Pith. sign in

REVIEW 5 major objections 6 minor 27 references

Spatially-Adaptive Hash Encodings For Neural Surface Reconstruction

T0 review · 5 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read A learned spatial mask over multi-resolution hash-grid features lets neural surface reconstruction use fine grids where detail exists and coarse grids elsewhere, achieving state-of-the-art Chamfer-L1 of 0.56 mm on DTU and F1 of 0.53 on…

desk verdict A clean, plausible extension of Neuralangelo with a per-level learned spatial mask over hash features, but the SOTA claim is not yet attributable to the mask: the re-run baseline is unverified and no mask-disabled control exists. read the letter →

arxiv 2412.05179 v1 pith:RRPF7PXZ submitted 2024-12-06 cs.CV

classification cs.CV
keywords neuralsurfacereconstructionhashencodingspatiallyadaptivesigneddistancefunctionvolumerenderingmulti-resolutiongridDTUdatasetTanksandTemples
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

Surface reconstruction from images with neural signed distance functions usually relies on a fixed stack of hash grids, which biases the network toward either coarse or fine geometry everywhere at once. This paper argues that the right bias is spatial: smooth regions should be encoded with coarse grids, while edges and fine texture should use high-resolution grids. It introduces a learned mask field that multiplies the features of each hash-grid level by a per-location weight, so the network effectively chooses its encoding basis as a function of position, together with a progressive schedule that unveils finer grids over training. On the DTU and Tanks and Temples benchmarks, the method reports state-of-the-art accuracy, including a mean Chamfer-L1 of 0.56 mm on DTU against 0.96 mm for the re-run Neuralangelo baseline.

What carries the argument

The load-bearing machinery is the learned spatial mask field. A small hash grid (8 levels, feature dimension 4, hash size $2^{18}$) plus a one-hidden-layer MLP with sigmoid activation outputs $N$ scalar weights $s_\ell(x)$, one per level $\ell$ of the main 16-level SDF hash grid (resolutions $32^3$ to $2048^3$, feature dimension 8, hash size $2^{22}$). The final encoding multiplies each level's feature vector by its mask weight before concatenation, so the network can silence any resolution in any region. Progressive unveiling of fine grids, and gradient blocking to masks of inactive grids, prevents the mask from permanently suppressing high frequencies early in training. Numerical-gradient $\epsilon$ scheduling is inherited from the backbone to provide coarse-to-fine surface optimization.

What would settle it

Run the proposed method and the official Neuralangelo baseline from identical code, seeds, iteration counts, and hyperparameters, and check whether the mean DTU Chamfer-L1 gap stays near the reported 0.40 mm; if the gap shrinks or the method no longer beats the published 0.61 mm, the spatial-mask explanation for the gain fails. A complementary check is to inspect the learned high-frequency mask over a known flat region in a synthetic scene, where it should be near zero.

Watch

Extended reading notes

Core claim

The central discovery is that a spatially varying mask over hash-grid resolutions is enough to fit a wider range of frequencies without adding noise. The mask is produced by a small auxiliary hash grid and a shallow sigmoid MLP, yielding per-level scalars $s_\ell(x) \in (0,1)$; the SDF hash encoding becomes $h(x) = [s_1(x) f_1, \dots, s_N(x) f_N]$, and progressive unveiling blocks gradients to inactive mask levels early in training. Learned masks segment scenes by frequency need: fine grids activate on edges and gratings, while smooth or specular regions fall back to coarse grids. The paper shows this leads to cleaner low-frequency surfaces and improved fine details, with quantitative gains over Neuralangelo on both DTU and Tanks and Temples.

Load-bearing premise

The reported state-of-the-art result rests on the authors' re-run of the official Neuralangelo code, which scores 0.96 mm on DTU rather than the 0.61 mm in the original Neuralangelo paper; if that re-run is not a faithful like-for-like baseline, the improvement attributed to the spatial mask is not established.

