Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

FreNBRDF: A Frequency-Rectified Neural Material Representation

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read FreNBRDF claims that adding a spherical-harmonic frequency error term to the training loss improves the accuracy and robustness of neural BRDF reconstruction and editing on the MERL dataset.

desk verdict The frequency-rectified loss is built on a misparameterized spherical-harmonic transform; the idea is a genuine extension of FrePolad, but the central claim does not hold as written. read the letter →

arxiv 2507.00476 v1 pith:X3CGTBSX submitted 2025-07-01 cs.GR cs.CV

Chenliang Zhou , Zheyuan Hu , Cengiz Oztireli This is my paper · ORCID
classification cs.GRcs.CV
keywords neuralBRDFrepresentationfrequency-rectifiedlosssphericalharmonicsmaterialreconstructioneditingMERLdatasetautoencoderlatentspaceimplicit
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

The paper tries to establish that neural representations of surface reflectance can be made more accurate by training them with an explicit frequency-domain error, not only a spatial-sample error. It introduces FreNBRDF, an autoencoder pipeline that takes a material's reflectance function, slices it at a fixed azimuth angle, expands each slice in spherical harmonics, and adds the mean squared difference of the resulting coefficients to the reconstruction loss. The claim is that this frequency-rectified loss helps the network learn a more structured distribution of neural material weights, which shows up as better reconstruction fidelity and, especially, more stable interpolation between materials during editing. The payoff, if true, is that implicit neural materials become more reliable for downstream appearance tasks while keeping the compact latent representation.

What carries the argument

The central mechanism is the frequency-rectified loss $L_{\mathrm{fre}}$. To define it, the paper maps a BRDF to $S^2$ by fixing the azimuth argument $\varphi_H$ to a constant $\alpha$, interpolates the discrete measurements using normalized Gaussian weights over the $k$ nearest neighbors, and expands the result in spherical harmonics up to a truncation order $L_{\max}$. The squared difference of the resulting coefficients $c_{l,m}$ between reconstructed and target materials is then added to the $L_1$ reconstruction loss and weight regularizers, so the autoencoder is trained to keep neural material weights consistent in both spatial and frequency domains. The mechanism carries the argument because the claimed improvements all flow from this extra term; its behavior depends on the choices of interpolation bandwidth $\sigma$, neighbor count $k$, and truncation order $L_{\max}$.

What would settle it

Take a synthetic material whose true reflection function is known analytically, compute its exact spherical-harmonic coefficients, then recompute them from discretely sampled measurements using the paper's interpolation procedure. If the interpolated coefficients differ substantially from the exact ones at low frequencies, then the frequency loss is minimizing error against a distorted target rather than the material's real spectrum, and the claimed mechanism fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that FreNBRDF, a neural material representation trained with a frequency-rectified loss, reconstructs and edits measured materials more faithfully than the hypernetwork set-encoder baseline and the naive NBRDF pipeline. The construction treats a BRDF as a function on the sphere by fixing $\varphi_H = \alpha$, interpolating the discrete MERL samples with normalized Gaussian $k$-nearest-neighbor weights, and computing spherical-harmonic coefficients $c_{l,m}$ per RGB channel. The frequency-rectified loss $L_{\mathrm{fre}}$ is the mean squared error between these coefficients for reconstructed and ground-truth BRDFs, added to the spatial reconstruction loss with a weighting hyperparameter $\eta$. With this term, the paper reports consistently lower frequency-compliance loss and better PSNR and SSIM on both reconstruction and editing, and interprets this as evidence that frequency rectification gives the latent space a more structured and interpretable organization.

Load-bearing premise

The method assumes that the frequency content of the interpolated material slices, formed by fixing one angle and averaging nearby measured samples, faithfully represents the material's true high-frequency detail, so that reducing the frequency error improves appearance instead of merely matching an artifact of the interpolation.

Editorial extensions

If this is right

  • If the central claim holds, FreNBRDF reconstructs unseen MERL materials with better PSNR and SSIM than the two baselines while cutting the frequency-compliance loss from about $6.80 \times 10^{-3}$ to $0.23 \times 10^{-3}$.
  • Latent-space interpolation becomes a dependable editing operation: interpolation metrics stay near reconstruction quality instead of degrading as they do for the baselines.
  • Because the autoencoder uses a permutation-invariant set encoder, the frequency-rectified loss applies to arbitrary-size and sparse sample sets, preserving the pipeline's generalizability.
  • The same frequency-rectification principle is positioned to extend to anisotropic, spatially varying, and physically-based materials, the paper's stated next targets.

