pith. sign in

arxiv: 2604.17155 · v1 · submitted 2026-04-18 · 💻 cs.CV · cs.GR

Instant Colorization of Gaussian Splats

Daniel Lieber , Alexander Mock , Nils Wandel This is my paper

Pith reviewed 2026-05-10 06:15 UTC · model grok-4.3

classification 💻 cs.CV cs.GR
keywords Gaussian splattingcolorizationleast squaresscene relightingsemantic segmentationdifferentiable renderingnormal equations
0
0 comments X

The pith

Mapping 2D image data onto 3D Gaussian splats is achieved by solving a visibility-weighted least squares problem independently for each Gaussian using the normal equation.

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

The paper develops a method to efficiently project 2D information such as colors, neural features, or segmentation masks back onto an existing collection of 3D Gaussian splats. It formulates the mapping as a separate visibility-weighted least squares problem per Gaussian and solves it directly with the normal equation rather than iterative gradient descent. This leverages existing differentiable rasterizers for implementation. The resulting speed enables practical use in tasks like updating scene colors from new views, adding semantic labels in 3D, and feature enrichment. A reader would care because prior optimization approaches were too slow for interactive or large-scale applications.

Core claim

We solve a visibility-weighted least squares problem for every Gaussian using the normal equation to map 2D image information onto 3D Gaussian splats. The approach handles occlusions and view-dependent effects through efficient computation with differentiable rasterizers and produces consistent results across views.

What carries the argument

Visibility-weighted least squares problem solved independently per Gaussian via the normal equation

If this is right

  • Scene relighting becomes possible by instantly updating Gaussian colors from new 2D images.
  • 3D semantic segmentation and feature enrichment can be performed directly on the splat representation.
  • Up to an order of magnitude speedup is obtained over gradient-descent baselines on the demonstrated tasks.
  • The method works with any existing differentiable rasterizer without custom code.

Where Pith is reading between the lines

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

  • The independent per-Gaussian computation opens the door to straightforward GPU parallelization across millions of splats.
  • Incremental updates to the scene could be supported by re-solving only the affected Gaussians when new views arrive.
  • Similar per-element closed-form solves might apply to other point-based or particle representations in graphics.

Load-bearing premise

Independently solving the visibility-weighted least squares problem for each Gaussian with the normal equation is enough to correctly account for occlusions and produce consistent multi-view colorizations without any global optimization.

What would settle it

Rendering the resulting colored splats from novel viewpoints that expose heavy Gaussian overlaps and checking for color inconsistencies or artifacts would test whether the independent per-Gaussian solves suffice.

Figures

Figures reproduced from arXiv: 2604.17155 by Alexander Mock, Daniel Lieber, Nils Wandel.

Figure 1
Figure 1. Figure 1: Architecture overview: First, camera poses and an initial point cloud are generated from a set of original scene photos using [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Examples of our captured scenes under different lighting [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Light fields can be linearly combined. (left): equal [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Train and Test Loss curves for different optimizers. Our [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Scenes colorized with 3 PCA components of DINOv2 [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Segmentation results on LLFF dataset by SAGA [ [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
read the original abstract

Gaussian Splatting has recently become one of the most popular frameworks for photorealistic 3D scene reconstruction and rendering. While current rasterizers allow for efficient mappings of 3D Gaussian splats onto 2D camera views, this work focuses on mapping 2D image information (e.g. color, neural features or segmentation masks) efficiently back onto an existing scene of Gaussian splats. This 'opposite' direction enables applications ranging from scene relighting and stylization to 3D semantic segmentation, but also introduces challenges, such as view-dependent colorization and occlusion handling. Our approach tackles these challenges using the normal equation to solve a visibility-weighted least squares problem for every Gaussian and can be implemented efficiently with existing differentiable rasterizers. We demonstrate the effectiveness of our approach on scene relighting, feature enrichment and 3D semantic segmentation tasks, achieving up to an order of magnitude speedup compared to gradient descent-based baselines.