Editorial extensions

If this is right

  • On the 15-scene DTU benchmark, the method reports a mean Chamfer-L1 of 0.56 mm, below both the re-run Neuralangelo (0.96 mm) and the published Neuralangelo number (0.61 mm).
  • On Tanks and Temples, the method's F1 score is 0.53, above the re-run (0.47) and published (0.50) Neuralangelo numbers, with visible gains on structures like the barn grating and truck wheels.
  • Ablations show the sigmoid mask activation is important: replacing it with softmax drops performance (Chamfer 0.74 vs 0.65) because the network collapses toward the coarsest resolution.
  • The mask network adds only about 8 million parameters (374M vs 366M), so the adaptive bias comes at negligible storage cost beyond the hash-grid features themselves.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the learned masks genuinely reflect frequency need, they could be reused after training as a pruning map: hash entries in regions where all fine-level masks are near zero are never read, so the about 4.2 GB footprint could be cut without retraining. The paper gestures at this as future work.
  • A controlled test of the mechanism would force the high-frequency masks to be active in a smooth region and measure the resulting chamfer error; the heat maps in Figure 4 suggest this should hurt, but the paper does not run that experiment.
  • The spatial-mask idea is orthogonal to the choice of encoding beyond hash grids; coupling it with spatially adaptive view-direction encodings could address reflective scenes, a limitation the paper acknowledges. This is a natural next step the authors name.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The manuscript proposes a spatially-adaptive variant of multi-resolution hash encoding for neural surface reconstruction. Building on Neuralangelo, it learns a per-level scalar mask field, computed from a small hash grid and a shallow MLP, that multiplies the features of each hash-grid level as a function of spatial position. Training uses an RGB rendering loss, eikonal and curvature regularization, and a progressive unveiling of high-resolution grids with gradient blocking for inactive masks. Experiments on DTU and Tanks and Temples report mean Chamfer-L1 of 0.56 mm and F1 of 0.53, which the paper states is state-of-the-art compared with published Neuralangelo numbers of 0.61 mm and 0.50. Ablations examine mask grid resolution, curvature regularization, and sigmoid vs. softmax activations.

Significance. If the claimed improvement is real, the idea is simple, inexpensive, and likely to transfer to other grid-based neural fields. The paper makes a clear empirical contribution: the mask heatmaps provide a useful diagnostic, and the ablations of design choices are appropriate. However, the central SOTA claim depends on a comparison to a re-run Neuralangelo baseline whose training protocol is not documented, on small margins over published numbers, and on single-run results with no variance. No code or data are provided for verification. These gaps mean the result is plausible but not yet established.

major comments (5)
  1. [Sec. 4.1, Table 2] The re-run of the official Neuralangelo repository (Neuralangelo †) reports a DTU mean Chamfer-L1 of 0.96 mm, compared with the published 0.61 mm. The manuscript does not state the number of training iterations, learning-rate schedule, loss weights, or other hyperparameters for this re-run or for the proposed method. If the re-run used a different or shorter training budget, the 0.40 mm difference between the re-run and the proposed method could be due to training budget rather than the spatial mask. Please provide the exact training configuration and, ideally, reproduce the published Neuralangelo numbers under the same protocol, and report the proposed method under the same protocol.
  2. [Sec. 3.2 and Tables 1-2] There is no ablation that disables the spatial mask within the proposed pipeline (e.g., fixing s_l(x)=1 for all levels). The progressive unveiling schedule, the gradient blocking to inactive masks, and the additional mask-network parameters are all confounded with the mask itself. A three-way comparison is needed: the full proposed method, the proposed method with the mask fixed to one, and the official Neuralangelo re-run under identical training settings. Without this, the reported gains cannot be attributed specifically to the spatially adaptive mask.
  3. [Sec. 4.3, Table 3] In the mask grid-resolution ablation, the default setting [5,11] yields Chamfer 0.654 mm, while [4,10] yields the best value of 0.648 mm. The text says the coarsest setting 'highlights the importance' of the mask but does not explain why [5,11] is preferred over [4,10]. Please justify the choice of default or update the main tables if the best configuration differs.
  4. [Tables 1-2] All results are from a single run with no error bars or significance testing. The mean margins over published Neuralangelo are small (0.05 mm Chamfer on DTU and 0.03 F1 on TNT), and scene-level differences (e.g., DTU Scan 69, where Ours gives 0.54 vs. published 0.53) are within typical run-to-run variation. Please report multiple seeds or a statistical analysis to support the claim that the differences are not noise.
  5. [Abstract and Tables 1-2] The state-of-the-art claim is scoped to the methods listed in the tables, which are mostly from 2022-2023. Please either clarify that the claim is restricted to the compared baselines or add comparisons to more recent surface reconstruction methods that report DTU/TNT metrics. As written, the abstract's 'state-of-the-art performance on two benchmark datasets' is stronger than the evidence.
minor comments (6)
  1. [Sec. 4.2] The sentence 'The results are detailed in Section 4.1' is incorrect; the Tanks and Temples results are in Section 4.2, while Section 4.1 presents DTU. Please fix the cross-reference.
  2. [Sec. 4.2 and Discussion] There are grammatical and typographical errors: 'offical' should be 'official', and 'on both DTU and as well as Tanks and Temples' should be 'on both DTU and Tanks and Temples'.
  3. [Sec. 4.1] The grid resolution range '323 to 20483' should read '32^3 to 2048^3' with proper superscripts.
  4. [Figure 4] The caption should state the color scale and the exact level groupings (e.g., levels 1-8, 8-14, 15-16) used for the Low/Mid/High-Frequency heatmaps, rather than leaving this to the body text.
  5. [General] The paper states that code will be released after acceptance; for reproducibility, please provide code or a detailed configuration listing all hyperparameters at submission time.
  6. [Eq. (4)] Consider clarifying that the mask s_l(x) is a scalar multiplier shared across all feature channels of level l, since the notation s_l(x) * f_l could also be read as channel-wise multiplication if f_l is treated as a vector.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the spatially adaptive mask is learned end-to-end and all reported metrics come from external benchmarks, so no prediction reduces to a fitted input or to a self-citation.