Reading between the lines

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

  • The reconstruction table shows RMSE slightly worse than the naive NBRDF baseline while PSNR and SSIM improve; a plausible reading is that the frequency term trades a small amount of pixel-level intensity accuracy for structural and perceptual fidelity, a trade-off the paper does not quantify.
  • Because the $\varphi_H$-fixed slice drops azimuthal dependence, the current gains on isotropic MERL data may partly reflect the simplification rather than the frequency loss itself; testing on anisotropic materials would separate these factors.
  • The paper reports no ablation of the weighting hyperparameter $\eta$, so the causal role of $L_{\mathrm{fre}}$ is inferred from comparisons against the no-frequency baseline; sweeping $\eta$ would show whether the benefit is monotonic or saturates.
  • Being differentiable in the network output, the frequency loss could be dropped into other neural appearance models, such as neural scene representations, as a regularizer rather than a training objective for a dedicated material autoencoder.
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.

Referee Report

4 major / 5 minor

Summary. The paper proposes FreNBRDF, a modification of the NBRDF autoencoder for measured BRDFs, adding a 'frequency-rectified' loss that penalizes differences between spherical harmonic coefficients of BRDF slices. The authors claim this improves reconstruction and editing quality over two baselines, and report quantitative results on the MERL dataset. The core technical contribution is an SH-based frequency analysis of BRDFs, obtained by fixing one angular coordinate and interpolating the tabulated MERL data onto a sphere.

Significance. If the frequency decomposition were well-defined, a principled frequency-domain regularizer for neural BRDFs could be a useful contribution to material modeling. The paper is clearly written and the idea of linking neural material representations to spherical harmonics is appealing. However, the central construction is mathematically defective (the isotropic co-ordinate statement is inverted and the SH transform is applied to a function that is not defined on the sphere), and the one independent rendering metric (RMSE in Table 1) contradicts the accuracy claim. The evaluation relies on the training loss as a success metric and omits all hyperparameter values. As a result, the paper's main claim is not supported.

major comments (4)
  1. [Sec. 3.1, Eq. (3)] The isotropy statement is inverted. In the Rusinkiewicz parameterization, an isotropic BRDF is independent of the half-vector azimuth φ_H, not of φ_D. This is confirmed by the paper's own Sec. 4.1, which describes the MERL array as being indexed by (θ_H, θ_D, φ_D). Fixing φ_H = α in Eq. (3) therefore leaves φ_D in the slice, and the SH coefficients do not encode the frequency content of an isotropic material as claimed. If the authors intended to fix φ_D instead, then the transform discards the one azimuthal coordinate that carries directional variation for isotropic materials; either way the construction is not a valid frequency analysis of the material.
  2. [Sec. 3.2, Eqs. (8)–(9)] The function f_r(θ, φ) defined in Eq. (3) is not a well-defined function on the unit sphere S². The variables θ_H and θ_D each only range over [0, π/2] for the MERL dataset, so a slice in these variables covers at most a quarter of the sphere. The paper gives no extension of the BRDF to the full sphere, yet Eqs. (8)–(9) integrate over the full sphere. Consequently the spherical harmonic coefficients c_{l,m} are mathematically undefined, and the frequency-rectified loss L_fre in Eq. (10) is not a meaningful quantity. This is a load-bearing defect because the central claim rests entirely on these coefficients.
  3. [Tables 1 and 2, Sec. 4.4] The reported RMSE for FreNBRDF in Table 1 (6.74×10⁻²) is worse than the naive NBRDF baseline (6.60×10⁻²), which directly contradicts the abstract's claim of improved accuracy. The single sentence after Table 1 that the higher RMSE is 'likely due to' the loss being designed for both spatial and frequency consistency is not a testable explanation. No error bars or statistical significance tests are provided anywhere in the paper. Furthermore, L_fre is the very loss minimized during training, so its drop from 6.80×10⁻³ to 0.23×10⁻³ in Table 1 is expected and cannot be used as independent evidence of frequency fidelity; the claim of improved frequency compliance is therefore circular.
  4. [Sec. 4.2, Eqs. (4), (6), (8), (11)] The method's key hyperparameters are never reported: the neighbour count k and Gaussian standard deviation σ in Eq. (4) and Eq. (6), the spherical harmonic truncation degree L_max in Eq. (8), the frequency-loss weight η in Eq. (11), and the regularization coefficients λ_1 and λ_2 in Eq. (1). The paper states that 'optimal hyperparameters are selected based on performance on the validation set' but gives no values. Without these, the experiments are not reproducible, and it is impossible to assess whether the reported results depend on an accidental choice of these parameters, especially σ and k, which determine the interpolated function that is supposedly transformed to the frequency domain.
minor comments (5)
  1. [Sec. 3.2 title] The heading 'Recitification' contains a typo; it should be 'Rectification'.
  2. [Sec. 4.3] The phrase 'accessing the frequency compliance' should read 'assessing the frequency compliance'.
  3. [Eqs. (3), (8)–(9)] The notation for the azimuthal coordinate is inconsistent: φ and ϕ are both used for what appears to be the same variable. Please standardize to a single symbol.
  4. [Figs. 4 and 5] The captions state 'with mean (solid) and variance (dashed line)' but do not specify whether the mean and variance are taken over the test set, over samples, or over rendering viewpoints. This should be clarified.
  5. [Sec. 4.4, Table 1] The phrase 'state-of-the-art baselines' is overstated: the comparison is only against one external method [6] and the authors' own naive pipeline. Please temper the 'state-of-the-art' claim or add comparisons to recent neural BRDF and frequency-aware methods.

