer is the cause of the observed improvements.
  2. [Sec. 3.3.1, Eq. (4)] The incompressibility constraint in Eq. (4) is stated as an assumption but is never incorporated into any loss term. Because the simplified momentum loss only penalizes spatial variance of the material derivative, it does not constrain the divergence of the predicted velocity field. The paper should either add a divergence-free penalty (readily computable with autodiff) or demonstrate on the synthetic scenes that the predicted velocities are approximately divergence-free; without this, the 'incompressible' claim in Sec. 3.3.1 is unsupported.
  3. [Sec. 4.8, Table 6] The ablation in Table 6 reports that adding viscosity or pressure fields increases L1 error, but attributes this to optimization difficulty. This does not validate the physical simplification of Eq. (5); it only shows the relative performance of training procedures. To substantiate the simplification, the paper should separate optimization difficulty from model fidelity, for example by reporting training convergence curves or loss values for the variants, or by evaluating on a case where pressure is important, such as the river-merging case mentioned in Sec. 11. As it stands, the ablation supports the design choice pragmatically but not the physical correctness claim.
  4. [Sec. 4.7, Table 5] The synthetic scene evaluation lacks essential details: the paper does not state how the ground-truth velocity fields for the two scenes are generated (e.g., by a Navier-Stokes solver, and with which parameters), nor does it report the scene resolution or the number of surface points sampled. Without this information, the reported L1 errors cannot be independently reproduced or compared across methods.
minor comments (5)
  1. [Sec. 3.5.1, Eq. (13)] The scene-flow loss is denoted L_sceneflow, but Eq. (12) defines it as L_flow; please unify the notation.
  2. [Sec. 4.8, Table 4] The configurations in the ablation are indicated by checkmarks, but the row labels are not explicitly stated; please spell out which components are enabled in each row.
  3. [Sec. 3.3.1, Eq. (3)] The phrase 'incompression constraint' appears to be a typo for 'incompressibility constraint.'
  4. [Sec. 4.2 and Sec. 7] The training-iteration statement differs: Sec. 4.2 says 50 epochs (250k iterations) for all methods, while Sec. 7 states 120k iterations for the physics-informed neural dynamics and 250k for the animation module. Please clarify the training schedule.
  5. [Sec. 4.7, Table 5] The L1 velocity errors are reported without units or confidence intervals; please specify the normalization and report variance across the synthetic scenes.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: velocity prediction is anchored by external scene-flow supervision and benchmarked against held-out validation and synthetic ground truth.

full rationale

The paper's central derivation chain is not circular. The predicted velocity field (Sec. 3.3) is primarily supervised by dense 3D scene flows lifted from RAFT optical flow and monocular depth (Eq. 12), an external signal independent of the physics loss. The physics loss (Eq. 6) is a regularizer, and although its learned per-image force f makes it weak (for fixed v the optimal f is the spatial mean of the material derivative, so Eq. 6 essentially penalizes only spatial variance of Dv/Dt), this is a model-capacity and physical-validity concern rather than a circular reduction: f is not an output whose value is later reported as a prediction, and the velocity itself is not defined in terms of the physics loss. The simplifications dropping pressure and viscosity are imported from external prior work [5, 48] and are tested against alternatives in Table 6; they are not justified by a self-citation chain. Evaluation is self-contained against external benchmarks: held-out Holynski validation videos and synthetic water-flow scenes with ground-truth velocities (Table 5). No load-bearing self-citation, uniqueness import, or renaming of a known result was found. Hence score 0.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The physics-consistency claim sits on several stated but unverified modeling choices: the simplified Navier-Stokes equation, a learned global external force, and the lifting of 2D optical flow to 3D scene flow. These are not invented entities, but they are free parameters or domain assumptions that the reader must accept. The two loss weights are reported nowhere, making the exact balance between data and physics unknown.

