Pith. sign in

REVIEW 4 major objections 4 minor 39 references

The paper claims that a physics-based simulation environment plus Neuralangelo can reconstruct static satellite models whose meshes closely match their CAD references, establishing a baseline for evaluating dynamic scene reconstruction of t

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

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

T0 review

2026-08-04 21:27 UTC pith:J6IW3K44

load-bearing objection Title promises dynamic reconstruction; the paper delivers only a static baseline and a simulator, with no quantitative evidence. the 4 major comments →

arxiv 2509.07932 v1 pith:J6IW3K44 submitted 2025-09-09 cs.CV

Dynamic Scene 3D Reconstruction of an Uncooperative Resident Space Object

Bala Prenith Reddy Gopu , Timothy Jacob Huber , George M. Nehma , Patrick Quinn , Madhur Tiwari , Matt Ueckermann , David Hinckley , Christopher McKenna This is my paper
classification cs.CV
keywords uncooperative resident space object3D reconstructionneural radiance fieldsNeuralangeloClohessy-Wiltshire relative motionIsaac Sim simulationspace debris characterizationcloud-to-mesh distance
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The authors aim to assess how well state-of-the-art dynamic 3D reconstruction algorithms can model an uncooperative tumbling satellite observed by a moving camera spacecraft, a scenario with dual motion that has not been studied before. To support that evaluation, they built a simulation environment that generates realistic synthetic imagery of satellites under orbital lighting and relative-motion dynamics. Their delivered results so far are on static scenes: using Neuralangelo, the reconstructed meshes of three satellite models closely match the original CAD geometry, with small signed cloud-to-mesh distances and capture of fine detail, except for thin protruding structures like a magnetometer. These static results are presented as a baseline for the planned dynamic evaluation, not as the final contribution.

Core claim

Using a custom simulation pipeline built on Isaac Sim and the Clohessy–Wiltshire relative-orbit equations, the authors generated synthetic monocular image sequences of three satellite CAD models (a 1RU CubeSat, DESDynI, and GOES-R) in a static scene. They then applied Neuralangelo, a neural surface reconstruction method, and compared the extracted meshes with the reference CAD models using signed cloud-to-mesh (C2M) distances. The meshes closely matched the references, with errors concentrated in small local regions and fine details preserved. The one systematic failure mode noted is that thin structures, exemplified by a GOES-R magnetometer, can be completely absent from the reconstruction.

What carries the argument

The key machinery is the simulation framework: Isaac Sim generates physically rendered images with ground-truth depth, pose, and lighting, while the observer spacecraft follows a 45-degree-inclined, stably bounded Clohessy–Wiltshire relative orbit whose initial conditions are derived from a rotation of the standard bounded-orbit condition. This provides camera poses that are assumed perfectly known. For the static baseline, Neuralangelo—a coarse-to-fine optimization applied to multi-hash encoding—carries the reconstruction, producing high-fidelity surface meshes that are quantitatively compared to the CAD model via signed C2M distances in CloudCompare.

Load-bearing premise

The pipeline assumes that the camera poses are known exactly from the simulator's ground truth, so the reconstruction never has to contend with pose-estimation error; in a real uncooperative rendezvous mission, pose uncertainty is a primary error source and likely to dominate reconstruction quality.

What would settle it

Run the same Neuralangelo reconstruction on real imagery of an uncooperative spacecraft (e.g., from a ground-based telescope or an inspection mission) where camera poses must be estimated from the images themselves; if the mesh-to-CAD error increases sharply or thin structures disappear entirely, the simulator-only transferability claim is falsified. A simpler check within the paper's own framework: add realistic pose noise to the ground-truth camera positions and measure how C2M error degrades.

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

Share X Bluesky LinkedIn Reddit HN

If this is right

  • If the static baseline holds, mission planners can expect that neural surface reconstruction from monocular imagery can recover satellite geometry well enough to support proximity operations, provided thin appendages are not mission-critical.
  • The 45-degree inclined relative orbit ensures that top and bottom surfaces of the target are imaged, which is a necessary condition for complete reconstruction; the paper suggests this trajectory design generalizes to other targets.
  • The evaluation pipeline, with its ground-truth poses and CAD references, provides a controlled way to compare dynamic reconstruction algorithms (D-NeRF, Nerfies, TiNeuVox, HexPlane) on a tumbling target, which is the paper's advertised next step.
  • The current C2M metric fails to penalize missing geometry, so the authors propose reversing the distance direction (reference-to-reconstruction) to catch such omissions; this would make the error metric more reliable for RSO characterization.
  • Neuralangelo's approximately 8-hour runtime per reconstruction indicates that the static baseline is not real-time, but the paper does not claim operational speed, only fidelity.

