pith. sign in

arxiv: 2606.20842 · v1 · pith:NDKHZF6Dnew · submitted 2026-06-18 · 💻 cs.CV

From Uncertainty to Stability and Fidelity: Guiding Sparse-View 3D Gaussian Splatting with Fisher Information

Junbao Zhou , Qingshan Xu , Yuan Zhou , Xiaolong Shen , Beier Zhu , Kesen Zhao , Yiming Zeng , Chen Bai
show 2 more authors
Cheng Lu Hanwang Zhang
This is my paper

Pith reviewed 2026-06-26 18:08 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D Gaussian SplattingSparse-view novel view synthesisFisher InformationStereo augmentationUncertainty-aware regularizationOverfitting mitigationDepth priors
0
0 comments X

The pith

Fisher Information selects informative views and adapts regularization to stabilize sparse-view 3D Gaussian Splatting.

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

The paper shows that Fisher Information can replace random sampling when adding depth-prior views and when deciding which Gaussians to drop during training. This guided selection reduces the uncertainty that previously caused overfitting and artifacts in sparse-input novel view synthesis. A reader would care because the change produces higher-fidelity renderings from fewer captured images without adding new hand-crafted priors. The method therefore turns an ad-hoc regularization step into a quantitatively driven one. Experiments on standard benchmarks confirm the resulting improvement in stability and image quality.

Core claim

By computing Fisher Information to choose the most informative supporting viewpoints for stereo augmentation and to scale the dropout probability of each 3D Gaussian according to its uncertainty, the optimization avoids the compounded randomness of prior methods, yielding more stable training and higher rendering fidelity on sparse-view inputs.

What carries the argument

Fisher Information, used both to rank candidate viewpoints by expected information gain and to quantify per-Gaussian uncertainty for adaptive regularization.

If this is right

  • Stereo augmentation becomes deterministic rather than random, producing more consistent pseudo ground truth.
  • Dropout regularization becomes uncertainty-aware, removing high-uncertainty Gaussians more often and low-uncertainty ones less often.
  • Overfitting is reduced because both augmentation and regularization are now driven by the same information-theoretic signal.
  • The resulting renderings achieve state-of-the-art scores on existing sparse-view novel view synthesis benchmarks.

Where Pith is reading between the lines

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

  • The same Fisher-guided selection principle could be tested on other radiance-field representations that also suffer from sparse-view instability.
  • If the uncertainty estimates prove reliable, they might serve as a stopping criterion or active-learning signal for acquiring additional real views.
  • The approach implicitly treats Fisher Information as a cheap proxy for model sensitivity; verifying this proxy against explicit Hessian computations would strengthen the method.

Load-bearing premise

Fisher Information can be computed reliably inside the 3DGS optimization loop and will select views and dropout rates that genuinely reduce uncertainty rather than introduce new bias.

What would settle it

A controlled run on the same sparse-view benchmarks in which random view selection and uniform dropout probability produce equal or higher PSNR and SSIM than the Fisher-guided versions.

Figures

Figures reproduced from arXiv: 2606.20842 by Beier Zhu, Chen Bai, Cheng Lu, Hanwang Zhang, Junbao Zhou, Kesen Zhao, Qingshan Xu, Xiaolong Shen, Yiming Zeng, Yuan Zhou.

