Pith. sign in

REVIEW 42 references

Proxy-GS: Unified Occlusion Priors for Training and Inference in Structured 3D Gaussian Splatting

T0 review · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read A proxy mesh rendered through hardware rasterization provides a cheap occlusion depth prior that culls hidden anchors at inference and guides densification at training, giving Octree-GS-like MLP splatting a 3 to 4x speedup on occluded city scenes with equal or slightly better quality.

arxiv 2509.24421 v5 pith:6UAS3F34 submitted 2025-09-29 cs.CV

Yuanyuan Gao , Yuning Gong , Yifei Liu , Li Jingfeng , Dingwen Zhang , Yanci Zhang , Dan Xu , Xiao Sun
show 1 more author
Zhihang Zhong
This is my paper
classification cs.CV
keywords renderinggaussianocclusionproxyproxy-gssplattingachievesapproach
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

MLP-based 3D Gaussian Splatting, such as Octree-GS, places thousands of anchor points in a scene and decodes each one through a neural network to create the Gaussians that get drawn. In a city street, many of those anchors sit inside buildings or on walls facing away from the camera, so the decoder wastes time computing things nobody can see. Earlier speed tricks, like pruning and level-of-detail, cut costs but tend to hurt quality.

Proxy-GS builds a rough triangle mesh of the scene from whatever geometry is available: sensor depth, a monocular depth network, or a prior reconstruction. It renders this mesh with the GPU's fixed-function rasterizer, which outputs a depth map in about a millisecond. The depth map is used twice. At inference, anchors whose depth lies behind the mesh surface are culled in the same kernel that does frustum culling. During training, image patches with persistently high error are selected, and new anchors are placed by back-projecting through the proxy depth, so densification targets visible surfaces instead of occluded space.

On the heavily occluded MatrixCity streets, Proxy-GS renders at 126 to 151 FPS versus 30 to 48 FPS for Octree-GS, with PSNR gains of 0.1 to 0.3 dB. On less occluded scenes, the speed advantage shrinks. The quality gain is modest and reported without error bars, and the code is promised but not released, so the timing and FPS claims cannot currently be checked independently. The method also inherits whatever errors the proxy mesh has; robustness to bad meshes is asserted, not measured.

Extended reading notes

Core claim

Central claim (Abstract): 'Proxy-GS... achieves more than 2.5x speedup over Octree-GS, and consistently delivers substantially higher rendering quality.' Mechanistically: proxy depth maps from hardware rasterization (claimed under 1 ms at 1000x1000) cull occluded anchors at inference (Eq. 9) and, with proxy-guided densification (Section 4.3), keep quality while reducing decoded anchors from roughly 800k to 80k-190k (Table 9). If the paper is correct, MLP-based 3DGS runs 3 to 4 times faster on occlusion-rich scenes at equal or better quality than Octree-GS.

Load-bearing premise

The load-bearing premise is that an approximately correct proxy mesh is available for the scene, and that its errors do not break the depth test. Eq. 9 culls every anchor whose center lies behind the proxy depth at its pixel, and Section 4.3 places new anchors by back-projecting through the same depth; any hole, misplaced wall, or moving object in the proxy directly corrupts both. Robustness is asserted only in Appendix A.4.4 ('Our method does not require highly accurate meshes; an approximate geometry is sufficient') with no ablation degrading mesh quality, decimation level, or completeness. For scenes without ground-truth depth or LiDAR, the mesh comes from MoGe2+PGSR (indoor) or the authors' own CityGS-X pipeline (outdoor), so practical applicability inherits those tools' reliability.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard graphics results (Z-buffering, QEM simplification, Hi-Z culling) plus one domain assumption that is asserted rather than measured: the proxy mesh's depth is a valid visibility prior. The method introduces no new physical or representational entities. Two hyperparameters (gamma, tau) are chosen empirically, gamma on a single dataset, and the densification grid parameters h and K are undisclosed. The heaviest unpaid debt is the robustness of Eq. 9 and the Section 4.3 back-projection to mesh error.