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

1 major / 1 minor

Summary. The paper proposes an efficient method to map 2D image data (colors, features, segmentation masks) back onto pre-existing 3D Gaussian splats by solving a visibility-weighted least-squares problem independently for each Gaussian via the normal equation. The approach is implemented using existing differentiable rasterizers and is evaluated on scene relighting, feature enrichment, and 3D semantic segmentation, with reported speedups of up to an order of magnitude over gradient-descent baselines.

Significance. If the per-Gaussian normal-equation solves prove accurate, the technique would offer a practical, non-iterative alternative for attribute assignment in Gaussian splatting pipelines, potentially enabling faster post-processing for relighting, stylization, and semantic tasks in 3D vision and graphics.

major comments (1)
  1. [3] The central technical claim—that independently solving the visibility-weighted least-squares problem per Gaussian via the normal equation correctly recovers colors/features despite alpha-blended overlaps—requires explicit justification. Because rasterization composites contributions nonlinearly through depth-ordered alpha blending, local per-Gaussian solutions may not invert the shared pixel observations without residual error or cross-Gaussian inconsistency; a derivation or counter-example analysis of the multi-Gaussian case is needed to support the claim.
minor comments (1)
  1. The abstract and method description would benefit from a short pseudocode listing or explicit matrix dimensions for the normal-equation solve to clarify implementation with the rasterizer.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive summary and for identifying an important point that merits explicit treatment in the manuscript. We address the major comment below and will revise accordingly.

read point-by-point responses

  1. Referee: The central technical claim—that independently solving the visibility-weighted least-squares problem per Gaussian via the normal equation correctly recovers colors/features despite alpha-blended overlaps—requires explicit justification. Because rasterization composites contributions nonlinearly through depth-ordered alpha blending, local per-Gaussian solutions may not invert the shared pixel observations without residual error or cross-Gaussian inconsistency; a derivation or counter-example analysis of the multi-Gaussian case is needed to support the claim.

    Authors: We agree that the nonlinear nature of alpha blending requires careful justification. Our formulation defines a per-Gaussian visibility-weighted least-squares objective whose weights are obtained directly from the differentiable rasterizer (alpha and transmittance terms). The normal equation then yields the exact closed-form minimizer of that objective for each Gaussian independently. In the revision we will add a dedicated subsection containing (i) the full derivation showing that the visibility weights linearize the contribution of each Gaussian to the observed pixels, (ii) a first-order error analysis bounding the residual due to overlapping splats, and (iii) a counter-example study on synthetic multi-Gaussian scenes that quantifies reconstruction error and cross-Gaussian consistency. These additions will make the correctness argument explicit while preserving the method’s non-iterative advantage. revision: yes

Circularity Check

0 steps flagged

No circularity; standard normal-equation solve on explicitly defined per-Gaussian least-squares problem.

full rationale

The paper's central step is to formulate a visibility-weighted least-squares objective per Gaussian (using weights from an existing differentiable rasterizer) and solve it in closed form via the normal equations. This is a direct, non-iterative application of linear algebra to a problem whose inputs (visibility weights, observed pixel values) are obtained externally and whose outputs (per-Gaussian colors or features) are not fed back into the definition of the objective or the weights. No self-definitional loop exists, no fitted parameter is relabeled as a prediction, and no load-bearing claim reduces to a self-citation whose validity depends on the present work. The approach is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach relies on standard linear algebra and assumptions from differentiable rendering without introducing new free parameters or invented entities in the abstract.

axioms (2)
  • standard math The normal equation provides the closed-form optimal solution to a linear least squares problem.
    Core mathematical tool invoked for solving the per-Gaussian problem.
  • domain assumption Visibility weights can be reliably obtained from existing differentiable rasterizers.
    The weighting in the least squares problem depends on this capability.

pith-pipeline@v0.9.0 · 5451 in / 1351 out tokens · 54580 ms · 2026-05-10T06:15:27.592676+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

61 extracted references · 61 canonical work pages · 2 internal anchors

  1. [1]

    Segment any 3d gaussians.Proceedings of the AAAI Conference on Artificial Intelligence, 39(2):1971–1979, 2025