Figure 1
Figure 1. Figure 1: Comparison between our proposed strategies and traditional depth-based aug￾mentation and Dropout-based regularization for sparse-view 3DGS. σPSNR and µPSNR denote the standard deviation and the mean PSNR across multiple runs, respectively. Fig. (a) illustrates that curating pseudo ground truths by warping a single train￾ing view fails to fully utilize geometric priors, resulting noticeable hollow areas in … view at source ↗
Figure 2
Figure 2. Figure 2: Uniform dropping rate possi￾bly remove under-optimized Gaussians, while preserving over-optimized ones. However, 3DGS requires a large num￾ber of input views during training to achieve satisfactory rendering quality. When restricted to a limited number of input views (i.e. sparse-view setting) [53], 3DGS suffers from drastic performance degradation due to insufficient geometric and appearance information. … view at source ↗
Figure 3
Figure 3. Figure 3: Overall pipeline of our proposed method. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Stereo Augmentation with Fisher Information. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Uncertainty-aware Dropout. We se￾lectively drop surface (over-optimized) Gaussians, while keeping occluded (under-optimized) ones. Uncertainty-aware Dropout. We quantify the optimization status of each 3D Gaussian Gi using an uncertainty score Ui derived from Fisher Informa￾tion [18] (illustrated in [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Soft-scale Dropout. We maintain the presence of Gaussian by scaling down their opacity with factor λ. Soft-scale Dropout. Previous Dropout-based regularizations [38, 56] employ a hard removal strategy that zeros out the opacity of selected Gaussians. However, this removal triggers an over￾compensation effect (il￾lustrated in [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative ablation study on LLFF dataset in 3-view setting. 3DGS Cor-GS DropGaussian Ours Ground truth [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative comparisons with baseline methods on LLFF dataset (3-view) [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparisons with baseline methods on Mip-NeRF 360 dataset (12- view). References 1. Barron, J.T., Mildenhall, B., Tancik, M., Hedman, P., Martin-Brualla, R., Srini￾vasan, P.P.: 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) 2. Barron, J.T., Mildenhall, B., Verbin, D… view at source ↗
read the original abstract

3D Gaussian Splatting (3DGS) has emerged as a promising technique for novel view synthesis. However, 3DGS requires dense input views to achieve high-quality rendering. In sparse-view scenarios, 3DGS often prones to overfitting, resulting in noticeable artifacts and degraded rendering quality. Previous methods explore to address this issue by introducing additional priors (e.g. depth priors) or integrating regularization techniques (e.g. Dropout). However, these methods are often applied without principled guidance. In particular, prior-based augmentation typically samples novel viewpoints randomly, while Dropout-based regularization randomly removes Gaussians. The compounded randomness introduces uncertainty and instability, limiting the fidelity of novel view synthesis. In this paper, we propose a novel method for sparse-view 3DGS that incorporates Fisher Information to quantitatively guide the utilization of geometric priors and regularization. Specifically, our method comprises two key components: (1) Stereo augmentation with Fisher Information. By leveraging Fisher Information, we actively select most informative supporting views and use depth priors to curate reliable pseudo ground truths, which reduces randomness in augmentation and improves stability and rendering fidelity; (2) Uncertainty-aware regularization. We reduce the instability of Dropout-based regularization by using Fisher Information to quantitatively measure the uncertainty of each 3D Gaussian, and adaptively adjust the removal probability, leading to more stable and effective regularization. With these two components, our method effectively mitigates overfitting and improves the stability of optimization in sparse-view 3DGS, resulting in superior rendering fidelity. Extensive experiments show that our method achieves state-of-the-art performance in sparse-view novel view synthesis benchmarks.

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 / 2 minor

Summary. The paper claims that Fisher Information can be used to guide two components in sparse-view 3D Gaussian Splatting: (1) active selection of the most informative supporting views for stereo augmentation with depth priors, reducing randomness compared to prior random sampling, and (2) per-Gaussian uncertainty measurement to adaptively modulate dropout probability in regularization. These are asserted to mitigate overfitting, improve optimization stability, and yield superior rendering fidelity, with extensive experiments demonstrating state-of-the-art performance on sparse-view novel view synthesis benchmarks.