Where Pith is reading between the lines

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

  • The dynamic-scene evaluation, the paper's central advertised goal, is not yet presented; the reader should treat the static results as a proof-of-concept of the simulator and reconstruction pipeline rather than evidence about dynamic algorithms.
  • The assumption of perfectly known camera poses from the simulator is the main barrier to transferability: in a real uncooperative RPO mission, pose estimation error would likely be the dominant source of reconstruction error, so a natural test is to add pose noise or run structure-from-motion on the same imagery.
  • If the dynamic evaluation proceeds with the same metric, the missing-thin-structures failure mode may be even more pronounced for a tumbling target, where fast motion and occlusion can cause appendages to be imaged only briefly.
  • One practical extension would be using the reconstructed static mesh as an initialization prior for dynamic Gaussian splatting, similar to existing coarse-shape-prior approaches, to accelerate convergence on the tumbling scenario.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper describes a simulation framework for 3D reconstruction of an uncooperative resident space object, using Isaac Sim to render images of three CAD models (CubeSat, DESDynI, GOES-R) under Clohessy–Wiltshire relative orbits. The stated goal is to evaluate dynamic-scene NeRF variants (D-NeRF, Nerfies, TiNeuVox, HexPlane) for a tumbling target observed during a fly-around. The only delivered results, however, are static-scene reconstructions using Neuralangelo, assessed through qualitative C2M heatmaps and histograms. The dynamic evaluation is explicitly deferred to future work in Section VI. The manuscript also describes planned PSNR/SSIM/LPIPS metrics that are not yet applied.

Significance. If completed, the proposed evaluation would address a genuinely useful gap: the combination of target tumbling and observer fly-around creates a dual-motion scenario that existing dynamic NeRF evaluations have not covered. The Isaac Sim-based synthetic data pipeline, with ground-truth poses/depth and an inclined relative orbit designed to observe all spacecraft surfaces, is a sensible building block. As it stands, however, the only present result is a qualitative static baseline with Neuralangelo; the central advertised contribution is absent. The paper therefore reads as a progress report rather than a completed study, and the current evidence is too weak to support the abstract's claim of reconstructions that 'closely match' the CAD models.

major comments (4)
  1. [Abstract, Section II, Section VI] The central claim that this study evaluates state-of-the-art dynamic-scene reconstruction algorithms is not supported by any experiment. Section VI explicitly states that the dynamic evaluation is future work. The only results are static reconstructions with Neuralangelo, while D-NeRF, Nerfies, TiNeuVox, and HexPlane are never tested. The title and abstract therefore overstate the delivered contribution. The manuscript must either include the promised dynamic experiments (with the stated PSNR/SSIM/LPIPS metrics) or be reframed as a static baseline and simulation-framework paper with appropriately revised claims.
  2. [Section V.A, Figs. 5-7] The claim that the generated meshes 'closely match the original CAD models with minimal errors and artifacts' is not quantitatively supported. No C2M summary statistics (mean error, RMS, percentiles, or confidence intervals) are reported; only example heatmaps and histograms are shown. Moreover, the one-way C2M direction used cannot detect missing geometry, as the authors acknowledge for the missing GOES-R magnetometer in Figure 6. The reverse direction (reference mesh to reconstruction) or a volumetric/directional metric is needed to support the claim of capturing 'critical fine details.' The paper also acknowledges that thin structures commonly fail to reconstruct, which further undercuts the current wording. Please report full quantitative metrics per model, preferably normalized by model scale and accompanied by error distributions.
  3. [Section IV.A, Section IV.B] Camera poses are taken as ground truth from the simulator, so the reconstruction pipeline never encounters pose estimation error. In uncooperative RPO, pose uncertainty is a primary error source and is likely to dominate reconstruction quality. This assumption is load-bearing for the static baseline because Neuralangelo requires accurate camera poses. The authors should either evaluate sensitivity to pose perturbations or clearly state this limitation and restrict the transferability claims accordingly. The same caveat applies to the planned dynamic experiments.
  4. [Equation (2), Section IV.B] The Clohessy–Wiltshire state-space equation appears incorrectly typeset or dimensionally inconsistent. The A matrix rows are not aligned with the stated state vector [x,y,z,xdot,ydot,zdot], and the standard CW coupling terms (the 2n*ydot term in x-double-dot and the -2n*xdot term in y-double-dot) are not visible as written. Since this equation defines the observer trajectory used in the simulations, the authors should correct it and, ideally, validate the generated relative orbit against the analytic CW solution.
minor comments (4)
  1. [Section V.A, Fig. 5/Fig. 7] Figure referencing is inconsistent: the text says 'Example heatmaps ... can be seen in Figure 7' but Figure 7 is the histogram; the heatmap is Figure 5. Please correct the cross-references.
  2. [Throughout] Several typographical and labeling issues: 'claasification' -> 'classification', 'a object' -> 'an object', 'results findings results' -> 'results', 'potentiality' -> 'potentially', 'DESDynl'/'DESTINL' vs. 'DESDynI', and 'Ecosytem' -> 'Ecosystem'. Reference [26] is also incomplete, missing venue and year.
  3. [Section III, Section IV.C] PSNR, SSIM, and LPIPS are image-rendering metrics and are not used in the static geometric evaluation presented in Section V. It would help to state explicitly whether these metrics will be used only for the dynamic evaluation, and how they relate to geometric accuracy.
  4. [Section IV.A] The description of the 45-degree inclined orbit is useful but says 'rotation of 90 degrees around o_h and a 45-degree rotation around o_r' while the DCM parameters are listed as theta=45, phi=0, psi=90. The relationship between the described rotations and the Euler-angle convention should be clarified.