free parameters (4)
  • Safety margin gamma (Eq. 8) = 0.3
    Tuned on the real Small City dataset (Table 3: 0.1, 0.3, 0.6, 1.0; 0.3 best) and applied to all other scenes without per-scene adjustment.
  • Patch selection threshold tau = 3 * mean patch loss = 3x mean patch loss
    Chosen multiplier for selecting high-error patches in proxy-guided densification (Section 4.3); no ablation or derivation reported.
  • Proxy-grid cell size h and capacity K = not disclosed
    Grid parameters limiting anchor growth in Section 4.3 ('cell size h and origin b_min, allow up to K anchors per cell'); values never given, affecting density control and reproducibility.
  • Hi-Z level snapping constant c and padding Delta = c in {1,2}, Delta in {0,1}
    Engineering constants in Appendix A.5.2 controlling conservativeness of cluster-level culling; chosen by hand, minor impact.
assumptions (5)
  • domain assumption Proxy mesh approximates true scene geometry closely enough that its depth is a valid occlusion prior
    Used in Eq. 9 (culling) and Section 4.3 (densification back-projection). The paper asserts robustness qualitatively in Appendix A.4.4 but runs no accuracy-degradation ablation.
  • standard math Hardware Z-buffer from a depth-only pass is conservative and correct for visibility
    Eqs. 2-9 assume standard perspective projection and z-buffer semantics; standard graphics, but the near-plane and precision choices (tau=1e-4, epsilon=1e-7) are unvalidated.
  • ad hoc to paper High-error patches (patch loss > 3 * mean) correspond to geometrically meaningful regions suitable for anchoring on the proxy surface
    Section 4.3 heuristic linking rendering error to occlusion; no evidence that error correlates with proxy-surface proximity rather than appearance or view-dependent effects.
  • domain assumption Octree-GS LOD and anchor initialization are a sound base, and re-implemented baselines are configured fairly
    Section 5.1: baselines re-trained with densification threshold 1e-4; FPS comparisons rest on an under-specified render-resolution and measurement protocol.
  • domain assumption The scene is static; proxy depth does not capture dynamic objects
    Real datasets (Berlin, Small City) contain moving vehicles and pedestrians; a static proxy could mis-cull their Gaussians, a concern the paper does not address.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Proxy-GS: Unified Occlusion Priors for Training and Inference in Structured 3D Gaussian Splatting." pith.science (2026). https://pith.science/paper/6UAS3F34

@misc{pith2026250924421,
  author       = {Pith},
  title        = {Pith review of: Proxy-GS: Unified Occlusion Priors for Training and Inference in Structured 3D Gaussian Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6UAS3F34}},
  note         = {Machine review of arXiv:2509.24421}
}
read the original abstract

3D Gaussian Splatting (3DGS) has emerged as an efficient approach for achieving photorealistic rendering. Recent MLP-based variants further improve visual fidelity but introduce substantial decoding overhead during rendering. To alleviate computation cost, several pruning strategies and level-of-detail (LOD) techniques have been introduced, aiming to effectively reduce the number of Gaussian primitives in large-scale scenes. However, our analysis reveals that significant redundancy still remains due to the lack of occlusion awareness. In this work, we propose Proxy-GS, a novel pipeline that exploits a proxy to introduce Gaussian occlusion awareness from any view. At the core of our approach is a fast proxy system capable of producing precise occlusion depth maps at a resolution of 1000x1000 under 1ms. This proxy serves two roles: first, it guides the culling of anchors and Gaussians to accelerate rendering speed. Second, it guides the densification towards surfaces during training, avoiding inconsistencies in occluded regions, and improving the rendering quality. In heavily occluded scenarios, such as the MatrixCity Streets dataset, Proxy-GS not only equips MLP-based Gaussian splatting with stronger rendering capability but also achieves faster rendering speed. Specifically, it achieves more than 2.5x speedup over Octree-GS, and consistently delivers substantially higher rendering quality. Code will be public upon acceptance.