free parameters (3)
  • Physics loss weight lambda_1 = not reported
    Scales the simplified Navier-Stokes residual loss in Eq. (10) or Eq. (13); the paper does not state its value, though it directly affects the balance between data fidelity and physics adherence.
  • Boundary loss weight lambda_2 = not reported
    Scales the no-through boundary penalty in Eq. (10); its value is unspecified, affecting the strength of the collision avoidance.
  • External force f = learned 3D vector per image
    A network head outputs a single global external force vector used in the simplified physics loss; it is fitted to minimize the residual and acts as a per-image free parameter.
assumptions (5)
  • domain assumption Natural fluid motion is governed by the incompressible Navier-Stokes equations (Eq. 3-4).
    The method assumes incompressible fluids; Eq. (4) is stated without verification for the images in the dataset.
  • ad hoc to paper Pressure and viscosity terms can be dropped, and the external force can absorb their effects (Eq. 5).
    This simplification is adopted from [5] to avoid extra network heads; it is load-bearing for the physics loss and is not independently validated for natural fluid imagery.
  • domain assumption RAFT optical flow lifted by monocular depth provides valid 3D scene-flow supervision.
    In Sec. 3.5.1, dense 2D optical flow from [41] and monocular depth are lifted to 3D scene flows used as ground truth for velocity; errors in either estimator are inherited by the model.
  • ad hoc to paper A single global external force vector per image captures the dominant forces (e.g., gravity).
    Sec. 3.3.2 predicts f in R^3; spatially varying forces are not modeled, which limits the physical fidelity.
  • domain assumption Boundaries are impermeable and the simplified boundary penalty approximates the no-through condition (Eq. 8-9).
    The boundary penalty omits normal estimation and uses a binary in/out fluid indicator; this is an approximation of the true Dirichlet condition.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning an Implicit Physics Model for Image-based Fluid Simulation." pith.science (2026). https://pith.science/paper/S43IVKUJ

@misc{pith2026250808254,
  author       = {Pith},
  title        = {Pith review of: Learning an Implicit Physics Model for Image-based Fluid Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S43IVKUJ}},
  note         = {Machine review of arXiv:2508.08254}
}
read the original abstract

Humans possess an exceptional ability to imagine 4D scenes, encompassing both motion and 3D geometry, from a single still image. This ability is rooted in our accumulated observations of similar scenes and an intuitive understanding of physics. In this paper, we aim to replicate this capacity in neural networks, specifically focusing on natural fluid imagery. Existing methods for this task typically employ simplistic 2D motion estimators to animate the image, leading to motion predictions that often defy physical principles, resulting in unrealistic animations. Our approach introduces a novel method for generating 4D scenes with physics-consistent animation from a single image. We propose the use of a physics-informed neural network that predicts motion for each surface point, guided by a loss term derived from fundamental physical principles, including the Navier-Stokes equations. To capture appearance, we predict feature-based 3D Gaussians from the input image and its estimated depth, which are then animated using the predicted motions and rendered from any desired camera perspective. Experimental results highlight the effectiveness of our method in producing physically plausible animations, showcasing significant performance improvements over existing methods. Our project page is https://physfluid.github.io/ .

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 36 canonical work pages

  1. [5]

    Physics informed neural fields for smoke reconstruction with sparse data

    Mengyu Chu, Lingjie Liu, Quan Zheng, Erik Franz, Hans- Peter Seidel, Christian Theobalt, and Rhaleb Zayer. Physics informed neural fields for smoke reconstruction with sparse data. ACM Transactions on Graphics (ToG) , 41(4):1–14,

    work page

  2. [1]

    Videophy: Evaluating physical commonsense for video generation

    Hritik Bansal, Zongyu Lin, Tianyi Xie, Zeshun Zong, Michal Yarom, Yonatan Bitton, Chenfanfu Jiang, Yizhou Sun, Kai- Wei Chang, and Aditya Grover. Videophy: Evaluating physical commonsense for video generation. arXiv preprint arXiv:2406.03520, 2024. 3

    arXiv 2024

  3. [2]

    Stable video diffusion: Scaling latent video diffusion models to large datasets

    Andreas Blattmann, Tim Dockhorn, Sumith Kulal, Daniel Mendelevitch, Maciej Kilian, Dominik Lorenz, Yam Levi, Zion English, Vikram V oleti, Adam Letts, et al. Stable video diffusion: Scaling latent video diffusion models to large datasets. arXiv preprint arXiv:2311.15127, 2023. 2

    arXiv 2023

  4. [3]

    Diffdreamer: Towards consistent unsupervised single-view scene extrapolation with conditional diffusion models

    Shengqu Cai, Eric Ryan Chan, Songyou Peng, Mohamad Shahbazi, Anton Obukhov, Luc Van Gool, and Gordon Wetzstein. Diffdreamer: Towards consistent unsupervised single-view scene extrapolation with conditional diffusion models. In ICCV, 2023. 2

    work page 2023

  5. [4]

