pith. sign in

arxiv: 2607.01164 · v1 · pith:PMY4WSAVnew · submitted 2026-07-01 · 💻 cs.LG

Efficient Compression of Structured and Unstructured Volumes via Learned 3D Gaussian Representation

Landon Dyken , Sharmistha Chakrabarti , Nathan Debardeleben , Steve Petruzza , Qi Wu , Will Usher , Sidharth Kumar This is my paper

Pith reviewed 2026-07-02 15:21 UTC · model grok-4.3

classification 💻 cs.LG
keywords volume compression3D Gaussian representationexplicit modelsstructured volumesunstructured volumesneural compressionscalar field representation
0
0 comments X

The pith

Collections of 3D Gaussians compress both structured and unstructured volume data by directly encoding scalar fields and domain geometry without separate mesh storage.

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

The paper develops an explicit representation for volume compression that replaces implicit neural networks with collections of 3D Gaussian primitives. Scalar values are recovered at query points by weighted aggregation of all Gaussians that intersect the location, and loss terms plus a sampling-error densification step encourage the primitives to capture both the field values and the domain boundaries. Because the Gaussians themselves delineate the geometry, unstructured volumes no longer require any auxiliary mesh storage. The resulting models train faster than implicit alternatives on structured data and deliver higher fidelity across all reported metrics on unstructured data.

Core claim

An explicit model based on 3D Gaussian primitives can serve as a complete representation for volume scalar fields. The primitives are trained so that scalar values are reconstructed through weighted aggregation of intersecting Gaussians, and the same primitives automatically encode domain geometry. This formulation eliminates the partial-mesh requirement that limits compression of unstructured volumes under implicit neural representations, while optimized CUDA sampling pipelines and sampling-error densification produce competitive quality with faster training on structured volumes and superior results on unstructured volumes.

What carries the argument

3D Gaussian primitives reinterpreted as an explicit scalar-field representation, reconstructed via weighted aggregation of intersecting primitives at query locations.

If this is right

  • Unstructured volumes can be stored and queried without any separate mesh data structure.
  • Training completes faster than implicit neural representations on structured volumes while matching reconstruction quality.
  • All standard quality metrics improve on unstructured volumes relative to implicit baselines.
  • Higher overall compression ratios become possible because geometry is carried inside the same primitives that store the field.
  • CUDA-accelerated sampling pipelines allow direct querying of the compressed representation.

Where Pith is reading between the lines

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

  • The same explicit primitives could support real-time volume rendering or direct isosurface extraction without an intermediate decompression step.
  • Extending the densification criterion to time-varying data might allow compression of dynamic volumes with the same mesh-free property.
  • Because geometry and field live in one data structure, the approach could simplify pipelines that combine compression with spatial queries or level-of-detail selection.

Load-bearing premise

Collections of 3D Gaussians trained via weighted aggregation can accurately encode both the scalar field values and the domain geometry for unstructured volumes without requiring any additional mesh storage or post-processing.

What would settle it

A head-to-head storage and quality comparison on multiple unstructured volumes in which the Gaussian model requires more total memory than an implicit network plus its partial mesh while also showing lower PSNR or higher error.

Figures

Figures reproduced from arXiv: 2607.01164 by Landon Dyken, Nathan Debardeleben, Qi Wu, Sharmistha Chakrabarti, Sidharth Kumar, Steve Petruzza, Will Usher.