Figures

Figures reproduced from arXiv: 2509.24421 by the authors.

Figure 1
Figure 1. We propose Proxy-GS, an occlusion-aware training and inference framework built upon lightweight proxies. By introducing proxy-guided densification, our method effectively guides an￾chors to grow in more geometrically meaningful regions. As a result, Proxy-GS not only achieves higher rendering quality but also delivers significantly faster rendering compared to state-of-the-art MLP-based 3DGS approaches. ABSTRACT 3D … view at source ↗
Figure 2
Figure 2. Proxy-GS Framework. We first construct a lightweight proxy mesh. During rendering, hardware rasterization produces a depth map in under 1 ms, which is then used to efficiently cull anchors that are occluded. During training, in addition to the same rendering pipeline, we further introduce structure-aware anchor densification, encouraging anchors to grow adaptively along the proxy mesh geometry. their anchor selectio… view at source ↗
Figure 3
Figure 3. Comparison of the time propor￾tion of each inference component (Render￾ing, anchor filter, depth rendering) with that of Octree-GS. In the original anchor-growing densification strategy, new anchors are generated around Gaussian splats that exhibit large gradients during training. However, this procedure may introduce redundant anchors behind the proxy mesh depth: although these Gaussians have large gradients, the n… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison. Visualization on different datasets (Li et al., 2023; Xiong et al., 2024; Barron et al., 2023; Kerbl et al., 2024). of Hierarchical-GS, we set the τ1, τ2, τ3 = 3, 6, 15. For approaches that do not employ MLPs, such as 3DGS and Hierarchical-GS, t…
Figure 5
Figure 5. Figure 5: Visualization on different safety margins. Safety margin of the occlusion culling. In the Ap￾pendix A.2, we report all the results with the hardware 3DGS as the default renderer Rendering time analysis. In [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Mesh visualization. Scenes include different datasets (Li et al., 2023; Xiong et al., 2024; Barron et al., 2023; Kerbl et al., 2024). A.4.3 OUTDOOR SCENES WITH SPARSE COLMAP POINT CLOUDS For outdoor environments where the reconstruction relies solely on sparse COLMAP p…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

42 extracted references · 15 linked inside Pith

  1. [1]

    GitHub, 2024

    Fast gaussian rasterization. GitHub, 2024. URL https://github.com/dendenxu/fast-gaussian-rasterization

    2024

  2. [2]

    Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Matthew Tancik, Peter Hedman, Ricardo Martin-Brualla, and Pratul P Srinivasan. Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields. In Proceedings of the IEEE/CVF international conference on computer vision, pp.\ 5855--5864, 2021

    2021

  3. [3]

    Barron, Ben Mildenhall, Dor Verbin, Pratul P

    Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, and Peter Hedman. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. CVPR, 2022

    2022

  4. [4]

    Zip-nerf: Anti-aliased grid-based neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. Zip-nerf: Anti-aliased grid-based neural radiance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 19697--19705, 2023

    2023

  5. [5]

    Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction

    Danpeng Chen, Hai Li, Weicai Ye, Yifan Wang, Weijian Xie, Shangjin Zhai, Nan Wang, Haomin Liu, Hujun Bao, and Guofeng Zhang. Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction. arXiv preprint arXiv:2406.06521, 2024

    arXiv 2024

  6. [6]

    Letsgo: Large-scale garage modeling and rendering via lidar-assisted gaussian primitives

    Jiadi Cui, Junming Cao, Fuqiang Zhao, Zhipeng He, Yifan Chen, Yuhui Zhong, Lan Xu, Yujiao Shi, Yingliang Zhang, and Jingyi Yu. Letsgo: Large-scale garage modeling and rendering via lidar-assisted gaussian primitives. ACM Transactions on Graphics (TOG), 43 0 (6): 0 1--18, 2024