    Everybody dance now

    Caroline Chan, Shiry Ginosar, Tinghui Zhou, and Alexei A Efros. Everybody dance now. In Proceedings of the IEEE/CVF international conference on computer vision , pages 5933–5942, 2019. 3

    work page 2019

  6. [6]

    Salesin, and Richard Szeliski

    Yung-Yu Chuang, Dan B Goldman, Ke Colin Zheng, Brian Curless, David H. Salesin, and Richard Szeliski. Animat- ing pictures with stochastic motion textures. ACM Trans. Graph., 24(3):853–860, 2005. 3

    work page 2005

  7. [7]

    Fluid simulation on neural flow maps

    Yitong Deng, Hong-Xing Yu, Diyang Zhang, Jiajun Wu, and Bo Zhu. Fluid simulation on neural flow maps. ACM Trans- actions on Graphics (TOG), 42(6):1–21, 2023. 3

    work page 2023

  8. [8]

    Flownet: Learning optical flow with convolutional networks

    Alexey Dosovitskiy, Philipp Fischer, Eddy Ilg, Philip Hausser, Caner Hazirbas, Vladimir Golkov, Patrick Van Der Smagt, Daniel Cremers, and Thomas Brox. Flownet: Learning optical flow with convolutional networks. In Pro- ceedings of the IEEE international conference on computer vision, pages 2758–2766, 2015. 3

    2015

Show all 58 references
  1. [9]

    An- imating landscape: self-supervised learning of decoupled motion and appearance for single-image video synthesis

    Yuki Endo, Yoshihiro Kanamori, and Shigeru Kuriyama. An- imating landscape: self-supervised learning of decoupled motion and appearance for single-image video synthesis. arXiv preprint arXiv:1910.07192, 2019. 3

    1910 arXiv

  2. [10]

    Simulating fluids in real-world still images

    Siming Fan, Jingtan Piao, Chen Qian, Hongsheng Li, and Kwan-Yee Lin. Simulating fluids in real-world still images. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 15922–15931, 2023. 3, 5, 7

    2023

  3. [11]

    Global transport for fluid reconstruction with learned self- supervision

    Erik Franz, Barbara Solenthaler, and Nils Thuerey. Global transport for fluid reconstruction with learned self- supervision. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 1632– 1642, 2021. 3

    2021

  4. [12]

    Learning to estimate single-view volumetric flow motions without 3d supervision

    Erik Franz, Barbara Solenthaler, and Nils Thuerey. Learning to estimate single-view volumetric flow motions without 3d supervision. arXiv preprint arXiv:2302.14470, 2023. 3

    2023 arXiv

  5. [13]

    Srinivasan, Jonathan T

    Ruiqi Gao*, Aleksander Holynski*, Philipp Henzler, Arthur Brussee, Ricardo Martin-Brualla, Pratul P. Srinivasan, Jonathan T. Barron, and Ben Poole*. Cat3d: Create any- thing in 3d with multi-view diffusion models. Advances in Neural Information Processing Systems, 2024. 2

    2024

  6. [14]

    Modeling the dy- namics of pde systems with physics-constrained deep auto- regressive networks

    Nicholas Geneva and Nicholas Zabaras. Modeling the dy- namics of pde systems with physics-constrained deep auto- regressive networks. Journal of Computational Physics, 403: 109056, 2020. 3

    2020

  7. [15]

    Neuroanimator: Fast neural network emulation and con- trol of physics-based models

    Radek Grzeszczuk, Demetri Terzopoulos, and Geoffrey Hin- ton. Neuroanimator: Fast neural network emulation and con- trol of physics-based models. In Proceedings of the 25th an- nual conference on Computer graphics and interactive tech- niques, pages 9–20, 1998. 3

    1998

  8. [16]

    Neurofluid: Fluid dynamics grounding with particle- driven neural radiance fields

    Shanyan Guan, Huayu Deng, Yunbo Wang, and Xiaokang Yang. Neurofluid: Fluid dynamics grounding with particle- driven neural radiance fields. In International Conference on Machine Learning, pages 7919–7929. PMLR, 2022. 3

    2022

  9. [17]