Significance. If the Fisher Information proxies are shown to reliably correlate with view informativeness and Gaussian sensitivity, the method would supply a principled, quantitative alternative to random augmentation and dropout in data-scarce 3DGS settings. This could meaningfully advance novel-view synthesis for applications where dense capture is impractical, while building on established information-theoretic tools without introducing new external priors.

major comments (2)
  1. [§3.1] §3.1 (Stereo augmentation with Fisher Information): The central claim that Fisher Information computed on the rendering loss ranks supporting views more reliably than random sampling requires explicit validation. The manuscript must include an ablation (e.g., Table comparing random vs. FI-selected views under identical depth priors) demonstrating lower final test loss or smaller train-test gap for the FI-guided selection; without this, the component reduces to guided random augmentation whose advantage is unproven.
  2. [§3.2] §3.2 (Uncertainty-aware regularization): The per-Gaussian uncertainty scalar derived from the empirical Fisher (almost always diagonal or block-diagonal in 3DGS due to alpha compositing) is used to modulate dropout probability. The paper must report quantitative evidence (e.g., ablation on the train-test PSNR gap and artifact metrics) that this adaptive scheme measurably outperforms standard random Dropout; otherwise the regularization benefit is not load-bearing for the stability claim.
minor comments (2)
  1. [Abstract] Abstract: 'prones to overfitting' should read 'is prone to overfitting'.
  2. [§3] Notation: clarify whether the Fisher Information is the full matrix, diagonal approximation, or empirical estimate, and specify the exact loss whose Hessian is approximated.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. The major comments correctly identify the need for component-specific ablations to substantiate the advantages of Fisher Information guidance. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [§3.1] §3.1 (Stereo augmentation with Fisher Information): The central claim that Fisher Information computed on the rendering loss ranks supporting views more reliably than random sampling requires explicit validation. The manuscript must include an ablation (e.g., Table comparing random vs. FI-selected views under identical depth priors) demonstrating lower final test loss or smaller train-test gap for the FI-guided selection; without this, the component reduces to guided random augmentation whose advantage is unproven.

    Authors: We agree that an explicit ablation isolating Fisher Information-guided view selection from random sampling (with fixed depth priors) is required to validate the ranking claim. The current manuscript reports overall benchmark gains but does not contain this isolated comparison. In the revised version we will add the requested table, reporting final test loss and train-test gap for both strategies. revision: yes

  2. Referee: [§3.2] §3.2 (Uncertainty-aware regularization): The per-Gaussian uncertainty scalar derived from the empirical Fisher (almost always diagonal or block-diagonal in 3DGS due to alpha compositing) is used to modulate dropout probability. The paper must report quantitative evidence (e.g., ablation on the train-test PSNR gap and artifact metrics) that this adaptive scheme measurably outperforms standard random Dropout; otherwise the regularization benefit is not load-bearing for the stability claim.

    Authors: We concur that a direct quantitative comparison of the uncertainty-modulated dropout against standard random dropout is necessary to establish the adaptive scheme's contribution. The manuscript emphasizes end-to-end improvements but omits this isolated ablation. We will include the requested metrics (train-test PSNR gap and artifact measures) in the revision. revision: yes

Circularity Check

0 steps flagged

No circularity: Fisher Information applied as external guidance without self-referential reduction

full rationale

The paper proposes two components that apply the established Fisher Information concept to guide view selection and per-Gaussian dropout in 3DGS. No equations, derivations, or claims in the abstract or described method reduce a 'prediction' or result to a fitted input or self-citation by construction. The central claims rest on empirical experiments rather than any load-bearing self-definition or imported uniqueness theorem. This is the common case of an applied method paper whose derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based on abstract only; no free parameters or invented entities explicitly identified. The approach relies on the domain assumption regarding the utility of Fisher Information in this setting.

axioms (1)
  • domain assumption Fisher Information provides a quantitative measure of uncertainty for 3D Gaussians and informativeness for novel views that can be used to guide augmentation and regularization effectively.
    This underpins both key components of the method.

pith-pipeline@v0.9.1-grok · 5856 in / 1310 out tokens · 30467 ms · 2026-06-26T18:08:19.062131+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

62 extracted references · 2 linked inside Pith

  1. [1]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Barron, J.T., Mildenhall, B., Tancik, M., Hedman, P., Martin-Brualla, R., Srini- vasan, P.P.: 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

