Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Tolerance-Aware Deep Optics

T0 review · 3 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read Deep optics usually assumes a perfect lens; this paper claims the first end-to-end framework that instead trains the lens and its reconstruction network against sampled manufacturing and assembly tolerances, recovering more than 2 dB of…

desk verdict First to co-optimize refractive-lens deep optics against manufacturing tolerances in the loop, but the headline robustness claim is in-sample and the '>2dB' gain is overstated. read the letter →

arxiv 2502.04719 v1 pith:J2BEEETR submitted 2025-02-07 cs.CV cs.GR

Jun Dai , Liqun Chen , Xinge Yang , Yuyao Hu , Jinwei Gu , Tianfan Xue This is my paper · ORCID
classification cs.CVcs.GR
keywords deepopticstolerance-awareoptimizationdifferentiableraytracinglensdesignmanufacturingtolerancescomputationalimagingpointspreadfunctionimagedeblurring
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

Deep optics designs a lens and a reconstruction network as one trainable system, but it usually assumes the manufactured lens matches the design exactly. This paper argues that the unavoidable errors of mass production—decentering, tilt, thickness, and curvature deviations on each lens—are precisely what makes fabricated deep-optics cameras underperform, and that these errors should be built into the design process itself. It proposes what it calls the first end-to-end tolerance-aware optimization framework: every training iteration randomly perturbs each lens surface by tolerances sampled from a Gaussian, traces rays through the perturbed system, and back-propagates the reconstruction loss into both the optical parameters and the decoder. The authors claim that the resulting design keeps its deblurring quality under random deviations, with over 2 dB PSNR improvement over conventional deep optics, and that the tolerance-aware design also raises manufacturing yield. The reason this matters is that it targets the design-to-manufacturing gap that keeps deep-optics lenses from being practical in volume production.

What carries the argument

The load-bearing object is a tolerance-aware differentiable ray tracer. In each forward pass, the four tolerance types are converted into per-surface spatial transformations—translation for decentration and central thickness, rotation for tilt, plus a curvature offset—and applied as an equivalent coordinate transform on the ray before intersection and refraction, keeping the whole pipeline differentiable. Training then minimizes three terms: a standard image-quality loss, a Spot loss that keeps the traced spot size within a reasonable range, and a PSF similarity loss that penalizes changes in the point spread function under random tolerances. The PSF-similarity term is what couples the optical design to the decoder's expectations, stabilizing the Monte Carlo training and forcing the lens to keep its encoding stable under perturbations.

What would settle it

Fabricate, say, 100 copies of the tolerance-optimized Lens 2, measure their actual decentration, tilt, thickness, and curvature errors, and compare the distribution of measured deblurring PSNR against the simulated 100-sample tolerance test in Table 2; if the measured performance spread matches the conventional design's spread rather than the tolerance-aware design's, the Gaussian independence model and the >2 dB claim are falsified. A cheaper preliminary check is to measure a real factory's tolerance statistics and test whether they fit the assumed normal distribution with the stated ranges.

Watch

Extended reading notes

Core claim

The central claim is that explicitly modeling manufacturing and assembly tolerances inside the differentiable ray tracer, and optimizing against them jointly with the decoder, produces a deep-optics system whose performance no longer collapses when the real lens deviates from the nominal design. Concretely, the paper shows that sampling decentration, tilt, central-thickness, and curvature errors per lens and per training iteration, rendering the perturbed point spread function map, and training the optics and reconstruction network together yields more than 2 dB higher average PSNR than the same pipeline trained without tolerances, when both are tested under random tolerances. It also shows that the perturbed ray tracing matches Zemax spot diagrams to within about 1 micrometer in root-mean-square spot size, that a tolerance-aware design has substantially better manufacturing yield, and that optimizing only the optics or only the decoder is not enough: the gain appears only when both are updated together. On the basis of its real-world experiment, the paper further claims that the optimized decoder is more robust to actual perturbations during image acquisition.

Load-bearing premise

The load-bearing assumption is that real manufacturing and assembly errors are well described by independent Gaussian deviations of each lens with the hand-set ranges in Table A1 (for example, ±0.04 mm decentration and ±0.05 degrees tilt), since the same distribution is used for training, evaluation, and yield analysis; if a factory's actual errors are larger, correlated, or non-Gaussian, the claimed >2 dB gain may not carry over to fabricated lenses.