    Jiazhong Cen, Jiemin Fang, Chen Yang, Lingxi Xie, Xi- aopeng Zhang, Wei Shen, and Qi Tian. Segment any 3d gaussians.Proceedings of the AAAI Conference on Artificial Intelligence, 39(2):1971–1979, 2025. 3, 8

    work page 1971

  2. [2]

    Textured Gaus- sians for enhanced 3d scene appearance modeling

    Brian Chao, Hung-Yu Tseng, Lorenzo Porzi, Chen Gao, Tuotuo Li, Qinbo Li, Ayush Saraf, Jia-Bin Huang, Johannes Kopf, Gordon Wetzstein, and Changil Kim. Textured Gaus- sians for enhanced 3d scene appearance modeling. InCVPR,

    work page

  3. [3]

    Quantifying and alleviating co-adaptation in sparse-view 3d gaussian splatting,

    Kangjie Chen, Yingji Zhong, Zhihao Li, Jiaqi Lin, Youyu Chen, Minghan Qin, and Haoqian Wang. Quantifying and alleviating co-adaptation in sparse-view 3d Gaussian splat- ting.arXiv preprint arXiv:2508.12720, 2025. 1

    work page arXiv 2025

  4. [4]

    Zwicker, and H

    Wei Chen, Liu Ren, M. Zwicker, and H. Pfister. Hardware- accelerated adaptive ewa volume splatting. InIEEE Visual- ization 2004, pages 67–74, 2004. 1

    work page 2004

  5. [5]

    S-nerf++: Autonomous driving simulation via neural reconstruction and generation,

    Yurui Chen, Junge Zhang, Ziyang Xie, Wenye Li, Feihu Zhang, Jiachen Lu, and Li Zhang. S-nerf++: Autonomous driving simulation via neural reconstruction and generation,

    work page

  6. [6]

    Don’t splat your Gaussians: V olumetric ray-traced primitives for mod- eling and rendering scattering and emissive media.ACM Trans

    Jorge Condor, Sebastien Speierer, Lukas Bode, Aljaz Bozic, Simon Green, Piotr Didyk, and Adrian Jarabo. Don’t splat your Gaussians: V olumetric ray-traced primitives for mod- eling and rendering scattering and emissive media.ACM Trans. Graph., 44(1), 2025. 1, 2

    work page 2025

  7. [7]

    Robbersmyr, and Kris- tian Muri Knausg ˚ard

    Anurag Dalal, Daniel Hagen, Kjell G. Robbersmyr, and Kris- tian Muri Knausg ˚ard. Gaussian splatting: 3d reconstruc- tion and novel view synthesis: A review.IEEE Access, 12: 96797–96820, 2024. 2

    work page 2024

  8. [8]

    The light stages and their applications to pho- toreal digital actors

    Paul Debevec. The light stages and their applications to pho- toreal digital actors. 2012. 5

    work page 2012

  9. [9]

    Paul E. Debevec. Rendering synthetic objects into real scenes: Bridging traditional and image-based graphics with global illumination and high dynamic range photography. In Proceedings of ACM SIGGRAPH ’98, page 189–198, Or- lando, FL, USA, 1998. ACM. 2

    work page 1998

  10. [10]

    Jan-Niklas Dihlmann, Andreas Engelhardt, Arjun Majum- dar, Raphael Braun, and Hendrik P.A. Lensch. Subsurface scattering for 3d Gaussian splatting. InNeurIPS, 2024. 2

    work page 2024

  11. [11]

    OpenNeRF: Open Set 3D Neural Scene Segmentation with Pixel-Wise Features and Rendered Novel Views

    Francis Engelmann, Fabian Manhardt, Michael Niemeyer, Keisuke Tateno, Marc Pollefeys, and Federico Tombari. OpenNeRF: Open Set 3D Neural Scene Segmentation with Pixel-Wise Features and Rendered Novel Views. InInterna- tional Conference on Learning Representations, 2024. 3

    work page 2024

  12. [12]

    Filip, M

    J. Filip, M. Kolafov ´a, M. Havl ´ıˇcek, R. V ´avra, M. Haindl, and Rushmeier H. Evaluating Physical and Rendered Mate- rial Appearance.The Visual Computer (Computer Graphics International 2018), 2018. 5

    work page 2018

  13. [13]