Circularity Check

1 steps flagged · score 4.0 of 10

The frequency-rectified loss Lfre is defined as the training objective (Eq. 11) and then reported in Sec. 4.3 as the evaluation metric for 'frequency compliance', so the dramatic Lfre reduction in Tables 1-2 is expected by construction; independent PSNR/SSIM metrics provide real but separate support.

  1. fitted input called prediction [Eq. (11) (training objective) and Sec. 4.3 'Metrics' (evaluation), Tables 1 and 2.]
    "L(fr, f′r) := Lrec(fr, f′r) + ηLfre(fr, f′r) ... To evaluate the performance of material reconstruction and editing, we report the frequency-rectified loss (Eq. (10)) accessing the frequency compliance."

    Lfre is not a held-out diagnostic: it is the second term of the training loss L in Eq. (11), so the network is explicitly optimized to minimize it. Reporting Lfre itself as the primary 'frequency compliance' metric therefore guarantees a favorable value relative to baselines trained without that term; the 0.23 vs 6.7 numbers in Tables 1 and 2 are expected by construction. This is only a partial circularity because PSNR and SSIM are independent rendering-based metrics and do improve, but the Lfre column cannot independently validate the claim that the SH coefficients capture the material's true frequency content.

full rationale

The central construction—using spherical-harmonic coefficients of a k-NN-interpolated BRDF slice to define a frequency loss—is not circular in itself; it is a concrete feature-space loss. The only circular element I can exhibit by quotation is the evaluation: the paper adds Lfre to the training loss (Eq. 11) and then uses the same Lfre as the headline 'frequency compliance' metric in Sec. 4.3 and Tables 1-2. That metric is by construction minimized by the proposed method, so it cannot serve as independent evidence. The paper also cites its own prior FrePolad work [7] for the frequency-rectification idea, but this citation is inspirational rather than load-bearing: the spherical-harmonic transform is standard external mathematics, and no uniqueness or correctness claim is imported from [7]. The deeper mathematical concerns about Eq. (3) and the S2 integration domain are correctness risks, not circularity, because the loss is not defined in terms of the evaluation outcome. Overall, the independent PSNR/SSIM gains mean the central claim does not reduce to the fitted objective, so the circularity burden is moderate rather than severe.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The method's central claim depends on several unreported hyperparameters and heuristic assumptions. The most significant is the assertion that SH coefficient matching improves NBRDF learning, which is stated but not derived or ablated.