  2. [2]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Barron, J.T., Mildenhall, B., Verbin, D., Srinivasan, P.P., Hedman, P.: Mip- nerf 360: Unbounded anti-aliased neural radiance fields. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 5470–5479 (2022)

    2022

  3. [3]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Barron, J.T., Mildenhall, B., Verbin, D., Srinivasan, P.P., Hedman, P.: 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

  4. [4]

    In: European conference on computer vision

    Chen, A., Xu, Z., Geiger, A., Yu, J., Su, H.: Tensorf: Tensorial radiance fields. In: European conference on computer vision. pp. 333–350. Springer (2022)

    2022

  5. [5]

    In: European Conference on Computer Vision

    Chen, D., Liu, Y., Huang, L., Wang, B., Pan, P.: Geoaug: Data augmentation for few-shot nerf with geometry constraints. In: European Conference on Computer Vision. pp. 322–337. Springer (2022)

    2022

  6. [6]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Chen, T., Wang, P., Fan, Z., Wang, Z.: Aug-nerf: Training stronger neural radiance fields with triple-level physically-grounded augmentations. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15191– 15202 (2022)

    2022

  7. [7]

    In: European Conference on Computer Vision

    Chen,Y.,Wu,Q.,Lin,W.,Harandi,M.,Cai,J.:Hac:Hash-gridassistedcontextfor 3d gaussian splatting compression. In: European Conference on Computer Vision. pp. 422–438. Springer (2024)

    2024

  8. [8]

    In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition

    Chung, J., Oh, J., Lee, K.M.: Depth-regularized optimization for 3d gaussian splat- ting in few-shot images. In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition. pp. 811–820 (2024) 16 Junbao Zhou et al

    2024

  9. [9]

    In: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp

    Debevec, P.E., Taylor, C.J., Malik, J.: Modeling and rendering architecture from photographs: A hybrid geometry-and image-based approach. In: Seminal Graphics Papers: Pushing the Boundaries, Volume 2, pp. 465–474 (2023)

    2023

  10. [10]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Deng,K.,Liu,A.,Zhu,J.Y.,Ramanan,D.:Depth-supervisednerf:Fewerviewsand faster training for free. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 12882–12891 (2022)

    2022

  11. [11]

    Advances in neural information processing systems37, 140138–140158 (2024)

    Fan, Z., Wang, K., Wen, K., Zhu, Z., Xu, D., Wang, Z., et al.: Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps. Advances in neural information processing systems37, 140138–140158 (2024)

    2024

  12. [12]

    In: European Conference on Computer Vision

    Fang,G.,Wang,B.:Mini-splatting:Representingsceneswithaconstrainednumber of gaussians. In: European Conference on Computer Vision. pp. 165–181. Springer (2024)

    2024

  13. [13]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Fridovich-Keil, S., Yu, A., Tancik, M., Chen, Q., Recht, B., Kanazawa, A.: Plenox- els: Radiance fields without neural networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 5501–5510 (2022)

    2022

  14. [14]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Garbin, S.J., Kowalski, M., Johnson, M., Shotton, J., Valentin, J.: Fastnerf: High- fidelity neural rendering at 200fps. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 14346–14355 (2021)

    2021

  15. [15]

    arXiv preprint arXiv:2511.14633 (2025)

    Gu, M., Zhang, J., Li, J., Yu, X., Luo, H., Zheng, J., Bai, X.: Sparsesurf: Sparse-view 3d gaussian splatting for surface reconstruction. arXiv preprint arXiv:2511.14633 (2025)

    arXiv 2025

  16. [16]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Guo, Y.C., Kang, D., Bao, L., He, Y., Zhang, S.H.: Nerfren: Neural radiance fields with reflections. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 18409–18418 (2022)

    2022

  17. [17]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Hamdi, A., Melas-Kyriazi, L., Mai, J., Qian, G., Liu, R., Vondrick, C., Ghanem, B., Vedaldi, A.: Ges: Generalized exponential splatting for efficient radiance field rendering. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 19812–19822 (2024)