    Relightable 3d Gaussians: Re- alistic point cloud relighting with BRDF decomposition and ray tracing

    Jian Gao, Chun Gu, Youtian Lin, Zhihao Li, Hao Zhu, Xun Cao, Li Zhang, and Yao Yao. Relightable 3d Gaussians: Re- alistic point cloud relighting with BRDF decomposition and ray tracing. InECCV, pages 73–89, 2024. 2

    work page 2024

  14. [14]

    Semantic Gaussians: Open-vocabulary scene understanding with 3d Gaussian splatting, 2024

    Jun Guo, Xiaojian Ma, Yue Fan, Huaping Liu, and Qing Li. Semantic Gaussians: Open-vocabulary scene understanding with 3d Gaussian splatting, 2024. 3

    work page 2024

  15. [15]

    A scene graph backed approach to open set semantic mapping, 2026

    Martin G ¨unther, Felix Igelbrink, Oscar Lima, Lennart Niecksch, Marian Renz, and Martin Atzmueller. A scene graph backed approach to open set semantic mapping, 2026. 3

    work page 2026

  16. [16]

    Semantic abstraction: Open- world 3D scene understanding from 2D vision-language models

    Huy Ha and Shuran Song. Semantic abstraction: Open- world 3D scene understanding from 2D vision-language models. InProceedings of the 2022 Conference on Robot Learning, 2022. 8

    work page 2022

  17. [17]

    Efficient perspective-correct 3d Gaussian splatting using hybrid trans- parency.Computer Graphics Forum, 44(2), 2025

    Florian Hahlbohm, Fabian Friederichs, Tim Weyrich, Li- nus Franke, Moritz Kappel, Susana Castillo, Marc Stam- minger, Martin Eisemann, and Marcus Magnor. Efficient perspective-correct 3d Gaussian splatting using hybrid trans- parency.Computer Graphics Forum, 44(2), 2025. 2

    work page 2025

  18. [18]

    3DGS-LM: Faster Gaussian-splatting optimization with Levenberg-Marquardt

    Lukas H ¨ollein, Aljaˇz Boˇziˇc, Michael Zollh¨ofer, and Matthias Nießner. 3DGS-LM: Faster Gaussian-splatting optimization with Levenberg-Marquardt. InICCV, 2025. 1

    work page 2025

  19. [19]

    2d Gaussian splatting for geometrically ac- curate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d Gaussian splatting for geometrically ac- curate radiance fields. InSIGGRAPH 2024 Conference Pa- pers. Association for Computing Machinery, 2024. 1, 2, 3

    work page 2024

  20. [20]

    Textured-GS: Gaussian splatting with spatially defined color and opacity, 2024

    Zhentao Huang and Minglun Gong. Textured-GS: Gaussian splatting with spatially defined color and opacity, 2024. 2

    work page 2024

  21. [21]

    DISC: Dense integrated semantic con- text for large-scale open-set semantic mapping, 2026

    Felix Igelbrink, Lennart Niecksch, Martin Atzmueller, and Joachim Hertzberg. DISC: Dense integrated semantic con- text for large-scale open-set semantic mapping, 2026. 3

    work page 2026

  22. [22]

    Lierex: Language-image embeddings for robotic exploration.KI - K ¨unstliche Intelli- genz, pages 1610–1987, 2026

    Felix Igelbrink, Lennart Niecksch, Marian Renz, Martin G¨unther, and Martin Atzmueller. Lierex: Language-image embeddings for robotic exploration.KI - K ¨unstliche Intelli- genz, pages 1610–1987, 2026. 8

    work page 1987

  23. [23]

    Mitsuba 3 renderer, 2022

    Wenzel Jakob, S ´ebastien Speierer, Nicolas Roussel, Merlin Nimier-David, Delio Vicini, Tizian Zeltner, Baptiste Nicolet, Miguel Crespo, Vincent Leroy, and Ziyi Zhang. Mitsuba 3 renderer, 2022. https://mitsuba-renderer.org. 1

    work page 2022

  24. [24]

    Tenenbaum, Celso Miguel de Melo, Madhava Krishna, Liam Paull, Florian Shkurti, and Antonio Torralba