Figure 1
Figure 1. Figure 1: Volume rendering of the 17.9 GB unstructured Impact dataset with a custom transfer function using the ground truth and our [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visualization of sampling for structured (left) and unstructured [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: An illustration of our sampling algorithm on a [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of how we define false negative (A) and false positive [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: An illustration of our sampling algorithm on a 16 cell unstructured dataset, where a sample is taken at each cell center. (A) We first create a [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Renderings of ground truth, 1024× compressed W-VEG models, loss, and W-VEG density for Miranda and Richtmyer datasets. While loss and W-VEG density are spatially uniform for the Miranda model, they are heavily clustered on the mixing surface of the Richtmyer model, due to our method’s explicit nature and adaptive density control during training. 6.1 Structured Volume Evaluation Experimental Setup We first … view at source ↗
Figure 9
Figure 9. Figure 9: Neural volume rendering using samples taken from UGINR, W-VEG, and W-VEG+S models at [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
read the original abstract

Recent work has shown that implicit neural representations (INRs) can be trained to effectively compress structured and unstructured volume data, allowing for direct data querying with a reduced memory footprint. However, as existing INRs for unstructured volumes do not encode geometry, they require partial mesh storage for later sampling, limiting achievable compression. At the same time, novel view synthesis methods have shown that explicit collections of 3D Gaussians can be used to accurately visualize volume data. In this work, we introduce an explicit model for volume data compression based on 3D Gaussian primitives. We reinterpret collections of 3D Gaussians as an explicit representation of a scalar field and use a sampling strategy that reconstructs scalar values at spatial locations through weighted aggregation of intersecting Gaussians. We develop optimized CUDA-accelerated pipelines for structured and unstructured model sampling, loss functions that encourage accurate domain encoding by our models, and a novel sampling-error based densification strategy. Our explicit formulation naturally encodes domain geometry, eliminating the need for mesh storage in unstructured volumes and introducing significantly higher compression opportunities. Compared to existing INRs, we demonstrate that our explicit model achieves competitive reconstruction quality with significant training speedups on structured volumes, while markedly outperforming in all metrics on unstructured volumes.

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 / 0 minor

Summary. The paper proposes an explicit representation for compressing structured and unstructured volume data using collections of 3D Gaussian primitives. Scalar values are reconstructed via weighted aggregation of intersecting Gaussians, supported by CUDA-accelerated sampling pipelines, loss functions that encourage domain encoding, and a sampling-error-based densification strategy. The central claims are that this approach achieves competitive reconstruction quality with faster training on structured volumes and markedly superior performance on unstructured volumes, while naturally encoding domain geometry to eliminate the need for separate mesh storage.

Significance. If the performance claims and geometry-encoding property hold under rigorous verification, the method could advance volume compression by offering an explicit, mesh-free alternative to implicit neural representations, with potential gains in training speed and compression ratios for unstructured data.

major comments (2)
  1. Abstract: the claim that the model 'naturally encodes domain geometry, eliminating the need for mesh storage' is load-bearing for the unstructured-volume advantage, yet the provided text gives no concrete description of how weighted aggregation assigns background values to points outside the true domain support or how this is validated on irregular geometries; without this, the reported metric gains versus mesh-dependent INRs cannot be directly compared.
  2. Abstract: the assertion of 'competitive reconstruction quality with significant training speedups' and 'markedly outperforming in all metrics' on unstructured volumes is presented without any experimental setup details, baseline definitions, error bars, or data-exclusion rules, rendering the central performance claims unverifiable from the manuscript text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the two major comments point-by-point below. Where the abstract lacks sufficient detail for verifiability, we will revise it and cross-reference the methods and experiments sections.

read point-by-point responses

  1. Referee: Abstract: the claim that the model 'naturally encodes domain geometry, eliminating the need for mesh storage' is load-bearing for the unstructured-volume advantage, yet the provided text gives no concrete description of how weighted aggregation assigns background values to points outside the true domain support or how this is validated on irregular geometries; without this, the reported metric gains versus mesh-dependent INRs cannot be directly compared.

    Authors: We agree the abstract is too terse on this mechanism. Section 3.2 details the weighted aggregation and the domain-encoding loss terms that drive background values outside the support; Section 4.3 validates this on irregular unstructured geometries by comparing against mesh-dependent baselines. We will expand the abstract with one sentence summarizing the background assignment process and ensure the irregular-geometry validation is explicitly called out, enabling direct metric comparison. revision: yes

  2. Referee: Abstract: the assertion of 'competitive reconstruction quality with significant training speedups' and 'markedly outperforming in all metrics' on unstructured volumes is presented without any experimental setup details, baseline definitions, error bars, or data-exclusion rules, rendering the central performance claims unverifiable from the manuscript text.