Circularity Check

0 steps flagged

No circularity; the paper is an empirical evaluation using an off-the-shelf algorithm against independent CAD ground truth.

full rationale

No circular derivation is present. The paper's delivered claim is an empirical baseline: images rendered in Isaac Sim from public CAD models are fed to Neuralangelo, an off-the-shelf algorithm, and reconstructed meshes are compared to the same CAD reference using C2M distances. The CAD reference is not an input to Neuralangelo; it enters only through rendered images, so the evaluation is not definitionally equivalent to the output. Camera poses are taken from known simulation states, which is an idealization but not a circular inference. Self-references [8, 9, 17] appear only in related-work context and do not carry the reconstruction claim. The one-way C2M metric and missing numeric values weaken the 'close match' claim, but that is an evidence-quality issue, not circularity. Dynamic-scene claims are explicitly deferred to future work, so no predicted result is being asserted from fitted inputs.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

No new physical entities are introduced. The load-bearing assumptions are standard orbital mechanics approximations and simulation fidelity assumptions. The hand-chosen scenario parameters (range, inclination, rotation rate) affect the coverage and difficulty of the reconstruction but are not fitted to data. They are not used to back-derive a claimed result, so they do not create circularity, but they do limit generalizability.

free parameters (3)
  • Observer initial standoff distance x0 = 40 m
    Hand-chosen scenario parameter. It controls image scale and reconstruction quality but is not justified by any stated optimization.
  • Relative orbit inclination = 45 degrees
    Hand-chosen to maximize surface coverage of the CubeSat. The paper notes planar orbits miss top and bottom surfaces.
  • Target rotation rate (preliminary) = 3 deg/s about z-axis
    Chosen as a starting point for the dynamic dataset; no analysis of sensitivity to this rate is provided.
axioms (4)
  • domain assumption Clohessy-Wiltshire linearized relative motion is valid for the simulated scenario
    Assumed circular LEO orbit, small relative distance, and no perturbations (Section IV.B). Violations would change the camera trajectory and the geometry observed.
  • domain assumption Isaac Sim renderer is physically accurate for space lighting
    The paper assumes a distant light approximating the sun produces realistic conditions. No validation against actual space imagery is presented.
  • domain assumption Perfect knowledge of camera poses
    The reconstruction pipeline uses simulator-provided ground-truth poses. Real RPO missions have pose uncertainty from navigation, which is not modeled.
  • domain assumption Neuralangelo is a valid black-box reconstruction method for this domain
    The paper treats Neuralangelo as reliable and does not compare against a baseline or analyze its failure modes beyond a brief mention of thin objects.