    Imagen video: High definition video generation with diffusion mod- els

    Jonathan Ho, William Chan, Chitwan Saharia, Jay Whang, Ruiqi Gao, Alexey Gritsenko, Diederik P Kingma, Ben Poole, Mohammad Norouzi, David J Fleet, et al. Imagen video: High definition video generation with diffusion mod- els. arXiv preprint arXiv:2210.02303, 2022. 2

    arXiv

  10. [18]

    Video dif- fusion models

    Jonathan Ho, Tim Salimans, Alexey Gritsenko, William Chan, Mohammad Norouzi, and David J Fleet. Video dif- fusion models. Advances in Neural Information Processing Systems, 35:8633–8646, 2022. 2

    2022

  11. [19]

    Animating pictures with eulerian mo- tion fields

    Aleksander Holynski, Brian L Curless, Steven M Seitz, and Richard Szeliski. Animating pictures with eulerian mo- tion fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 5810– 5819, 2021. 2, 3, 5, 6, 7, 8

    2021

  12. [20]

    Cogvideo: Large-scale pretraining for text-to-video generation via transformers

    Wenyi Hong, Ming Ding, Wendi Zheng, Xinghan Liu, and Jie Tang. Cogvideo: Large-scale pretraining for text-to-video generation via transformers. arXiv preprint arXiv:2205.15868, 2022. 2

    2022 arXiv

  13. [21]

    Animating still land- scape photographs through cloud motion creation

    Wei-Cih Jhou and Wen-Huang Cheng. Animating still land- scape photographs through cloud motion creation. IEEE Transactions on Multimedia, 18(1):4–13, 2015. 3

    2015

  14. [22]

    3d gaussian splatting for real-time radiance field rendering

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

  15. [23]

    3d cinemagraphy from a sin- gle image

    Xingyi Li, Zhiguo Cao, Huiqiang Sun, Jianming Zhang, Ke Xian, and Guosheng Lin. 3d cinemagraphy from a sin- gle image. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 4595– 4605, 2023. 2, 3, 5, 6, 7, 8, 1

    2023

  16. [24]

    Graph neural network- accelerated lagrangian fluid simulation

    Zijie Li and Amir Barati Farimani. Graph neural network- accelerated lagrangian fluid simulation. Computers & Graphics, 103:201–211, 2022. 3

    2022

  17. [25]

    Wonderland: Nav- igating 3d scenes from a single image

    Hanwen Liang, Junli Cao, Vidit Goel, Guocheng Qian, Sergei Korolev, Demetri Terzopoulos, Konstantinos N Pla- taniotis, Sergey Tulyakov, and Jian Ren. Wonderland: Nav- igating 3d scenes from a single image. arXiv preprint arXiv:2412.12091, 2024. 2

    2024 arXiv

  18. [26]

    Physgen: Rigid-body physics-grounded image- to-video generation

    Shaowei Liu, Zhongzheng Ren, Saurabh Gupta, and Shen- long Wang. Physgen: Rigid-body physics-grounded image- to-video generation. In European Conference on Computer Vision, pages 360–378. Springer, 2024. 2

    2024

  19. [27]

    Liquid warping gan: A unified framework for human motion imitation, appearance transfer and novel view synthesis

    Wen Liu, Zhixin Piao, Jie Min, Wenhan Luo, Lin Ma, and Shenghua Gao. Liquid warping gan: A unified framework for human motion imitation, appearance transfer and novel view synthesis. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 5904–5913,

  20. [28]

    Controllable animation of fluid elements in still images

    Aniruddha Mahapatra and Kuldeep Kulkarni. Controllable animation of fluid elements in still images. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3667–3676, 2022. 3

    2022

  21. [29]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    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

    2021

  22. [30]

    Semantic image synthesis with spatially-adaptive nor- malization

    Taesung Park, Ming-Yu Liu, Ting-Chun Wang, and Jun-Yan Zhu. Semantic image synthesis with spatially-adaptive nor- malization. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2337–2346,

  23. [31]

    Deepmimic: Example-guided deep reinforce- ment learning of physics-based character skills

    Xue Bin Peng, Pieter Abbeel, Sergey Levine, and Michiel Van de Panne. Deepmimic: Example-guided deep reinforce- ment learning of physics-based character skills. ACM Trans- actions On Graphics (TOG), 37(4):1–14, 2018. 3

    2018

  24. [32]