    Authors: The abstract is a high-level summary; full experimental protocols (including INR baselines, multiple-run error bars, and data-exclusion criteria) appear in Section 4 and the supplement. We will revise the abstract to include a brief parenthetical reference to these sections and, if space permits, note that all metrics include standard deviations, thereby improving verifiability without altering the summary nature of the abstract. revision: partial

Circularity Check

0 steps flagged

No circularity: formulation introduces independent sampling, loss, and densification mechanisms

full rationale

The paper presents an explicit 3D Gaussian model for volume compression that reinterprets Gaussians as a scalar field via weighted aggregation sampling, augmented by custom CUDA pipelines, domain-encoding loss terms, and sampling-error densification. No equations or performance claims in the provided text reduce reported metrics to quantities defined by the same fitted parameters or prior self-citations; the geometry-encoding claim is tied to the newly proposed losses rather than being presupposed by construction. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that 3D Gaussians can serve as a complete explicit representation of scalar fields; no free parameters are explicitly named in the abstract, but training likely involves hyperparameters for Gaussian count and optimization.

free parameters (1)
  • Gaussian count and initialization density
    The number and placement of Gaussians are optimized during training and act as a tunable capacity parameter.
axioms (1)
  • domain assumption Weighted aggregation of intersecting 3D Gaussians reconstructs accurate scalar values at query points
    Core reinterpretation stated in the abstract as the basis for the explicit model.

pith-pipeline@v0.9.1-grok · 5772 in / 1257 out tokens · 24296 ms · 2026-07-02T15:21:57.230610+00:00 · methodology

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

54 extracted references · 33 canonical work pages

  1. [1]

    https://github.com/NVIDIA/cuBQL

    cuBQL: A CUDA BVH Build and Query Library. https://github.com/NVIDIA/cuBQL. 6

  2. [2]

    K. Ai, K. Tang, and C. Wang. NLI4V olVis: Natural language interaction for vol- ume visualization via llm multi-agents and editable 3d gaussian splatting.IEEE Transactions on Visualization and Computer Graphics (IEEE VIS 2025), 32(1), 2026. Preprint available athttps://arxiv.org/abs/2507.12621. 2

    work page arXiv 2025

  3. [3]

    W. K. Anderson, R. T. Biedron, J.-R. Carlson, J. M. Derlaga, B. Diskin, C. T. Druyor Jr, P. A. Gnoffo, D. P. Hammond, K. E. Jacobson, W. T. Jones, et al. Fun3d manual: 14.1. 2024. 1

    2024

  4. [4]

    Arndt, W

    D. Arndt, W. Bangerth, D. Davydov, T. Heister, L. Heltai, M. Kronbichler, M. Maier, J.-P. Pelteret, B. Turcksin, and D. Wells. The deal.ii finite element library: Design, features, and insights.Computers & Mathematics with Applications, 81:407–422,

  5. [5]

    doi: 10.1016/j.camwa.2020.02.022 1

    Development and Application of Open-source Software for Problems with Numerical PDEs. doi: 10.1016/j.camwa.2020.02.022 1

    work page doi:10.1016/j.camwa.2020.02.022 2020

  6. [6]

    Ayachit.The ParaView Guide: A Parallel Visualization Application

    U. Ayachit.The ParaView Guide: A Parallel Visualization Application. Kitware, Inc., Clifton Park, NY , USA, 2015. 7

    2015

  7. [7]

    Banesh, R

    D. Banesh, R. Linn, and J. Patchett. V orticity-driven lateral spread ensemble data set. Dataset documentation (PDF), Oct. 2021. Accessed 2026-03-05. 8

    2021

  8. [8]