    Krishna Murthy Jatavallabhula, Alihusein Kuwajerwala, Qiao Gu, Mohd Omama, Tao Chen, Shuang Li, Ganesh Iyer, Soroush Saryazdi, Nikhil Keetha, Ayush Tewari, Joshua B. Tenenbaum, Celso Miguel de Melo, Madhava Krishna, Liam Paull, Florian Shkurti, and Antonio Torralba. ConceptFu- sion: Open-set multimodal 3d mapping.Robotics: Science and Systems, 2023. 3, 7

    work page 2023

  25. [25]

    Rt-gs2: Real-time generalizable seman- tic segmentation for 3d gaussian representations of radiance fields, 2024

    Mihnea-Bogdan Jurca, Remco Royen, Ion Giosan, and Adrian Munteanu. Rt-gs2: Real-time generalizable seman- tic segmentation for 3d gaussian representations of radiance fields, 2024. 3

    work page 2024

  26. [26]

    LumiGauss: Relightable Gaussian splatting in the wild

    Joanna Kaleta, Kacper Kania, Tomasz Trzci ´nski, and Marek Kowalski. LumiGauss: Relightable Gaussian splatting in the wild. InProceedings of WACV 2025, 2025. 2

    work page 2025

  27. [27]

    3d Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics, 42 (4), 2023

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d Gaussian splatting for real-time radiance field rendering.ACM Transactions on Graphics, 42 (4), 2023. 1, 2, 3, 5

    work page 2023

  28. [28]

    LERF: Language embed- ded radiance fields

    Justin* Kerr, Chung Min* Kim, Ken Goldberg, Angjoo Kanazawa, and Matthew Tancik. LERF: Language embed- ded radiance fields. InICCV, 2023. 3, 7

    work page 2023

  29. [29]

    Robot see robot do: Imitating articulated object manipulation with 9 monocular 4d reconstruction

    Justin Kerr, Chung Min Kim, Mingxuan Wu, Brent Yi, Qian- qian Wang, Ken Goldberg, and Angjoo Kanazawa. Robot see robot do: Imitating articulated object manipulation with 9 monocular 4d reconstruction. In8th Annual Conference on Robot Learning, 2024. 8

    work page 2024

  30. [30]

    GARField: Group anything with radiance fields

    Chung Min Kim, Mingxuan Wu, Justin Kerr, Matthew Tan- cik, Ken Goldberg, and Angjoo Kanazawa. GARField: Group anything with radiance fields. InConference on Com- puter Vision and Pattern Recognition (CVPR), 2024. 3, 7

    work page 2024

  31. [31]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization, 2017. 1, 6

    work page 2017

  32. [32]

    Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross Girshick

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C. Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross Girshick. Segment anything. InICCV, pages 4015– 4026, 2023. 3

    work page 2023

  33. [33]

    OMCL: Open-vocabulary monte carlo localization

    Evgenii Kruzhkov, Raphael Memmesheimer, and Sven Behnke. OMCL: Open-vocabulary monte carlo localization. IEEE Robotics and Automation Letters, 2026. 8

    work page 2026

  34. [34]

    Modular primitives for high-performance differentiable rendering, 2020

    Samuli Laine, Janne Hellsten, Tero Karras, Yeongho Seol, Jaakko Lehtinen, and Timo Aila. Modular primitives for high-performance differentiable rendering, 2020. 1

    work page 2020

  35. [35]

    Soft ras- terizer: A differentiable renderer for image-based 3d reason- ing.ICCV, 2019

    Shichen Liu, Tianye Li, Weikai Chen, and Hao Li. Soft ras- terizer: A differentiable renderer for image-based 3d reason- ing.ICCV, 2019. 1

    work page 2019

  36. [36]

    A survey of 3d reconstruction: The evolution from multi- view geometry to nerf and 3dgs.Sensors, 25(18):5748, 2025

    Shuai Liu, Mengmeng Yang, Tingyan Xing, and Ran Yang. A survey of 3d reconstruction: The evolution from multi- view geometry to nerf and 3dgs.Sensors, 25(18):5748, 2025. 1

    work page 2025

  37. [37]