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Dynamic Scene 3D Reconstruction of an Uncooperative Resident Space Object." pith.science (2026). https://pith.science/paper/J6IW3K44

@misc{pith2026250907932,
  author       = {Pith},
  title        = {Pith review of: Dynamic Scene 3D Reconstruction of an Uncooperative Resident Space Object},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J6IW3K44}},
  note         = {Machine review of arXiv:2509.07932}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Characterization of uncooperative Resident Space Objects (RSO) play a crucial role in On-Orbit Servicing (OOS) and Active Debris Removal (ADR) missions to assess the geometry and motion properties. To address the challenges of reconstructing tumbling uncooperative targets, this study evaluates the performance of existing state-of-the-art 3D reconstruction algorithms for dynamic scenes, focusing on their ability to generate geometrically accurate models with high-fidelity. To support our evaluation, we developed a simulation environment using Isaac Sim to generate physics-accurate 2D image sequences of tumbling satellite under realistic orbital lighting conditions. Our preliminary results on static scenes using Neuralangelo demonstrate promising reconstruction quality. The generated 3D meshes closely match the original CAD models with minimal errors and artifacts when compared using Cloud Compare (CC). The reconstructed models were able to capture critical fine details for mission planning. This provides a baseline for our ongoing evaluation of dynamic scene reconstruction.

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

39 extracted references · 31 canonical work pages

  1. [1]

    Space Debris by the Numbers,

    Agency, E. S., “Space Debris by the Numbers,”https://www.esa.int/Safety_Security/Space_Debris/Space_ debris_by_the_numbers, 2024. Accessed: 2025-05-09

    work page 2024

  2. [2]

    An active debris removal parametric study for LEO environment remediation,

    Liou, J.-C., “An active debris removal parametric study for LEO environment remediation,”Advances in Space Research, Vol. 47, No. 11, 2011, pp. 1865–1876

    work page 2011

  3. [3]

    Review of active space debris capturing methods,

    Shan, M., Guo, J., and Gill, E., “Review of active space debris capturing methods,”Progress in Aerospace Sciences, Vol. 80, 2016, pp. 18–32

    work page 2016

  4. [4]

    Fehse, W.,Automated Rendezvous and Docking of Spacecraft, Cambridge University Press, 2003

    work page 2003

  5. [5]

    Vision-Based Navigation for Autonomous Rendezvous with Uncooperative Spacecraft,

    Psomopoulos, E., and Tsiotras, P., “Vision-Based Navigation for Autonomous Rendezvous with Uncooperative Spacecraft,” Acta Astronautica, Vol. 173, 2020, pp. 167–180

    work page 2020

  6. [6]

    2 Private Satellites Undock After Pioneering Life-extension Mission,

    Jones, A., “2 Private Satellites Undock After Pioneering Life-extension Mission,” , 2025. URL https: //www.space.com/space-exploration/satellites/2-private-satellites-undock-after-pioneering- life-extension-mission

    work page 2025

  7. [7]

    Astroscale’s ADRAS-J Achieves Historic 15-Meter Approach to Space Debris,

    Astroscale, “Astroscale’s ADRAS-J Achieves Historic 15-Meter Approach to Space Debris,” , 2024. URLhttps:// astroscale.com/astroscales-adras-j-achieves-historic-15-meter-approach-to-space-debris/

    work page 2024

  8. [8]

    Spaceyolo: A human-inspired model for real-time, on-board spacecraft feature detection,

    Mahendrakar, T., White, R. T., Wilde, M., and Tiwari, M., “Spaceyolo: A human-inspired model for real-time, on-board spacecraft feature detection,”2023 IEEE Aerospace Conference, IEEE, 2023, pp. 01–11

    work page 2023

  9. [9]

    Unknown non-cooperative spacecraft characterization with lightweight convolutional neural networks,

    Mahendrakar, T., White, R. T., Tiwari, M., and Wilde, M., “Unknown non-cooperative spacecraft characterization with lightweight convolutional neural networks,”Journal of Aerospace Information Systems, Vol. 21, No. 5, 2024, pp. 455–460

    work page 2024

  10. [10]

    A review of space robotics technologies for on-orbit servicing,

    Flores-Abad, A., Ma, O., Pham, K., and Ulrich, S., “A review of space robotics technologies for on-orbit servicing,”Progress in Aerospace Sciences, Vol. 68, 2014, pp. 1–26

    work page 2014

  11. [11]

    High-Fidelity Shape Estimation for Non-Cooperative Space Targets,

    Seube, N., Courmontagne, J., and Petit, N., “High-Fidelity Shape Estimation for Non-Cooperative Space Targets,”Journal of Guidance, Control, and Dynamics, Vol. 42, No. 2, 2019, pp. 308–322

    work page 2019

  12. [12]

    A review on 3D reconstruction techniques from 2D images,

    Aharchi, M., and Ait Kbir, M., “A review on 3D reconstruction techniques from 2D images,”Innovations in Smart Cities Applications Edition 3: The Proceedings of the 4th International Conference on Smart City Applications 4 , Springer, 2020, pp. 510–522

    work page 2020

  13. [13]