    Raissi, P

    M. Raissi, P. Perdikaris, and G.E. Karniadakis. Physics- informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019. 3

    2019

  25. [33]

    Vi- sion transformers for dense prediction

    Ren ´e Ranftl, Alexey Bochkovskiy, and Vladlen Koltun. Vi- sion transformers for dense prediction. In Proceedings of the IEEE/CVF international conference on computer vision, pages 12179–12188, 2021. 1

    2021

  26. [34]

    Deep image spatial transformation for person image generation

    Yurui Ren, Xiaoming Yu, Junming Chen, Thomas H Li, and Ge Li. Deep image spatial transformation for person image generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7690– 7699, 2020. 3

    2020

  27. [35]

    Make-it-4d: Synthesizing a consistent long-term dynamic scene video from a single image

    Liao Shen, Xingyi Li, Huiqiang Sun, Juewen Peng, Ke Xian, Zhiguo Cao, and Guosheng Lin. Make-it-4d: Synthesizing a consistent long-term dynamic scene video from a single image. In Proceedings of the 31st ACM International Con- ference on Multimedia, pages 8167–8175, 2023. 3, 6, 7

    2023

  28. [36]

    3d photography using context-aware layered depth inpainting

    Meng-Li Shih, Shih-Yang Su, Johannes Kopf, and Jia-Bin Huang. 3d photography using context-aware layered depth inpainting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 6, 1

    2020

  29. [37]

    First order motion model for image animation

    Aliaksandr Siarohin, St ´ephane Lathuili`ere, Sergey Tulyakov, Elisa Ricci, and Nicu Sebe. First order motion model for image animation. Advances in neural information processing systems, 32, 2019. 3

    2019

  30. [38]

    Very deep con- volutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep con- volutional networks for large-scale image recognition. In 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Confer- ence Track Proceedings, 2015. 5

    2015

  31. [39]

    Make-a-video: Text-to-video generation without text-video data

    Uriel Singer, Adam Polyak, Thomas Hayes, Xi Yin, Jie An, Songyang Zhang, Qiyuan Hu, Harry Yang, Oron Ashual, Oran Gafni, et al. Make-a-video: Text-to-video generation without text-video data. arXiv preprint arXiv:2209.14792 ,

    arXiv

  32. [40]

    Physmotion: Physics- grounded dynamics from a single image

    Xiyang Tan, Ying Jiang, Xuan Li, Zeshun Zong, Tianyi Xie, Yin Yang, and Chenfanfu Jiang. Physmotion: Physics- grounded dynamics from a single image. arXiv preprint arXiv:2411.17189, 2024. 2

    2024 arXiv

  33. [41]

    Raft: Recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. Raft: Recurrent all-pairs field transforms for optical flow. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part II 16, pages 402–419. Springer,

    2020

  34. [42]

    Accelerating eulerian fluid simulation with convolutional networks

    Jonathan Tompson, Kristofer Schlachter, Pablo Sprechmann, and Ken Perlin. Accelerating eulerian fluid simulation with convolutional networks. In International conference on ma- chine learning, pages 3424–3433. PMLR, 2017. 3

    2017

  35. [43]

    Learning incompressible fluid dynamics from scratch– towards fast, differentiable fluid models that generalize

    Nils Wandel, Michael Weinmann, and Reinhard Klein. Learning incompressible fluid dynamics from scratch– towards fast, differentiable fluid models that generalize. arXiv preprint arXiv:2006.08762, 2020

    2006 arXiv

  36. [44]

    Teaching the incompressible navier–stokes equations to fast neural surrogate models in three dimensions

    Nils Wandel, Michael Weinmann, and Reinhard Klein. Teaching the incompressible navier–stokes equations to fast neural surrogate models in three dimensions. Physics of Flu- ids, 33(4), 2021. 3

    2021

  37. [45]

    Sv4d: Dynamic 3d content generation with multi-frame and multi-view consistency

    Yiming Xie, Chun-Han Yao, Vikram V oleti, Huaizu Jiang, and Varun Jampani. Sv4d: Dynamic 3d content generation with multi-frame and multi-view consistency. arXiv preprint arXiv:2407.17470, 2024. 2

    2024 arXiv

  38. [46]

    Mo- tion dreamer: Realizing physically coherent video genera- tion through scene-aware motion reasoning