    Berger and P

    M. Berger and P. Colella. Local adaptive mesh refinement for shock hydrodynamics. Journal of Computational Physics, 82(1):64–84, 1989. doi: 10.1016/0021-9991(89) 90035-1 1

    work page doi:10.1016/0021-9991(89 1989

  9. [9]

    Berger, J

    M. Berger, J. Li, and J. A. Levine. A Generative Model for V olume Rendering . IEEE Transactions on Visualization & Computer Graphics, 25(04):1636–1650, Apr

  10. [10]

    doi: 10.1109/TVCG.2018.2816059 2

    work page doi:10.1109/tvcg.2018.2816059 2018

  11. [11]

    Burstedde, L

    C. Burstedde, L. C. Wilcox, and O. Ghattas. p4est: Scalable algorithms for parallel adaptive mesh refinement on forests of octrees.SIAM J. Sci. Comput., 33(3):1103–1133, 31 pages, May 2011. doi: 10.1137/100791634 1

    work page doi:10.1137/100791634 2011

  12. [12]

    Dubey, K

    A. Dubey, K. Antypas, A. C. Calder, C. Daley, B. Fryxell, J. B. Gallagher, D. Q. Lamb, D. Lee, K. Olson, L. B. Reid, P. Rich, P. M. Ricker, K. M. Riley, R. Rosner, A. Siegel, N. T. Taylor, K. Weide, F. X. Timmes, N. Vladimirova, and J. ZuHone. Evolution of flash, a multi-physics scientific simulation code for high-performance computing.The International J...

    work page doi:10.1177/1094342013505656 2014

  13. [13]

    Dyken, S

    L. Dyken, S. Andres, W. Usher, N. Debardeleben, S. Petruzza, and S. Kumar. V olume encoding gaussians: Transfer function-agnostic 3d gaussians for volume rendering. IEEE Transactions on Visualization and Computer Graphics, 2026. To appear. 2, 3, 5

    2026

  14. [14]

    T. D. Economon, F. Palacios, S. R. Copeland, T. W. Lukaczyk, and J. J. Alonso. Su2: An open-source suite for multiphysics simulation and design.AIAA Journal, 54(3):828–846, 2016. doi: 10.2514/1.J053813 1

    work page doi:10.2514/1.j053813 2016

  15. [15]

    Z. Fan, K. Wang, K. Wen, Z. Zhu, D. Xu, and Z. Wang. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps. In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, eds.,Advances in Neural Information Processing Systems, vol. 37, pp. 140138–140158. Curran Associates, Inc., 2024. doi: 10.52202/079017-4447 9

    work page doi:10.52202/079017-4447 2024

  16. [16]

    Griffin, A

    W. Griffin, A. Godil, J. Bullard, J. Terrill, A. Varshney, and S. Jain. Compressed volume rendering using deep learning. Proceedings of the Large Scale Data Analysis and Visualization (LDA V) Symposium, Phoenix, AZ, 2017-10-02 2017. 2

    2017

  17. [17]

    Hahlbohm, L

    F. Hahlbohm, L. Franke, M. Eisemann, and M. Magnor. Faster-gs: Analyzing and improving gaussian splatting optimization, 2026. 5

    2026

  18. [19]

    M. Han, A. Sewell, J. Insley, J. Knowles, V . A. Mateevitsi, M. E. Papka, S. Petruzza, and S. Rizzi. Toward distributed 3d gaussian splatting for high-resolution isosurface visualization. In2025 IEEE International Conference on eScience (eScience), pp. 331–332, 2025. doi: 10.1109/eScience65000.2025.00056 2

    work page doi:10.1109/escience65000.2025.00056 2025

  19. [20]

    Kerbl, G

    B. Kerbl, G. Kopanas, T. Leimkuehler, and G. Drettakis. 3d gaussian splatting for real-time radiance field rendering.ACM Trans. Graph., 42(4), article no. 139, 14 pages, July 2023. doi: 10.1145/3592433 2, 3, 4, 5

    work page doi:10.1145/3592433 2023

  20. [21]