    A Point Set Generation Network for 3D Object Reconstruction From a Single Image,

    Fan, H., Su, H., and Guibas, L. J., “A Point Set Generation Network for 3D Object Reconstruction From a Single Image,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2017

    work page 2017

  14. [14]

    Pix2Surf: Learning Parametric 3D Surface Models of Objects from Images,

    Lei, J., Sridhar, S., Guerrero, P., Sung, M., Mitra, N., and Guibas, L. J., “Pix2Surf: Learning Parametric 3D Surface Models of Objects from Images,” , 2020. URLhttps://arxiv.org/abs/2008.07760

    Pith/arXiv arXiv 2020

  15. [15]

    3D reconstruction from multiple images part 1: Principles,

    Moons, T., Van Gool, L., Vergauwen, M., et al., “3D reconstruction from multiple images part 1: Principles,”Foundations and Trends® in Computer Graphics and Vision, Vol. 4, No. 4, 2010, pp. 287–404

    work page 2010

  16. [16]

    FineRecon: Depth-aware Feed-forward Network for Detailed 3D Reconstruction,

    Stier, N., Ranjan, A., Colburn, A., Yan, Y., Yang, L., Ma, F., and Angles, B., “FineRecon: Depth-aware Feed-forward Network for Detailed 3D Reconstruction,” , 2023. URLhttps://arxiv.org/abs/2304.01480

    Pith/arXiv arXiv 2023

  17. [17]

    High-Fidelity3DReconstructionofSpaceBodiesUsingMachineLearningandNeuralRadianceFields,

    Huber,T.J.,“High-Fidelity3DReconstructionofSpaceBodiesUsingMachineLearningandNeuralRadianceFields,”Master’s thesis, Florida Institute of Technology, 2024. URLhttps://repository.fit.edu/etd/1442

    work page 2024

  18. [18]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    Mildenhall, B., Srinivasan, P. P., Tancik, M., Barron, J. T., Ramamoorthi, R., and Ng, R., “Nerf: Representing scenes as neural radiance fields for view synthesis,”Communications of the ACM, Vol. 65, No. 1, 2021, pp. 99–106. 9

    work page 2021

  19. [19]

    3d gaussian splatting for real-time radiance field rendering

    Kerbl, B., Kopanas, G., Leimkühler, T., and Drettakis, G., “3d gaussian splatting for real-time radiance field rendering.”ACM Trans. Graph., Vol. 42, No. 4, 2023, pp. 139–1

    work page 2023

  20. [20]

    Vision-based neural scene representations for spacecraft,

    Mergy, A., Lecuyer, G., Derksen, D., and Izzo, D., “Vision-based neural scene representations for spacecraft,”Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 2002–2011

    2021

  21. [21]

    On Optimal Observation Orbits for Learning Gaussian Splatting-Based 3D Models of Unknown Resident Space Objects,

    Issitt, A., Mahendrakar, T., Alvarez, A., White, R. T., and Sizemore, A., “On Optimal Observation Orbits for Learning Gaussian Splatting-Based 3D Models of Unknown Resident Space Objects,”AIAA SCITECH 2025 Forum, 2025, p. 1780

    2025

  22. [22]

    3d reconstruction of non-cooperative resident space objects using instant ngp-accelerated nerf and d-nerf,

    Caruso, B., Mahendrakar, T., Nguyen, V. M., White, R. T., and Steffen, T., “3d reconstruction of non-cooperative resident space objects using instant ngp-accelerated nerf and d-nerf,”arXiv preprint arXiv:2301.09060, 2023

    Pith/arXiv arXiv 2023

  23. [23]

    Neural Field-Based Space Target 3D Reconstruction with Predicted Depth Priors,

    Fu, T., Zhou, Y., Wang, Y., Liu, J., Zhang, Y., Kong, Q., and Chen, B., “Neural Field-Based Space Target 3D Reconstruction with Predicted Depth Priors,”Aerospace, Vol. 11, No. 12, 2024, p. 997

    2024

  24. [24]

    Rapid abstraction of spacecraft 3d structure from single 2d image,

    Park, T. H., and D’Amico, S., “Rapid abstraction of spacecraft 3d structure from single 2d image,”AIAA SCITECH 2024 Forum, 2024, p. 2768

    2024

  25. [25]

    Removing Ambiguities in Concurrent Monocular Single-Shot Spacecraft Shape and Pose Estimation Using a Deep Neural Network,