    2024

  18. [18]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Hanson, A., Tu, A., Singla, V., Jayawardhana, M., Zwicker, M., Goldstein, T.: Pup 3d-gs: Principled uncertainty pruning for 3d gaussian splatting. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 5949–5958 (2025)

    2025

  19. [19]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Hu, W., Wang, Y., Ma, L., Yang, B., Gao, L., Liu, X., Ma, Y.: Tri-miprf: Tri-mip representation for efficient anti-aliasing neural radiance fields. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 19774–19783 (2023)

    2023

  20. [20]

    In: ACM SIGGRAPH 2024 conference papers

    Huang, B., Yu, Z., Chen, A., Geiger, A., Gao, S.: 2d gaussian splatting for geo- metrically accurate radiance fields. In: ACM SIGGRAPH 2024 conference papers. pp. 1–11 (2024)

    2024

  21. [21]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Jain, A., Tancik, M., Abbeel, P.: Putting nerf on a diet: Semantically consistent few-shot view synthesis. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 5885–5894 (2021)

    2021

  22. [22]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Jensen, R., Dahl, A., Vogiatzis, G., Tola, E., Aanæs, H.: Large scale multi-view stereopsis evaluation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 406–413 (2014)

    2014

  23. [23]

    ACM Transactions on Graphics42(4) (July 2023),https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics42(4) (July 2023),https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/

    2023

  24. [24]

    arXiv preprint arXiv:2208.00549 (2022) Title Suppressed Due to Excessive Length 17

    Kirsch, A., Gal, Y.: Unifying approaches in active learning and active sam- pling via fisher information and information-theoretic quantities. arXiv preprint arXiv:2208.00549 (2022) Title Suppressed Due to Excessive Length 17

    arXiv 2022

  25. [25]

    arXiv preprint arXiv:2301.10941 (2023)

    Kwak, M.S., Song, J., Kim, S.: Geconerf: Few-shot neural radiance fields via geo- metric consistency. arXiv preprint arXiv:2301.10941 (2023)

    arXiv 2023

  26. [26]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Lee, J.C., Rho, D., Sun, X., Ko, J.H., Park, E.: 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

  27. [27]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Li, J., Zhang, J., Bai, X., Zheng, J., Ning, X., Zhou, J., Gu, L.: Dngaussian: Opti- mizing sparse-view 3d gaussian radiance fields with global-local depth normaliza- tion. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 20775–20785 (2024)

    2024

  28. [28]

    arXiv preprint arXiv:2511.10647 (2025)

    Lin, H., Chen, S., Liew, J.H., Chen, D.Y., Li, Z., Shi, G., Feng, J., Kang, B.: Depth anything 3: Recovering the visual space from any views. arXiv preprint arXiv:2511.10647 (2025)

    Pith/arXiv arXiv 2025

  29. [29]

    Advances in Neural Information Processing Systems33, 15651–15663 (2020)

    Liu, L., Gu, J., Zaw Lin, K., Chua, T.S., Theobalt, C.: Neural sparse voxel fields. Advances in Neural Information Processing Systems33, 15651–15663 (2020)

    2020

  30. [30]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Meshry, M., Goldman, D.B., Khamis, S., Hoppe, H., Pandey, R., Snavely, N., Martin-Brualla, R.: Neural rerendering in the wild. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 6878– 6887 (2019)

    2019

  31. [31]

    ACM Transactions on Graphics (ToG)38(4), 1–14 (2019)

    Mildenhall, B., Srinivasan, P.P., Ortiz-Cayon, R., Kalantari, N.K., Ramamoorthi, R., Ng, R., Kar, A.: Local light field fusion: Practical view synthesis with pre- scriptive sampling guidelines. ACM Transactions on Graphics (ToG)38(4), 1–14 (2019)

    2019

  32. [32]

    Commu- nications of the ACM65(1), 99–106 (2021)

    Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.: Nerf: Representing scenes as neural radiance fields for view synthesis. Commu- nications of the ACM65(1), 99–106 (2021)