    Kheradmand, D

    S. Kheradmand, D. Rebain, G. Sharma, W. Sun, Y .-C. Tseng, H. Isack, A. Kar, A. Tagliasacchi, and K. M. Yi. 3d gaussian splatting as markov chain monte carlo. In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, eds.,Advances in Neural Information Processing Systems, vol. 37, pp. 80965–80986. Curran Associates, Inc., 2024....

    work page doi:10.52202/079017-2573 2024

  21. [22]

    D. P. Kingma and J. Ba. Adam: A method for stochastic optimization, 2017. 7

    2017

  22. [23]

    C. C. Kiris, J. A. Housman, M. F. Barad, C. Brehm, E. Sozer, and S. Moini-Yekta. Computational framework for launch, ascent, and vehicle aerodynamics (lava). Aerospace Science and Technology, 55:189–219, 2016. doi: 10.1016/j.ast.2016. 05.008 1

    work page doi:10.1016/j.ast.2016 2016

  23. [24]

    Klacansky

    P. Klacansky. Open scivis datasets, December 2017. https://klacansky.com/open-scivis-datasets/. 7

    2017

  24. [25]

    R. Li, H. Gao, M. Tancik, and A. Kanazawa. Nerfacc: Efficient sampling accelerates nerfs.arXiv preprint arXiv:2305.04966, 2023. 8

    work page arXiv 2023

  25. [26]

    K. Liu, C. Jiao, X. Gao, and C. Bi. Uginr: large-scale unstructured grid reduction via implicit neural representation.J. Vis., 27(5):983–996, 14 pages, June 2024. doi: 10. 1007/s12650-024-01003-y 1, 2, 6, 8

    2024

  26. [27]

    Y . Lu, K. Jiang, J. A. Levine, and M. Berger. Compressive neural representations of volumetric scalar fields.Computer Graphics Forum, 40(3):135–146, 2021. doi: 10. 1111/cgf.14295 1, 2

    2021

  27. [28]

    S. S. Mallick, R. Goel, B. Kerbl, M. Steinberger, F. V . Carrasco, and F. De La Torre. Taming 3dgs: High-quality radiance fields with limited resources. InSIGGRAPH Asia 2024 Conference Papers, SA ’24, article no. 2, 11 pages. Association for Computing Machinery, New York, NY , USA, 2024. doi: 10.1145/3680528.3687694 5

    work page doi:10.1145/3680528.3687694 2024

  28. [29]

    Moenne-Loccoz, A

    N. Moenne-Loccoz, A. Mirzaei, O. Perel, R. de Lutio, J. Martinez Esturo, G. State, S. Fidler, N. Sharp, and Z. Gojcic. 3d gaussian ray tracing: Fast tracing of particle scenes.ACM Trans. Graph., 43(6), article no. 232, 19 pages, Nov. 2024. doi: 10. 1145/3687934 9

    2024

  29. [30]

    Morrical, A

    N. Morrical, A. Sahistan, U. Güdükbay, I. Wald, and V . Pascucci. Quick clusters: A gpu-parallel partitioning for efficient path tracing of unstructured volumetric grids. IEEE Transactions on Visualization and Computer Graphics, 29(1):537–547, 2023. doi: 10.1109/TVCG.2022.3209418 2, 8

    work page doi:10.1109/tvcg.2022.3209418 2023

  30. [31]

    T. Müller. tiny-cuda-nn, 4 2021. 7

    2021

  31. [32]

    Müller, A

    T. Müller, A. Evans, C. Schied, and A. Keller. Instant neural graphics primitives with a multiresolution hash encoding.ACM Trans. Graph., 41(4), article no. 102, 15 pages, July 2022. doi: 10.1145/3528223.3530127 2

    work page doi:10.1145/3528223.3530127 2022

  32. [33]