full rationale

I find no circularity in the paper's derivation chain. The proposed construction in Eq. (4) is a feature-wise multiplication h(x)=[s1(x)*f1,...,sN(x)*fN], where the mask network and the SDF/color networks are trained jointly in an end-to-end manner using the rendering loss, the eikonal loss, and the curvature loss in Eq. (8). The benchmark numbers (DTU Chamfer-L1 and Tanks and Temples F1) are obtained from external datasets and official evaluation scripts, so the reported state-of-the-art performance is an empirical outcome rather than a quantity defined by the method's own parameters. The mask grid-resolution choices in Table 3 are hyperparameter selections, not fitted parameters that are subsequently renamed as predictions. The only author self-citation is reference [21] in the related-work passage about explicit parametric surfaces, and it is not load-bearing for the proposed masking scheme, the progressive unveiling schedule, or any performance claim. The discrepancy between the published Neuralangelo numbers and the dagger re-run is a baseline attribution and reproducibility concern, not circularity, because the method's derivation does not assume or enforce that re-run. Likewise, the absence of a mask-disabled control weakens causal attribution but does not make the central claim equivalent to its input. The paper is self-contained against external benchmarks and no step reduces, by construction, to its own inputs.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The method introduces a learned spatial mask field, a new modeling construct parameterized by a small hash grid and MLP. It is not a physical entity and has no independent evidence outside the reported benchmarks. The free parameters are mostly hand-chosen or inherited from Neuralangelo, and the key heuristic of gradient blocking for progressive unveiling is not independently validated.

free parameters (6)
  • spatial mask grid resolutions [dmin, dmax] = [5, 11] (32^3 to 2048^3)
    Chosen by hand. The paper's own ablation reports [4,10] giving lower Chamfer (0.648 vs 0.654) on scan 37, so the default is not the best in the sweep.
  • mask hash grid feature dimension = 4
    Chosen by hand; not ablated in the paper.
  • mask hash dictionary size = 2^18
    Chosen by hand; not ablated in the paper.
  • mask MLP hidden units = 16
    Chosen by hand; not ablated in the paper.
  • loss weights w_eik and w_curv = 0.1 and 5e-4
    Inherited from Neuralangelo; they affect the balance between photometric and geometric losses.
  • numerical gradient epsilon schedule = not specified
    Coarse-to-fine epsilon scheduling is inherited from Neuralangelo and described qualitatively, but the exact schedule used in experiments is not stated.
assumptions (5)
  • domain assumption NeuS volume rendering with logistic SDF-to-density conversion is a valid way to optimize SDFs from images.
    Invoked in Sec. 3.1, Eq. (1)-(2), as the optimization framework; this is a standard assumption in the field.
  • domain assumption Numerical gradients with scheduled epsilon enforce surface normal consistency and coarse-to-fine optimization.
    Sec. 3.1, taken from Neuralangelo [9]; the method depends on this to avoid relearning details across epsilon scales.
  • ad hoc to paper Multiplying hash-grid features by a learned per-level scalar mask preserves the SDF network's ability to represent geometry while adjusting frequency bias.
    Sec. 3.2, Eq. (4); no formal argument is given that masking features is equivalent to a spatially varying frequency selection, only empirical support.
  • domain assumption Hash collisions in the spatial mask grid do not corrupt the mask field.
    Sec. 3.2 uses a 2^18-entry hash table for the mask; the paper notes hash-collision artifacts as a motivation but does not analyze collisions in the mask grid itself.
  • ad hoc to paper Blocking gradients to masks of inactive grid resolutions prevents early convergence of masks to zero.
    Sec. 3.2, Progressive Encoding paragraph; this is a heuristic to avoid the mask learning to discard fine grids early, and it is not ablated in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spatially-Adaptive Hash Encodings For Neural Surface Reconstruction." pith.science (2026). https://pith.science/paper/RRPF7PXZ