    Tianshuo Xu, Zhifei Chen, Leyi Wu, Hao Lu, Yuying Chen, Lihui Jiang, Bingbing Liu, and Yingcong Chen. Mo- tion dreamer: Realizing physically coherent video genera- tion through scene-aware motion reasoning. arXiv preprint arXiv:2412.00547, 2024. 2

    2024

  39. [47]

    Cogvideox: Text-to-video diffusion models with an expert transformer

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiao- han Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072, 2024. 2

    2024 arXiv

  40. [48]

    Inferring hybrid neural fluid fields from videos

    Hong-Xing Yu, Yang Zheng, Yuan Gao, Yitong Deng, Bo Zhu, and Jiajun Wu. Inferring hybrid neural fluid fields from videos. In NeurIPS, 2023. 3, 4

    2023

  41. [49]

    Freeman, Forrester Cole, Deqing Sun, Noah Snavely, Jiajun Wu, and Charles Her- rmann

    Hong-Xing Yu, Haoyi Duan, Junhwa Hur, Kyle Sargent, Michael Rubinstein, William T. Freeman, Forrester Cole, Deqing Sun, Noah Snavely, Jiajun Wu, and Charles Her- rmann. Wonderjourney: Going from anywhere to every- where. In CVPR, 2024. 2

    2024

  42. [50]

    Viewcrafter: Taming video diffusion models for high-fidelity novel view synthesis.arXiv preprint arXiv:2409.02048, 2024

    Wangbo Yu, Jinbo Xing, Li Yuan, Wenbo Hu, Xiaoyu Li, Zhipeng Huang, Xiangjun Gao, Tien-Tsin Wong, Ying Shan, and Yonghong Tian. Viewcrafter: Taming video diffusion models for high-fidelity novel view synthesis.arXiv preprint arXiv:2409.02048, 2024. 2

    2024 arXiv

  43. [51]

    Tomofluid: Reconstructing dynamic fluid from sparse view videos

    Guangming Zang, Ramzi Idoughi, Congli Wang, Anthony Bennett, Jianguo Du, Scott Skeen, William L Roberts, Peter Wonka, and Wolfgang Heidrich. Tomofluid: Reconstructing dynamic fluid from sparse view videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Patter...

    2020

  44. [52]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 2018. 5, 6

    2018

  45. [53]

    Physdreamer: Physics-based interac- tion with 3d objects via video generation

    Tianyuan Zhang, Hong-Xing Yu, Rundi Wu, Brandon Y Feng, Changxi Zheng, Noah Snavely, Jiajun Wu, and William T Freeman. Physdreamer: Physics-based interac- tion with 3d objects via video generation. In European Con- ference on Computer Vision, pages 388–406. Springer, 2024. 2 L...

    2024

  46. [54]

    Both stages use the Adam optimizer with a 1e-4 learning rate and (0, 0.9) for betas

    Implementation Details Our training process consists of two stages: first, the physics-informed neural dynamics model is trained for 120k iterations, followed by the animation module for 250k iterations. Both stages use the Adam optimizer with a 1e-4 learning rate and (0, 0.9)...

  47. [55]

    User Hints Our network can be extended to incorporate sparse flow hints from users, as shown in Figure 9. These hints pro- vide valuable information, such as the general direction of fluid flow, and help to reduce ambiguity in velocity predic- tion (e.g., a river could flow bo...

  48. [56]

    The animation part contains an inpainter to inpaint each LDI, an encoder to extract features for each LDI, and a decoder to decode the feature maps rendered by 3D Gaussians

    Model Architecture Our model consists of two parts: physics-informed neu- ral dynamics and the animation part. The animation part contains an inpainter to inpaint each LDI, an encoder to extract features for each LDI, and a decoder to decode the feature maps rendered by 3D Gau...

  49. [57]

    Video Results We include video results generated by our methods and baseline methods in the supplementary materials

    Additional Results 10.1. Video Results We include video results generated by our methods and baseline methods in the supplementary materials. These results comprise: (a) video outputs from the Holynski et al. [19] validation set, showcasing both original and novel viewpoints, ...

  50. [58]

    In particular, the lack of pressure fields limits its ability to capture interactions such as river merging

    Limitations Our method focuses on natural fluids and may be less ef- fective in some scenarios. In particular, the lack of pressure fields limits its ability to capture interactions such as river merging. Static fluid masks also pose challenges for scenes with evolving boundar...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.