    Decoupled Weight Decay Regularization

    Ilya Loshchilov and Frank Hutter. Fixing weight decay reg- ularization in adam.CoRR, abs/1711.05101, 2017. 6

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  38. [38]

    Clio: Real-time task- driven open-set 3d scene graphs.IEEE Robotics and Au- tomation Letters, 9(10):8921–8928, 2024

    Dominic Maggio, Yun Chang, Nathan Hughes, Matthew Trang, Dan Griffith, Carlyn Dougherty, Eric Cristofalo, Lukas Schmid, and Luca Carlone. Clio: Real-time task- driven open-set 3d scene graphs.IEEE Robotics and Au- tomation Letters, 9(10):8921–8928, 2024. 3, 7

    work page 2024

  39. [39]

    Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar

    Ben Mildenhall, Pratul P. Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar. Local light field fusion: Practical view syn- thesis with prescriptive sampling guidelines.ACM Transac- tions on Graphics (TOG), 2019. 8

    work page 2019

  40. [40]

    Srinivasan, Matthew Tancik, Jonathan T

    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. InECCV, 2020. 1

    work page 2020

  41. [41]

    Bench- marking neural radiance fields for autonomous robots: An overview.Engineering Applications of Artificial Intelli- gence, 140:109685, 2025

    Yuhang Ming, Xingrui Yang, Weihan Wang, Zheng Chen, Jinglun Feng, Yifan Xing, and Guofeng Zhang. Bench- marking neural radiance fields for autonomous robots: An overview.Engineering Applications of Artificial Intelli- gence, 140:109685, 2025. 3

    work page 2025

  42. [42]

    3d gaussian ray trac- ing: Fast tracing of particle scenes.ACM Trans

    Nicolas Moenne-Loccoz, Ashkan Mirzaei, Or Perel, Ric- cardo de Lutio, Janick Martinez Esturo, Gavriel State, Sanja Fidler, Nicholas Sharp, and Zan Gojcic. 3d gaussian ray trac- ing: Fast tracing of particle scenes.ACM Trans. Graph., 43 (6), 2024. 2

    work page 2024

  43. [43]

    Instant neural graphics primitives with a multires- olution hash encoding.ACM Trans

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a multires- olution hash encoding.ACM Trans. Graph., 41(4):102:1– 102:15, 2022. 1

    work page 2022

  44. [44]

    DINOv2: Learning Robust Visual Features without Supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193, 2023. 1, 2, 7

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  45. [45]

    Demonstrating ok-robot: What really matters in integrating open-knowledge models for robotics.Robotics: Science and Systems, 2024

    Liu Peiqi, Orru Yaswanth, Vakil Jay, Paxton Chris, Shafiul- lah Nur, and Pinto Lerrel. Demonstrating ok-robot: What really matters in integrating open-knowledge models for robotics.Robotics: Science and Systems, 2024. 8

    work page 2024

  46. [46]

    OpenScene: 3d scene understanding with open vocabularies

    Songyou Peng, Kyle Genova, Chiyu ”Max” Jiang, An- drea Tagliasacchi, Marc Pollefeys, and Thomas Funkhouser. OpenScene: 3d scene understanding with open vocabularies. InCVPR, 2023. 3, 7

    work page 2023

  47. [47]

    SAM 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Dollar, and Christoph Feicht- enhofer. SAM 2: Segment anything in images and videos. In ICLR, 2025. 1, 2, 3, 7

    work page 2025

  48. [48]

    A versatile scene model with differentiable visibility applied to generative pose esti- mation

    Helge Rhodin, Nadia Robertini, Christian Richardt, Hans- Peter Seidel, and Christian Theobalt. A versatile scene model with differentiable visibility applied to generative pose esti- mation. InICCV, 2015. 1

    work page 2015

  49. [49]

    M. Rump, R. Sarlette, and R. Klein. Groundtruth data for multispectral bidirectional texture functions. InCGIV 2010, pages 326–330. Society for Imaging Science and Technol- ogy, 2010. 5

    work page 2010

  50. [50]