    2024

  7. [7]

    Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps

    Zhiwen Fan, Kevin Wang, Kairun Wen, Zehao Zhu, Dejia Xu, Zhangyang Wang, et al. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps. Advances in neural information processing systems, 37: 0 140138--140158, 2024

    2024

  8. [8]

    Flashgs: Efficient 3d gaussian splatting for large-scale and high-resolution rendering

    Guofeng Feng, Siyan Chen, Rong Fu, Zimu Liao, Yi Wang, Tao Liu, Zhiling Pei, Hengjie Li, Xingcheng Zhang, and Bo Dai. Flashgs: Efficient 3d gaussian splatting for large-scale and high-resolution rendering. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 26652--26662, 2024. URL https://api.semanticscholar.org/CorpusID:271874687

    2025

Show all 42 references
  1. [9]

    Cosurfgs: Collaborative 3d surface gaussian splatting with distributed learning for large scene reconstruction

    Yuanyuan Gao, Yalun Dai, Hao Li, Weicai Ye, Junyi Chen, Danpeng Chen, Dingwen Zhang, Tong He, Guofeng Zhang, and Junwei Han. Cosurfgs: Collaborative 3d surface gaussian splatting with distributed learning for large scene reconstruction. arXiv preprint arXiv:2412.17612, 2024

    2024 arXiv

  2. [10]

    Citygs-x: A scalable architecture for efficient and geometrically accurate large-scale scene reconstruction

    Yuanyuan Gao, Hao Li, Jiaqi Chen, Zhengyu Zou, Zhihang Zhong, Dingwen Zhang, Xiao Sun, and Junwei Han. Citygs-x: A scalable architecture for efficient and geometrically accurate large-scale scene reconstruction. arXiv preprint arXiv:2503.23044, 2025

    2025 arXiv

  3. [11]

    Hierarchical z-buffer visibility

    Ned Greene, Michael Kass, and Gavin Miller. Hierarchical z-buffer visibility. In Proceedings of the 20th annual conference on Computer graphics and interactive techniques, pp.\ 231--238, 1993

    1993

  4. [12]

    Neural kernel surface reconstruction

    Jiahui Huang, Zan Gojcic, Matan Atzmon, Or Litany, Sanja Fidler, and Francis Williams. Neural kernel surface reconstruction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 4369--4379, 2023

    2023

  5. [13]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk \"u hler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42 0 (4): 0 139--1, 2023 a

    2023

  6. [14]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk \"u hler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics, July 2023 b

    2023

  7. [15]

    A hierarchical 3d gaussian representation for real-time rendering of very large datasets

    Bernhard Kerbl, Andreas Meuleman, Georgios Kopanas, Michael Wimmer, Alexandre Lanvin, and George Drettakis. A hierarchical 3d gaussian representation for real-time rendering of very large datasets. ACM Transactions on Graphics (TOG), 43 0 (4): 0 1--15, 2024

    2024

  8. [16]

    Compact 3d gaussian representation for radiance field

    Joo Chan Lee, Daniel Rho, Xiangyu Sun, Jong Hwan Ko, and Eunbyung Park. Compact 3d gaussian representation for radiance field. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 21719--21728, 2024

    2024

  9. [17]

    Dgtr: Distributed gaussian turbo-reconstruction for sparse-view vast scenes

    Hao Li, Yuanyuan Gao, Haosong Peng, Chenming Wu, Weicai Ye, Yufeng Zhan, Chen Zhao, Dingwen Zhang, Jingdong Wang, and Junwei Han. Dgtr: Distributed gaussian turbo-reconstruction for sparse-view vast scenes. arXiv preprint arXiv:2411.12309, 2024 a

    2024 arXiv

  10. [18]

    Matrixcity: A large-scale city dataset for city-scale neural rendering and beyond

    Yixuan Li, Lihan Jiang, Linning Xu, Yuanbo Xiangli, Zhenzhi Wang, Dahua Lin, and Bo Dai. Matrixcity: A large-scale city dataset for city-scale neural rendering and beyond. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 3205--3215, 2023