free parameters (6)
  • eta (frequency-rectified loss weight) = not reported
    Hyperparameter controlling the strength of the frequency-rectified loss in Eq. (11); selected by validation but value not given.
  • lambda1, lambda2 (regularization coefficients) = not reported
    Regularization coefficients in Eq. (1) for NBRDF weights and latent embeddings; not specified.
  • sigma (Gaussian weight standard deviation) = not reported
    Controls the k-NN interpolation weights in Eq. (6); not specified.
  • k (nearest neighbor count) = not reported
    Number of neighbors used in Eq. (4) to interpolate BRDF values; not specified.
  • SH truncation degree L_max = not reported
    Maximum spherical harmonic degree used in the infinite sum in Eq. (8); not specified.
  • alpha values / number of phi_H slices = not reported
    The method fixes phi_H to alpha and averages over multiple values; the set of alpha is not reported.
assumptions (4)
  • standard math Spherical harmonics form an orthonormal basis for square-integrable functions on the unit sphere S2
    Used for the SH expansion in Eqs. (8)-(9).
  • domain assumption BRDFs in MERL are isotropic and thus independent of phi_D, so a 3D BRDF can be reduced to a 2D function on a sphere by fixing the remaining azimuthal angle phi_H
    Underlies Eq. (3), the construction of fr(theta, phi) from fr(theta_H, theta_D, phi_H).
  • domain assumption The k-NN Gaussian interpolation in Eqs. (4)-(7) produces a continuous approximation of the discrete BRDF that preserves its frequency content
    Needed for computing SH coefficients, but no validation is provided.
  • ad hoc to paper Minimizing MSE of SH coefficients will improve NBRDF weight learning and downstream reconstruction/editing
    This is the core motivation for the frequency-rectified loss; it is stated as an observation in Sec. 3.2 but not derived or ablated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FreNBRDF: A Frequency-Rectified Neural Material Representation." pith.science (2026). https://pith.science/paper/X3CGTBSX

@misc{pith2026250700476,
  author       = {Pith},
  title        = {Pith review of: FreNBRDF: A Frequency-Rectified Neural Material Representation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X3CGTBSX}},
  note         = {Machine review of arXiv:2507.00476}
}
read the original abstract

Accurate material modeling is crucial for achieving photorealistic rendering, bridging the gap between computer-generated imagery and real-world photographs. While traditional approaches rely on tabulated BRDF data, recent work has shifted towards implicit neural representations, which offer compact and flexible frameworks for a range of tasks. However, their behavior in the frequency domain remains poorly understood. To address this, we introduce FreNBRDF, a frequency-rectified neural material representation. By leveraging spherical harmonics, we integrate frequency-domain considerations into neural BRDF modeling. We propose a novel frequency-rectified loss, derived from a frequency analysis of neural materials, and incorporate it into a generalizable and adaptive reconstruction and editing pipeline. This framework enhances fidelity, adaptability, and efficiency. Extensive experiments demonstrate that FreNBRDF improves the accuracy and robustness of material appearance reconstruction and editing compared to state-of-the-art baselines, enabling more structured and interpretable downstream tasks and applications.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Hybrid Neural-Microfacet BRDF Model for Real-Time Rendering

    cs.GR 2026-08 conditional novelty 6.0 of 10

    A hybrid BRDF model, combining a GGX analytical term with a tiny learned residual and gating network, fits measured materials more accurately than fully neural models at equal memory cost.

Reference graph

Works this paper leans on

28 extracted references · 27 canonical work pages · cited by 1 Pith paper

  1. [1]

    INTRODUCTION Material properties play a crucial role in visual computing, serving as a fundamental component in applications such as rendering, augmented reality, and scene understanding [1]. This is usually realized through the modeling and reconstruc- tion of the bidirectional reflectance distribution functions (BRDFs) [2], which describe the relationsh...

    work page Pith review arXiv 2025

  2. [2]

    Using this technique, RGL [9] captured 62 materials, including anisotropic ones

    RELATED WORK Material acquisition Gonio-photometers [8] are widely used for measuring retro-reflection, enabling precise char- acterization of material appearance. Using this technique, RGL [9] captured 62 materials, including anisotropic ones. In contrast, image-based measurement devices [10] were used to construct the MERL dataset [3], which consists of...

    work page

  3. [3]