    2021

  33. [33]

    ACM transactions on graphics (TOG)41(4), 1–15 (2022)

    Müller,T.,Evans,A.,Schied,C.,Keller,A.:Instantneuralgraphicsprimitiveswith a multiresolution hash encoding. ACM transactions on graphics (TOG)41(4), 1–15 (2022)

    2022

  34. [34]

    In: Euro- pean Conference on Computer Vision

    Navaneet, K., Pourahmadi Meibodi, K., Abbasi Koohpayegani, S., Pirsiavash, H.: Compgs: Smaller and faster gaussian splatting with vector quantization. In: Euro- pean Conference on Computer Vision. pp. 330–349. Springer (2024)

    2024

  35. [35]

    In: Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition

    Niedermayr, S., Stumpfegger, J., Westermann, R.: Compressed 3d gaussian splat- ting for accelerated novel view synthesis. In: Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition. pp. 10349–10358 (2024)

    2024

  36. [36]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Niemeyer, M., Barron, J.T., Mildenhall, B., Sajjadi, M.S., Geiger, A., Radwan, N.: Regnerf: Regularizing neural radiance fields for view synthesis from sparse in- puts. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 5480–5490 (2022)

    2022

  37. [37]

    Proceedings of the ACM on Computer Graphics and Interactive Techniques7(1), 1–17 (2024)

    Papantonakis, P., Kopanas, G., Kerbl, B., Lanvin, A., Drettakis, G.: Reducing the memory footprint of 3d gaussian splatting. Proceedings of the ACM on Computer Graphics and Interactive Techniques7(1), 1–17 (2024)

    2024

  38. [38]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Park, H., Ryu, G., Kim, W.: Dropgaussian: Structural regularization for sparse- view gaussian splatting. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 21600–21609 (2025)

    2025

  39. [39]

    In: International conference on machine learning

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: International conference on machine learning. pp. 8748–8763. PmLR (2021)

    2021

  40. [40]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Reiser, C., Peng, S., Liao, Y., Geiger, A.: Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 14335–14345 (2021) 18 Junbao Zhou et al

    2021

  41. [41]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Roessle, B., Barron, J.T., Mildenhall, B., Srinivasan, P.P., Nießner, M.: Dense depth priors for neural radiance fields from sparse input views. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 12892– 12901 (2022)

    2022

  42. [42]

    In: 2006 IEEE computer society conference on computer vision and pattern recognition (CVPR’06)

    Seitz, S.M., Curless, B., Diebel, J., Scharstein, D., Szeliski, R.: A comparison and evaluation of multi-view stereo reconstruction algorithms. In: 2006 IEEE computer society conference on computer vision and pattern recognition (CVPR’06). vol. 1, pp. 519–528. IEEE (2006)

    2006

  43. [43]

    Song, J., Park, S., An, H., Cho, S., Kwak, M.S., Cho, S., Kim, S.: Därf: Boosting radiancefieldsfromsparseinputviewswithmonoculardepthadaptation.Advances in Neural Information Processing Systems36, 68458–68470 (2023)

    2023

  44. [44]

    The journal of machine learning research15(1), 1929–1958 (2014)

    Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.: Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research15(1), 1929–1958 (2014)

    1929

  45. [45]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion

    Suhail, M., Esteves, C., Sigal, L., Makadia, A.: Light field neural rendering. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion. pp. 8269–8279 (2022)

    2022

  46. [46]

    In: Proceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition

    Sun, C., Sun, M., Chen, H.T.: Direct voxel grid optimization: Super-fast conver- gence for radiance fields reconstruction. In: Proceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition. pp. 5459–5469 (2022)

    2022

  47. [47]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Truong, P., Rakotosaona, M.J., Manhardt, F., Tombari, F.: Sparf: Neural radiance fields from sparse and noisy poses. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4190–4200 (2023)