    2023

  11. [19]

    Garc \'i a-Fern \'a ndez, Yong Yue, Eng Gee Lim, and Xiaohui Zhu

    Zhuoxiao Li, Shanliang Yao, Yijie Chu, \'A ngel F. Garc \'i a-Fern \'a ndez, Yong Yue, Eng Gee Lim, and Xiaohui Zhu. Mvg-splatting: Multi-view guided gaussian splatting with adaptive quantile-based geometric consistency densification. ArXiv, abs/2407.11840, 2024 b . URL https:...

    2024 arXiv

  12. [20]

    Vastgaussian: Vast 3d gaussians for large scene reconstruction

    Jiaqi Lin, Zhihao Li, Xiao Tang, Jianzhuang Liu, Shiyong Liu, Jiayue Liu, Yangdi Lu, Xiaofei Wu, Songcen Xu, Youliang Yan, et al. Vastgaussian: Vast 3d gaussians for large scene reconstruction. arXiv preprint arXiv:2402.17427, 2024

    2024 arXiv

  13. [21]

    Citygaussian: Real-time high-quality large-scale scene rendering with gaussians

    Yang Liu, Chuanchen Luo, Lue Fan, Naiyan Wang, Junran Peng, and Zhaoxiang Zhang. Citygaussian: Real-time high-quality large-scale scene rendering with gaussians. In European Conference on Computer Vision, pp.\ 265--282. Springer, 2024

    2024

  14. [22]

    Maskgaussian: Adaptive 3d gaussian representation from probabilistic masks

    Yifei Liu, Zhihang Zhong, Yifan Zhan, Sheng Xu, and Xiao Sun. Maskgaussian: Adaptive 3d gaussian representation from probabilistic masks. In Proceedings of the Computer Vision and Pattern Recognition Conference, pp.\ 681--690, 2025

    2025

  15. [23]

    Scaffold-gs: Structured 3d gaussians for view-adaptive rendering

    Tao Lu, Mulin Yu, Linning Xu, Yuanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 20654--20664, 2024

    2024

  16. [24]

    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 synthesis. In ECCV, 2020

    2020

  17. [25]

    Anicrafter: Customizing realistic human-centric animation via avatar-background conditioning in video diffusion models

    Muyao Niu, Mingdeng Cao, Yifan Zhan, Qingtian Zhu, Mingze Ma, Jiancheng Zhao, Yanhong Zeng, Zhihang Zhong, Xiao Sun, and Yinqiang Zheng. Anicrafter: Customizing realistic human-centric animation via avatar-background conditioning in video diffusion models. arXiv preprint arXiv...

    2025 arXiv

  18. [26]

    Octree-gs: Towards consistent real-time rendering with lod-structured 3d gaussians

    Kerui Ren, Lihan Jiang, Tao Lu, Mulin Yu, Linning Xu, Zhangkai Ni, and Bo Dai. Octree-gs: Towards consistent real-time rendering with lod-structured 3d gaussians. arXiv preprint arXiv:2403.17898, 2024

    2024 arXiv

  19. [27]

    Ags-mesh: Adaptive gaussian splatting and meshing with geometric priors for indoor room reconstruction using smartphones

    Xuqian Ren, Matias Turkulainen, Jiepeng Wang, Otto Seiskari, Iaroslav Melekhov, Juho Kannala, and Esa Rahtu. Ags-mesh: Adaptive gaussian splatting and meshing with geometric priors for indoor room reconstruction using smartphones. In 2025 International Conference on 3D Vision ...

    2025

  20. [28]

    Gs-cache: A gs-cache inference framework for large-scale gaussian splatting models

    Miao Tao, Yuanzhen Zhou, Haoran Xu, Zeyu He, Zhenyu Yang, Yuchang Zhang, Zhongling Su, Linning Xu, Zhenxiang Ma, Rong Fu, Hengjie Li, Xingcheng Zhang, and Jidong Zhai. Gs-cache: A gs-cache inference framework for large-scale gaussian splatting models. ArXiv, abs/2502.14938, 20...