    METHOD 3.1. NBRDF autoencoder pipeline Under Rusinkiewicz reparameterization of BRDFs [20], we represent the incoming and outgoing directions using the half H and difference D vectors or their spherical coordinates θH , ϕH and the difference vector θD, ϕD, respectively. Note that for isotropic materials, the BRDFs are independent of ϕD. Therefore, a BRDF ...

    work page

  4. [4]

    Dataset The MERL dataset [3] is used in this study for its diversity and data-driven nature, making it suitable for both statistical and neural-network-based methods

    EXPERIMENTS 4.1. Dataset The MERL dataset [3] is used in this study for its diversity and data-driven nature, making it suitable for both statistical and neural-network-based methods. It contains 100 measured real-world materials. Each BRDF is represented as a90×90× 180 × 3 floating-point array, mapping uniformly sampled in- put angles (θH , θD, ϕD) under...

    work page 2000

  5. [5]

    CONCLUSION AND FUTURE WORK In this work, we propose FreNBRDF, a frequency-rectified neural material representation. Building on this represen- tation, we develop a material reconstruction and editing pipeline based on an autoencoder architecture, incorporat- ing a novel frequency-rectified loss to enforce accuracy in the frequency domain. We evaluate our ...

    work page

  6. [6]

    Neumadiff: Neural mate- rial synthesis via hyperdiffusion,

    Chenliang Zhou, Zheyuan Hu, Alejandro Sztrajman, Yancheng Cai, Yaru Liu, and Cengiz Oztireli, “Neumadiff: Neural mate- rial synthesis via hyperdiffusion,” 2024

    work page 2024

  7. [7]

    Tech- niques for brdf evaluation,

    Mislene da Silva Nunes, Fernando Melo Nascimento, Gastão Florêncio Miranda Jr., and Beatriz Trinchão Andrade, “Tech- niques for brdf evaluation,” The Visual Computer, vol. 38, pp. 573 – 589, 2021

    work page 2021

  8. [8]

    A data- driven reflectance model,

    W. Matusik, H. Pfister, M. Brand, and L. McMillan, “A data- driven reflectance model,” ACM Transactions on Graphics (TOG), vol. 22, no. 3, pp. 759–769, July 2003

    work page 2003

Show all 28 references
  1. [9]

    Neural brdf representation and importance sam- pling,

    Alejandro Sztrajman, Gilles Rainer, Tobias Ritschel, and Tim Weyrich, “Neural brdf representation and importance sam- pling,” Computer Graphics Forum, 2021

    2021

  2. [10]

    Physically based neural bidirectional reflectance distribution function,

    Chenliang Zhou, Alejandro Sztrajman, Gilles Rainer, Fangcheng Zhong, Fazilet Gokbudak, Zhilin Guo, Weihao Xia, Rafal Mantiuk, and Cengiz Oztireli, “Physically based neural bidirectional reflectance distribution function,” arXiv preprint arXiv:2411.02347, 2024

    2024 arXiv

  3. [11]

    Hy- pernetworks for generalizable brdf representation,

    Fazilet Gokbudak, Alejandro Sztrajman, Chenliang Zhou, Fangcheng Zhong, Rafal Mantiuk, and Cengiz Oztireli, “Hy- pernetworks for generalizable brdf representation,” The 18th European Conference on Computer Vision (ECCV), 2024

    2024

  4. [12]

    Frepolad: Frequency-rectified point latent diffusion for point cloud generation,

    Chenliang Zhou, Fangcheng Zhong, Param Hanji, Zhilin Guo, Kyle Fogarty, Alejandro Sztrajman, Hongyun Gao, and Cengiz Oztireli, “Frepolad: Frequency-rectified point latent diffusion for point cloud generation,” in ECCV 2024, 2024

    2024

  5. [13]

    Pab pgii gonio- photometer,

    PAB Advanced Technologies Ltd., “Pab pgii gonio- photometer,” 2018, Accessed: 2018-01-09

    2018

  6. [14]

    An adaptive parameteri- zation for efficient material acquisition and rendering,

    Jonathan Dupuy and Wenzel Jakob, “An adaptive parameteri- zation for efficient material acquisition and rendering,” Trans- actions on Graphics (Proceedings of SIGGRAPH Asia) , vol. 37, no. 6, pp. 274:1–274:18, Nov. 2018

    2018

  7. [15]