    Bates, E., and D’Amico, S., “Removing Ambiguities in Concurrent Monocular Single-Shot Spacecraft Shape and Pose Estimation Using a Deep Neural Network,”47th Rocky Mountain AAS Guidance, Naviga-tion and Control Conference, 2025

    work page 2025

  26. [26]

    FAST LEARNING OF NON-COOPERATIVE SPACECRAFT 3D MODELS THROUGH PRIMITIVE INITIALIZATION,

    Huc, P. F., Bates, E., and D’Amico, S., “FAST LEARNING OF NON-COOPERATIVE SPACECRAFT 3D MODELS THROUGH PRIMITIVE INITIALIZATION,” ????

    work page

  27. [27]

    Characterizingsatellitegeometryviaaccelerated3DGaussian splatting,

    Nguyen,V.M.,Sandidge,E.,Mahendrakar,T.,andWhite,R.T.,“Characterizingsatellitegeometryviaaccelerated3DGaussian splatting,” Aerospace, Vol. 11, No. 3, 2024, p. 183

    work page 2024

  28. [28]

    D-nerf: Neural radiance fields for dynamic scenes,

    Pumarola, A., Corona, E., Pons-Moll, G., and Moreno-Noguer, F., “D-nerf: Neural radiance fields for dynamic scenes,” Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 10318–10327

    work page 2021

  29. [29]

    Nerfies: Deformable neural radiance fields,

    Park, K., Sinha, U., Barron, J. T., Bouaziz, S., Goldman, D. B., Seitz, S. M., and Martin-Brualla, R., “Nerfies: Deformable neural radiance fields,”Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 5865–5874

    work page 2021

  30. [30]

    Fastdynamicradiancefieldswithtime-aware neural voxels,

    Fang, J., Yi, T.,Wang,X., Xie, L., Zhang,X., Liu, W.,Nießner,M., andTian, Q., “Fastdynamicradiancefieldswithtime-aware neural voxels,”SIGGRAPH Asia 2022 Conference Papers, 2022, pp. 1–9

    work page 2022

  31. [31]

    Hexplane: A fast representation for dynamic scenes,

    Cao, A., and Johnson, J., “Hexplane: A fast representation for dynamic scenes,”Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 130–141

    work page 2023

  32. [32]

    Accessed: May 08, 2025

    NASA, https://github.com/nasa/nasa-3d-resources, n.d.. Accessed: May 08, 2025

    work page 2025

  33. [33]

    Accessed: May 08, 2025

    NASA, https://nasa3d.arc.nasa.gov/models, n.d.. Accessed: May 08, 2025

    work page 2025

  34. [34]

    Accessed: May 20, 2023

    TurboSquid, https://www.turbosquid.com/3d-models, n.d. Accessed: May 20, 2023

    work page 2023

  35. [35]

    L.,Analytical Mechanics of Space Systems, American Institute of Aeronautics and Astronautics, Incorporated, 2014

    Schaub, H., and Junkins, J. L.,Analytical Mechanics of Space Systems, American Institute of Aeronautics and Astronautics, Incorporated, 2014

    work page 2014

  36. [36]

    Relative orbit geometry through classical orbit element differences,

    Schaub, H., “Relative orbit geometry through classical orbit element differences,”J. Guid. Control Dyn., Vol. 27, No. 5, 2004, pp. 839–848

    work page 2004

  37. [37]

    Neuralangelo: High-fidelityneuralsurface reconstruction,

    Li,Z.,Müller,T.,Evans,A.,Taylor,R.H.,Unberath,M.,Liu,M.-Y.,andLin,C.-H.,“Neuralangelo: High-fidelityneuralsurface reconstruction,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 8456–8465

    work page 2023

  38. [38]

    MeshLab: an Open-Source Mesh Processing Tool,

    Cignoni, P., Callieri, M., Corsini, M., Dellepiane, M., Ganovelli, F., and Ranzuglia, G., “MeshLab: an Open-Source Mesh Processing Tool,”Eurographics Italian Chapter Conference, edited by V. Scarano, R. De Chiara, and U. Erra, The Eurographics Association, 2008. https://doi.org/10.2312/LocalChapterEvents/ItalChap/ItalianChapConf2008/129-136

    doi:10.2312/localchapterevents/italchap/italianchapconf2008/129-136 2008

  39. [39]

    CloudCompare,

    CloudCompare, “CloudCompare,” , 2021. URL http://www.cloudcompare.org, gPL software. 10

    work page 2021

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.