    Niedermayr, J

    S. Niedermayr, J. Stumpfegger, and R. Westermann. Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis . In2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10349–10358. IEEE Computer Society, Los Alamitos, CA, USA, June 2024. doi: 10.1109/CVPR52733. 2024.00985 9

    work page doi:10.1109/cvpr52733 2024

  33. [34]

    Offermans, A

    N. Offermans, A. Peplinski, O. Marin, and P. Schlatter. Adaptive mesh refinement for steady flows in Nek5000.Computers & Fluids, 2020. doi: 10.1016/j.compfluid. 2019.104352 1

    work page doi:10.1016/j.compfluid 2020

  34. [35]

    Palacios, J

    F. Palacios, J. Alonso, K. Duraisamy, M. Colonno, J. Hicken, A. Aranake, A. Campos, S. Copeland, T. Economon, A. Lonkar, T. Lukaczyk, and T. Taylor. Stanford Univer- sity Unstructured (SU^2): An open-source integrated computational environment for multi-physics simulation and design. In51st AIAA Aerospace Sciences Meeting including the New Horizons Forum ...

    work page doi:10.2514/6.2013-287 2013

  35. [36]

    Papantonakis, G

    P. Papantonakis, G. Kopanas, B. Kerbl, A. Lanvin, and G. Drettakis. Reducing the memory footprint of 3d gaussian splatting.Proc. ACM Comput. Graph. Interact. Tech., 7(1), article no. 16, 17 pages, May 2024. doi: 10.1145/3651282 9

    work page doi:10.1145/3651282 2024

  36. [37]

    J. M. Patchett and G. R. Gisler. Deep water impact ensemble data set, 05 2017. doi: 10.25583/2588620 8

    work page doi:10.25583/2588620 2017

  37. [38]

    Rathke, I

    B. Rathke, I. Wald, K. Chiu, and C. Brownlee. SIMD Parallel Ray Tracing of Homogeneous Polyhedral Grids. InEurographics Symposium on Parallel Graphics and Visualization, 2015. doi: 10.2312/pgv.20151153 8

    work page doi:10.2312/pgv.20151153 2015

  38. [39]

    Rota Bulò, L

    S. Rota Bulò, L. Porzi, and P. Kontschieder. Revising densification in gaussian splatting. InComputer Vision – ECCV 2024: 18th European Conference, Milan, Italy, September 29–October 4, 2024, Proceedings, Part LXIII, 16 pages, p. 347–362. Springer-Verlag, Berlin, Heidelberg, 2024. doi: 10.1007/978-3-031-73036-8_20 5

    work page doi:10.1007/978-3-031-73036-8_20 2024

  39. [40]

    Sitzmann, J

    V . Sitzmann, J. N. P. Martel, A. W. Bergman, D. B. Lindell, and G. Wetzstein. Implicit neural representations with periodic activation functions, 2020. 2

    2020

  40. [41]

    H. Son, J. Noh, S. Jeon, C. Wang, and W.-K. Jeong. Mc-inr: Efficient encoding of multivariate scientific simulation data using meta-learning and clustered implicit neural representations, 2025. 1, 2, 6, 8

    2025

  41. [42]

    SuperSplat.https://github.com/playcanvas/supersplat. 8

  42. [43]

    Lambourne, Karl D

    T. Takikawa, J. Litalien, K. Yin, K. Kreis, C. Loop, D. Nowrouzezahrai, A. Jacobson, M. McGuire, and S. Fidler. Neural Geometric Level of Detail: Real-time Rendering with Implicit 3D Shapes . In2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11353–11362. IEEE Computer Society, Los Alamitos, CA, USA, June 2021. doi: 10.1109/...

    work page doi:10.1109/cvpr46437.2021.01120 2021

  43. [44]

    K. Tang, S. Yao, and C. Wang. ivr-gs: Inverse volume rendering for explorable visu- alization via editable 3d gaussian splatting.IEEE Transactions on Visualization and Computer Graphics, 31(6):3783–3795, 2025. doi: 10.1109/TVCG.2025.3567121 2

    work page doi:10.1109/tvcg.2025.3567121 2025

  44. [45]