    Image-based bidirectional re- flectance distribution function measurement,

    Stephen R. Marschner, Stephen H. Westin, Eric P. F. Lafor- tune, and Kenneth E. Torrance, “Image-based bidirectional re- flectance distribution function measurement,” Appl. Opt., vol. 39, no. 16, pp. 2592–2600, Jun 2000

    2000

  8. [16]

    Design and implementation of practical bidirectional texture function measurement devices focusing on the developments at the university of bonn,

    Christopher Schwartz, Ralf Sarlette, Michael Weinmann, Mar- tin Rump, and Reinhard Klein, “Design and implementation of practical bidirectional texture function measurement devices focusing on the developments at the university of bonn,” Sen- sors, vol. 14, no. 5, 2014

    2014

  9. [17]

    Illumination for computer generated pic- tures,

    Bui Tuong Phong, “Illumination for computer generated pic- tures,” Commun. ACM, vol. 18, no. 6, pp. 311–317, jun 1975

    1975

  10. [18]

    Non-linear approximation of re- flectance functions,

    Eric P. F. Lafortune, Sing-Choong Foo, Kenneth E. Torrance, and Donald P. Greenberg, “Non-linear approximation of re- flectance functions,” inProceedings of the 24th Annual Confer- ence on Computer Graphics and Interactive Techniques, USA, 1997, SIGGRAPH ’97, p. 117–126, ACM P...

    1997

  11. [19]

    Measuring and modeling anisotropic re- flection,

    Gregory J. Ward, “Measuring and modeling anisotropic re- flection,” SIGGRAPH Comput. Graph. , vol. 26, no. 2, pp. 265–272, jul 1992

    1992

  12. [20]

    A reflectance model for com- puter graphics,

    R. L. Cook and K. E. Torrance, “A reflectance model for com- puter graphics,” ACM Trans. Graph., vol. 1, no. 1, pp. 7–24, jan 1982

    1982

  13. [21]

    Microfacet models for refraction through rough surfaces,

    Bruce Walter, Stephen R. Marschner, Hongsong Li, and Ken- neth E. Torrance, “Microfacet models for refraction through rough surfaces,” in Proceedings of the 18th Eurographics Conference on Rendering Techniques , Goslar, DEU, 2007, EGSR’07, p. 195–206, Eurographics Association

    2007

  14. [22]

    Physically-based shading at disney,

    Brent Burley, “Physically-based shading at disney,” SIG- GRAPH Comput. Graph., 2012

    2012

  15. [23]

    On optimal, minimal brdf sampling for re- flectance acquisition,

    Jannik Boll Nielsen, Henrik Wann Jensen, and Ravi Ra- mamoorthi, “On optimal, minimal brdf sampling for re- flectance acquisition,” ACM Transactions on Graphics (TOG), vol. 34, no. 6, pp. 186:1–186:11, November 2015

    2015

  16. [24]

    Spherical fourier neural operators: Learning stable dy- namics on the sphere,

    Boris Bonev, Thorsten Kurth, Christian Hundt, Jaideep Pathak, Maximilian Baust, Karthik Kashinath, and Anima Anandku- mar, “Spherical fourier neural operators: Learning stable dy- namics on the sphere,” 2023

    2023

  17. [25]

    A new change of variables for efficient brdf representation,

    Szymon M. Rusinkiewicz, “A new change of variables for efficient brdf representation,” in Rendering Techniques ’98 , George Drettakis and Nelson Max, Eds., Vienna, 1998, pp. 11– 22, Springer Vienna

    1998

  18. [26]

    Deep sets,

    Manzil Zaheer, Satwik Kottur, Siamak Ravanbhakhsh, Barn- abás Póczos, Ruslan Salakhutdinov, and Alexander J Smola, “Deep sets,” in Proceedings of the 31st International Confer- ence on Neural Information Processing Systems , Red Hook, NY , USA, 2017, NIPS’17, p. 3394–3404, Cur...

    2017

  19. [27]

    Adam: A method for stochastic optimization,

    Diederik P. Kingma and Jimmy Ba, “Adam: A method for stochastic optimization,” CoRR, vol. abs/1412.6980, 2014

    2014 arXiv

  20. [28]

    Image quality assessment: from error visibility to structural similarity,

    Zhou Wang, A.C. Bovik, H.R. Sheikh, and E.P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,” IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600–612, 2004

    2004

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.