    2023

  48. [48]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Verbin, D., Hedman, P., Mildenhall, B., Zickler, T., Barron, J.T., Srinivasan, P.P.: Ref-nerf: Structured view-dependent appearance for neural radiance fields. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  49. [49]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Wang, G., Chen, Z., Loy, C.C., Liu, Z.: Sparsenerf: Distilling depth ranking for few-shot novel view synthesis. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 9065–9076 (2023)

    2023

  50. [50]

    In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition

    Wang, P., Liu, Y., Chen, Z., Liu, L., Liu, Z., Komura, T., Theobalt, C., Wang, W.: F2-nerf: Fast neural radiance field training with free camera trajectories. In: ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecog- nition. pp. 4150–4159 (2023)

    2023

  51. [51]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Wu, R., Mildenhall, B., Henzler, P., Park, K., Gao, R., Watson, D., Srinivasan, P.P., Verbin, D., Barron, J.T., Poole, B., et al.: Reconfusion: 3d reconstruction with diffusion priors. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 21551–21561 (2024)

    2024

  52. [52]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wynn, J., Turmukhambetov, D.: Diffusionerf: Regularizing neural radiance fields with denoising diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4180–4189 (2023)

    2023

  53. [53]

    arXiv preprint arXiv:2312.00206 (2023)

    Xiong, H., Muttukuru, S., Upadhyay, R., Chari, P., Kadambi, A.: Sparsegs: Real- time 360{\deg}sparse view synthesis using gaussian splatting. arXiv preprint arXiv:2312.00206 (2023)

    Pith/arXiv arXiv 2023

  54. [54]

    IEEE transactions on pattern analysis and machine intelligence45(10), 12148–12166 (2023)

    Xu,H.,Yuan,J.,Ma,J.:Murf:Mutuallyreinforcingmulti-modalimageregistration and fusion. IEEE transactions on pattern analysis and machine intelligence45(10), 12148–12166 (2023)

    2023

  55. [55]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Xu, Q., Xu, Z., Philip, J., Bi, S., Shu, Z., Sunkavalli, K., Neumann, U.: Point-nerf: Point-based neural radiance fields. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 5438–5448 (2022) Title Suppressed Due to Excessive Length 19

    2022

  56. [56]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Xu, Y., Wang, L., Chen, M., Ao, S., Li, L., Guo, Y.: Dropoutgs: Dropping out gaussians for better sparse-view rendering. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 701–710 (2025)

    2025

  57. [57]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yang, J., Pavone, M., Wang, Y.: Freenerf: Improving few-shot neural rendering with free frequency regularization. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 8254–8263 (2023)

    2023

  58. [58]

    In: Proceedings of the 32nd ACM International Conference on Multimedia

    Ye, Z., Li, W., Liu, S., Qiao, P., Dou, Y.: Absgs: Recovering fine details in 3d gaussian splatting. In: Proceedings of the 32nd ACM International Conference on Multimedia. pp. 1053–1061 (2024)

    2024

  59. [59]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Yu, A., Li, R., Tancik, M., Li, H., Ng, R., Kanazawa, A.: Plenoctrees for real-time rendering of neural radiance fields. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 5752–5761 (2021)

    2021

  60. [60]

    In: European Conference on Computer Vision

    Zhang, J., Li, J., Yu, X., Huang, L., Gu, L., Zheng, J., Bai, X.: Cor-gs: sparse-view 3d gaussian splatting via co-regularization. In: European Conference on Computer Vision. pp. 335–352. Springer (2024)

    2024

  61. [61]

    In: European Conference on Computer Vision

    Zhang, Z., Hu, W., Lao, Y., He, T., Zhao, H.: Pixel-gs: Density control with pixel- aware gradient for 3d gaussian splatting. In: European Conference on Computer Vision. pp. 326–342. Springer (2024)

    2024

  62. [62]

    In: European conference on computer vision

    Zhu, Z., Fan, Z., Jiang, Y., Wang, Z.: Fsgs: Real-time few-shot view synthesis using gaussian splatting. In: European conference on computer vision. pp. 145–