@misc{pith2026241205179,
  author       = {Pith},
  title        = {Pith review of: Spatially-Adaptive Hash Encodings For Neural Surface Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RRPF7PXZ}},
  note         = {Machine review of arXiv:2412.05179}
}
read the original abstract

Positional encodings are a common component of neural scene reconstruction methods, and provide a way to bias the learning of neural fields towards coarser or finer representations. Current neural surface reconstruction methods use a "one-size-fits-all" approach to encoding, choosing a fixed set of encoding functions, and therefore bias, across all scenes. Current state-of-the-art surface reconstruction approaches leverage grid-based multi-resolution hash encoding in order to recover high-detail geometry. We propose a learned approach which allows the network to choose its encoding basis as a function of space, by masking the contribution of features stored at separate grid resolutions. The resulting spatially adaptive approach allows the network to fit a wider range of frequencies without introducing noise. We test our approach on standard benchmark surface reconstruction datasets and achieve state-of-the-art performance on two benchmark datasets.

Figures

Figures reproduced from arXiv: 2412.05179 by the authors.

Figure 1
Figure 1. In order to modulate typical hash grid features fields (top [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Qualitative DTU results. Our approach improves surface details as well as surface accuracy of coarser components. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Qualitative Tanks and Temples results. Our approach produces improved surface accuracy, attaining generally cleaner surfaces [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Renderings of spatial mask heat maps. Red indicates the corresponding spatial masks take the value 1. Blue indicates the mask [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 6
Figure 6. Figure 6: Ablation results of models with and without curva [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Ablation results of models using sigmoid and softmax [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 5
Figure 5. Figure 5: Close-up comparison of details in scene 37. Left: spatial [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 16 canonical work pages

  1. [1]

    Large-scale data for multiple-view stereopsis

    Henrik Aanæs, Rasmus Ramsbøl Jensen, George V ogiatzis, Engin Tola, and Anders Bjorholm Dahl. Large-scale data for multiple-view stereopsis. International Journal of Computer Vision, pages 1–16, 2016. 4

  2. [2]

    Shape reconstruction by learn- ing differentiable surface representations

    Jan Bednarik, Shaifali Parashar, Erhan Gundogdu, Mathieu Salzmann, and Pascal Fua. Shape reconstruction by learn- ing differentiable surface representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4716–4725, 2020. 2

  3. [3]

    Accurate, dense, and robust multiview stereopsis

    Yasutaka Furukawa and Jean Ponce. Accurate, dense, and robust multiview stereopsis. IEEE transactions on pattern analysis and machine intelligence , 32(8):1362–1376, 2009. 1

  4. [4]

    Implicit geometric regularization for learning shapes

    Amos Gropp, Lior Yariv, Niv Haim, Matan Atzmon, and Yaron Lipman. Implicit geometric regularization for learning shapes. arXiv preprint arXiv:2002.10099, 2020. 2, 4

  5. [5]

    A papier-m ˆach´e ap- proach to learning 3d surface generation

    Thibault Groueix, Matthew Fisher, Vladimir G Kim, Bryan C Russell, and Mathieu Aubry. A papier-m ˆach´e ap- proach to learning 3d surface generation. In Proceedings of the IEEE conference on computer vision and pattern recog- nition, pages 216–224, 2018. 2

  6. [6]

    Multiscale tensor decomposition and rendering equation encoding for view synthesis

    Kang Han and Wei Xiang. Multiscale tensor decomposition and rendering equation encoding for view synthesis. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 4232–4241, June

  7. [7]

    Sape: Spatially-adaptive progressive encoding for neural optimization

    Amir Hertz, Or Perel, Raja Giryes, Olga Sorkine-Hornung, and Daniel Cohen-Or. Sape: Spatially-adaptive progressive encoding for neural optimization. Advances in Neural Infor- mation Processing Systems, 34:8820–8832, 2021. 2, 3, 4

  8. [8]

    Tanks and temples: Benchmarking large-scale scene reconstruction

    Arno Knapitsch, Jaesik Park, Qian-Yi Zhou, and Vladlen Koltun. Tanks and temples: Benchmarking large-scale scene reconstruction. ACM Transactions on Graphics, 36(4), 2017. 4, 6

Show all 27 references
  1. [9]

    Neuralangelo: High-fidelity neural surface reconstruction

    Zhaoshuo Li, Thomas M ¨uller, Alex Evans, Russell H Tay- lor, Mathias Unberath, Ming-Yu Liu, and Chen-Hsuan Lin. Neuralangelo: High-fidelity neural surface reconstruction. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 8456–8465, ...