    Structure-from-motion revisited

    Johannes Lutz Sch ¨onberger and Jan-Michael Frahm. Structure-from-motion revisited. InCVPR, 2016. 2

    work page 2016

  51. [51]

    Pixelwise view selection for un- structured multi-view stereo

    Johannes Lutz Sch ¨onberger, Enliang Zheng, Marc Pollefeys, and Jan-Michael Frahm. Pixelwise view selection for un- structured multi-view stereo. InECCV, 2016. 2

    work page 2016

  52. [52]

    CLIP-fields: Weakly supervised semantic fields for robotic memory

    Nur Muhammad Mahi Shafiullah, Chris Paxton, Lerrel Pinto, Soumith Chintala, and Arthur Szlam. CLIP-fields: Weakly supervised semantic fields for robotic memory. Robotics: Science and Systems, 2023. 3, 7

    work page 2023

  53. [53]

    Barron, and Ravi Ramamoorthi

    Tiancheng Sun, Zexiang Xu, Xiuming Zhang, Sean Fanello, Christoph Rhemann, Paul Debevec, Yun-Ta Tsai, Jonathan T. Barron, and Ravi Ramamoorthi. Light stage super- resolution: continuous high-frequency relighting.ACM Trans. Graph., 39(6), 2020. 2, 5

    work page 2020

  54. [54]

    arXiv preprint arXiv:2505.21502 , year=

    Yipengjing Sun, Chenyang Wang, Shunyuan Zheng, Zonglin Li, Shengping Zhang, and Xiangyang Ji. Generalizable and relightable Gaussian splatting for human novel view synthe- sis.arXiv, 2025. arXiv:2505.21502. 2

    work page arXiv 2025

  55. [55]

    Solutions of ill posed prob- lems

    Andrey Nikolayevich Tikhonov. Solutions of ill posed prob- lems. 1977. 4

    work page 1977

  56. [56]

    3d Gaussian splatting with normal-involved rendering

    Meng Wei, Qianyi Wu, Jianmin Zheng, Hamid Rezatofighi, and Jianfei Cai. 3d Gaussian splatting with normal-involved rendering. InNeurIPS, 2024. 2

    work page 2024

  57. [57]

    Splatting: a parallel, feed-forward volume rendering algorithm

    Lee Westover. Splatting: a parallel, feed-forward volume rendering algorithm. 1991. 1

    work page 1991

  58. [58]

    SuperGaussians: Enhancing Gaussian splat- ting using primitives with spatially varying colors.arXiv,

    Rui Xu, Wenyue Chen, Jiepeng Wang, Yuan Liu, Peng Wang, Lin Gao, Shiqing Xin, Taku Komura, Xin Li, and Wenping Wang. SuperGaussians: Enhancing Gaussian splat- ting using primitives with spatially varying colors.arXiv,

    work page

  59. [59]

    Spec-Gaussian: Anisotropic view-dependent ap- pearance for 3d Gaussian splatting.NeurIPS, 2024

    Ziyi Yang, Xinyu Gao, Yangtian Sun, Yihua Huang, Xi- aoyang Lyu, Wen Zhou, Shaohui Jiao, Xiaojuan Qi, and Xi- aogang Jin. Spec-Gaussian: Anisotropic view-dependent ap- pearance for 3d Gaussian splatting.NeurIPS, 2024. 2

    work page 2024

  60. [60]

    gsplat: An open-source library for Gaussian splatting.Journal of Ma- chine Learning Research, 26(34):1–17, 2025

    Vickie Ye, Ruilong Li, Justin Kerr, Matias Turkulainen, Brent Yi, Zhuoyang Pan, Otto Seiskari, Jianbo Ye, Jeffrey Hu, Matthew Tancik, and Angjoo Kanazawa. gsplat: An open-source library for Gaussian splatting.Journal of Ma- chine Learning Research, 26(34):1–17, 2025. 1, 3

    work page 2025

  61. [61]

    Ewa volume splatting

    Matthias Zwicker, Hanspeter Pfister, Jeroen van Baar, and Markus Gross. Ewa volume splatting. InProceedings of the Conference on Visualization ’01, page 29–36, USA, 2001. IEEE Computer Society. 1 11

    work page 2001