    I. Wald, N. Morrical, and S. Zellmann. A memory efficient encoding for ray tracing large unstructured data.IEEE Transactions on Visualization and Computer Graphics, 28(1):583–592, 2022. doi: 10.1109/TVCG.2021.3114869 2

    work page doi:10.1109/tvcg.2021.3114869 2022

  45. [46]

    Wang and J

    C. Wang and J. Han. Dl4scivis: A state-of-the-art survey on deep learning for scientific visualization.IEEE Transactions on Visualization and Computer Graphics, 29(8):3714–3733, 20 pages, Aug. 2023. doi: 10.1109/TVCG.2022.3167896 2

    work page doi:10.1109/tvcg.2022.3167896 2023

  46. [47]

    Weiss, P

    S. Weiss, P. Hermüller, and R. Westermann. Fast neural representations for direct volume rendering.Computer Graphics Forum, 41(6):196–211, June 2022. doi: 10. 1111/cgf.14578 1, 2

    2022

  47. [48]

    Q. Wu, D. Bauer, M. J. Doyle, and K.-L. Ma. Interactive V olume Visualization via Multi-Resolution Hash Encoding based Neural Representation.IEEE Transactions on Visualization and Computer Graphics, 2023. doi: 10.1109/TVCG.2023.3293121 1, 2

    work page doi:10.1109/tvcg.2023.3293121 2023

  48. [49]

    Q. Wu, D. Bauer, M. J. Doyle, and K.-L. Ma. Interactive volume visualization via multi-resolution hash encoding based neural representation.IEEE Transactions on Visualization and Computer Graphics, 30(8):5404–5418, 2024. doi: 10.1109/TVCG. 2023.3293121 2, 7

    work page doi:10.1109/tvcg 2024

  49. [50]

    S. W. Wurster and H.-W. Shen. Amgsrn++: Improved adaptive srn for scientific visualization. In2025 IEEE 18th Pacific Visualization Conference (PacificVis), pp. 182–191, 2025. doi: 10.1109/PacificVis64226.2025.00024 2, 7, 8

    work page doi:10.1109/pacificvis64226.2025.00024 2025

  50. [51]

    S. W. Wurster, T. Xiong, H.-W. Shen, H. Guo, and T. Peterka. Adaptively placed multi-grid scene representation networks for large-scale data visualization.IEEE Transactions on Visualization and Computer Graphics, 30(1):965–974, 2024. doi: 10.1109/TVCG.2023.3327194 1, 2

    work page doi:10.1109/tvcg.2023.3327194 2024

  51. [52]

    S. Yao, Y . Lu, and C. Wang. ViSNeRF: Efficient Multidimensional Neural Ra- diance Field Representation for Visualization Synthesis of Dynamic V olumetric Scenes . In2025 IEEE 18th Pacific Visualization Conference (PacificVis), pp. 235–

  52. [53]

    IEEE Computer Society, Los Alamitos, CA, USA, Apr. 2025. doi: 10.1109/ PacificVis64226.2025.00029 2

    work page arXiv 2025

  53. [54]

    Yao and C

    S. Yao and C. Wang. Revolve: Neural reconstruction of volumes for visualization enhancement of direct volume rendering.Computers & Graphics, 131:104295, 2025. doi: 10.1016/j.cag.2025.104295 2

    work page doi:10.1016/j.cag.2025.104295 2025

  54. [55]

    H. Zhao, H. Weng, D. Lu, A. Li, J. Li, A. Panda, and S. Xie. On scaling up 3d gaussian splatting training, 2024. 2 A DATASETINFORMATION Dataset Dimensions Total Size Vertebra 512×512×512 537 MB Miranda 1024×1024×1024 4.29 GB Chameleon 1024×1024×1080 4.53 GB Richtmyer 2048×2048×1920 32.21 GB Table 4: Structured volume datasets sorted by total size. Points ...

    2024