    2025 arXiv

  21. [29]

    Moge-2: Accurate monocular geometry with metric scale and sharp details

    Ruicheng Wang, Sicheng Xu, Yue Dong, Yu Deng, Jianfeng Xiang, Zelong Lv, Guangzhong Sun, Xin Tong, and Jiaolong Yang. Moge-2: Accurate monocular geometry with metric scale and sharp details. arXiv preprint arXiv:2507.02546, 2025

    2025 arXiv

  22. [30]

    Hyrf: Hybrid radiance fields for efficient and high-quality novel view synthesis

    Zipeng Wang and Dan Xu. Hyrf: Hybrid radiance fields for efficient and high-quality novel view synthesis. NeurIPS, 2025

    2025

  23. [31]

    Gauu-scene v2: Assessing the reliability of image-based metrics with expansive lidar image dataset using 3dgs and nerf

    Butian Xiong, Nanjun Zheng, Junhua Liu, and Zhen Li. Gauu-scene v2: Assessing the reliability of image-based metrics with expansive lidar image dataset using 3dgs and nerf. arXiv preprint arXiv:2404.04880, 2024

    2024 arXiv

  24. [32]

    Gast: Sequential gaussian avatars with hierarchical spatio-temporal context

    Wangze Xu, Yifan Zhan, Zhihang Zhong, and Xiao Sun. Gast: Sequential gaussian avatars with hierarchical spatio-temporal context. arXiv preprint arXiv:2411.16768, 2024

    2024 arXiv

  25. [33]

    When gaussian meets surfel: Ultra-fast high-fidelity radiance field rendering

    Keyang Ye, Tianjia Shao, and Kun Zhou. When gaussian meets surfel: Ultra-fast high-fidelity radiance field rendering. ACM Trans. Graph., 44: 0 113:1--113:15, 2025. URL https://api.semanticscholar.org/CorpusID:278032885

    2025

  26. [34]

    Tomie: Towards modular growth in enhanced smpl skeleton for 3d human with animatable garments

    Yifan Zhan, Qingtian Zhu, Muyao Niu, Mingze Ma, Jiancheng Zhao, Zhihang Zhong, Xiao Sun, Yu Qiao, and Yinqiang Zheng. Tomie: Towards modular growth in enhanced smpl skeleton for 3d human with animatable garments. arXiv preprint arXiv:2410.08082, 2024

    2024 arXiv

  27. [35]

    Nerf++: Analyzing and improving neural radiance fields

    Kai Zhang, Gernot Riegler, Noah Snavely, and Vladlen Koltun. Nerf++: Analyzing and improving neural radiance fields. arXiv preprint arXiv:2010.07492, 2020

    2010 arXiv

  28. [36]

    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 Proceedings of the IEEE conference on computer vision and pattern recognition, pp.\ 586--595, 2018

    2018

  29. [37]

    2dgs-room: Seed-guided 2d gaussian splatting with geometric constrains for high-fidelity indoor scene reconstruction

    Wanting Zhang, Haodong Xiang, Zhichao Liao, Xiansong Lai, Xinghui Li, and Long Zeng. 2dgs-room: Seed-guided 2d gaussian splatting with geometric constrains for high-fidelity indoor scene reconstruction. arXiv preprint arXiv:2412.03428, 2024

    2024 arXiv

  30. [38]

    On scaling up 3d gaussian splatting training

    Hexu Zhao, Haoyang Weng, Daohan Lu, Ang Li, Jinyang Li, Aurojit Panda, and Saining Xie. On scaling up 3d gaussian splatting training. In European Conference on Computer Vision, pp.\ 14--36. Springer, 2024

    2024

  31. [39]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  32. [40]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  33. [41]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  34. [42]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.