  2. [10]

    Neural sparse voxel fields

    Lingjie Liu, Jiatao Gu, Kyaw Zaw Lin, Tat-Seng Chua, and Christian Theobalt. Neural sparse voxel fields. Advances in Neural Information Processing Systems, 33:15651–15663,

  3. [11]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    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. Communications of the ACM, 65(1):99–106, 2021. 1, 7

  4. [12]

    Instant neural graphics primitives with a mul- tiresolution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a mul- tiresolution hash encoding. ACM transactions on graphics (TOG), 41(4):1–15, 2022. 2

  5. [13]

    Instant neural graphics primitives with a multires- olution hash encoding

    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, July 2022. 2, 3

  6. [14]

    Rectified linear units im- prove restricted boltzmann machines

    Vinod Nair and Geoffrey E Hinton. Rectified linear units im- prove restricted boltzmann machines. In Proceedings of the 27th international conference on machine learning (ICML- 10), pages 807–814, 2010. 4

  7. [15]

    Structure-from-motion revisited

    Johannes Lutz Sch ¨onberger and Jan-Michael Frahm. Structure-from-motion revisited. In Conference on Com- puter Vision and Pattern Recognition (CVPR), 2016. 1, 5

  8. [16]

    Photo tourism: exploring photo collections in 3d

    Noah Snavely, Steven M Seitz, and Richard Szeliski. Photo tourism: exploring photo collections in 3d. In ACM siggraph 2006 papers, pages 835–846. 2006. 1

  9. [17]

    Neural geometric level of detail: Real-time rendering with implicit 3D shapes

    Towaki Takikawa, Joey Litalien, Kangxue Yin, Karsten Kreis, Charles Loop, Derek Nowrouzezahrai, Alec Jacobson, Morgan McGuire, and Sanja Fidler. Neural geometric level of detail: Real-time rendering with implicit 3D shapes. 2021. 2

  10. [18]

    Fourier features let networks learn high frequency functions in low dimen- sional domains

    Matthew Tancik, Pratul Srinivasan, Ben Mildenhall, Sara Fridovich-Keil, Nithin Raghavan, Utkarsh Singhal, Ravi Ra- mamoorthi, Jonathan Barron, and Ren Ng. Fourier features let networks learn high frequency functions in low dimen- sional domains. Advances in neural information ...

  11. [19]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. 2

  12. [20]

    Ref-nerf: Struc- tured view-dependent appearance for neural radiance fields

    Dor Verbin, Peter Hedman, Ben Mildenhall, Todd Zickler, Jonathan T Barron, and Pratul P Srinivasan. Ref-nerf: Struc- tured view-dependent appearance for neural radiance fields. In 2022 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 5481–5490. IE...

  13. [21]

    Explicit neural surfaces: Learning contin- uous geometry with deformation fields

    Thomas Walker, Octave Mariotti, Amir Vaxman, and Hakan Bilen. Explicit neural surfaces: Learning contin- uous geometry with deformation fields. arXiv preprint arXiv:2306.02956, 2023. 2

  14. [22]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction,

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction,

  15. [23]

    Hf-neus: Improved surface reconstruction using high-frequency de- tails

    Yiqun Wang, Ivan Skorokhodov, and Peter Wonka. Hf-neus: Improved surface reconstruction using high-frequency de- tails. Advances in Neural Information Processing Systems , 35:1966–1978, 2022. 2, 3, 7

  16. [24]

    Deep geomet- ric prior for surface reconstruction

    Francis Williams, Teseo Schneider, Claudio Silva, Denis Zorin, Joan Bruna, and Daniele Panozzo. Deep geomet- ric prior for surface reconstruction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10130–10139, 2019. 2

  17. [25]

    Hollownerf: Pruning hashgrid-based nerfs with trainable collision mitigation, 2023

    Xiufeng Xie, Riccardo Gherardi, Zhihong Pan, and Stephen Huang. Hollownerf: Pruning hashgrid-based nerfs with trainable collision mitigation, 2023. 8

  18. [26]

    V ol- ume rendering of neural implicit surfaces

    Lior Yariv, Jiatao Gu, Yoni Kasten, and Yaron Lipman. V ol- ume rendering of neural implicit surfaces. Advances in Neu- ral Information Processing Systems, 34:4805–4815, 2021. 1, 2, 7

  19. [27]

    Improving deep neural networks using softplus units

    Hao Zheng, Zhanlei Yang, Wenju Liu, Jizhong Liang, and Yanpeng Li. Improving deep neural networks using softplus units. In 2015 International joint conference on neural net- works (IJCNN), pages 1–4. IEEE, 2015. 4

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.