Editorial extensions

If this is right

  • Under the paper's tolerance model, a deep-optics lens optimized with this framework keeps an average PSNR above 28 dB when a conventional design drops to around 26 dB on the paper's deblurring test set with random tolerances.
  • The same designs show higher manufacturing yield: for Lens 1, the tolerance-aware design reaches 26.26 dB PSNR with 90% confidence, versus 24.30 dB for the non-tolerant design.
  • Tolerance optimization by optics-only metrics, as done with Zemax, can degrade end-to-end deblurring because it ignores the encoder-decoder pairing; the paper's joint optimization avoids this mismatch, as seen for Lens 2 in its Table 2.
  • Optimizing only the decoder improves its robustness but leaves the design fragile, and optimizing only the optics also fails, so the framework's benefit requires joint optimization of both.
  • The two-stage recipe—pretrain without tolerances, then tolerance-aware fine-tuning—offers a stable path that protects the pre-trained design's image-quality goals while hardening the system against deviations.

Reading between the lines

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

  • A direct next test is physical fabrication: the paper's real-world experiment perturbs an off-the-shelf lens and optimizes only the decoder, so it never manufactures the tolerance-optimized lens designs; measuring actual point spread functions and tolerance statistics of fabricated copies of Lens 1 or Lens 2 would confirm or refute whether the Gaussian model and the claimed >2 dB gain survive real
  • Because the framework treats tolerances as differentiable random perturbations, the same recipe could be transferred to other fabrication errors, such as surface irregularity or refractive-index inhomogeneity, or to non-refractive optics like diffractive elements and metasurfaces where manufacturing errors have different distributions.
  • The PSF-similarity loss acts as a stability regularizer, so a plausible untested consequence is that it may also improve robustness to defocus, thermal drift, or other perturbations that alter the point spread function, not only assembly tolerances.
  • If the tolerance model is accurate, the proposed flow suggests an industrial recipe: start from a standard design that already meets image-quality goals, then harden it and its decoder against measured factory tolerance statistics rather than hand-set ranges.
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

3 major / 5 minor

Summary. The paper proposes a tolerance-aware deep optics framework that incorporates manufacturing and assembly tolerances (decentration, tilt, central thickness, curvature) into differentiable ray tracing, and jointly optimizes the lens design and the computational decoder. The method is evaluated in simulation on two lens designs against a non-tolerance-aware baseline and a Zemax-based tolerance optimization, and with a real-world experiment using an off-the-shelf lens. The main claims are improved deblurring robustness under tolerances, accurate tolerance modeling validated against Zemax, and a design flow that reduces the design-to-manufacturing gap.

Significance. The problem addressed is important: conventional deep optics assumes perfect fabrication, which creates a gap between simulation and physical systems. Incorporating multiple tolerance types into differentiable ray tracing is a sensible and practically relevant extension, and the Zemax comparison (Sec. 4.1) gives evidence that the perturbed ray tracing is physically accurate. The ablations (Secs. 5.1–5.3) are also useful, particularly the finding that joint optics-decoder optimization is needed for robustness. If the tolerance model were validated against real fabricated lenses, the framework would be a solid contribution to computational imaging. The paper also provides a useful analysis of loss terms and sampling numbers.

major comments (3)
  1. [Abstract; Sec. 4.2; Tab. 2] The abstract and Sec. 4.2 claim 'over 2dB improvement' in average deblurring performance, but the numbers in Tab. 2 show improvements of 0.36 dB for Lens1 (29.61 vs. 29.25) and 2.33 dB for Lens2 (28.08 vs. 25.75), averaging about 1.35 dB. This claim should be corrected to match the reported results, or the statistical basis for a 2 dB figure should be stated explicitly.
  2. [Sec. 4.3] The 'real-world experiment' does not fabricate or test the tolerance-optimized lens designs; it only retrains a decoder on an off-the-shelf lens and applies artificial perturbations to the acquired images. As a result, this experiment does not validate the central claim of design-to-manufacturing transfer. The section should be reframed as a decoder-robustness study, or supplemented with a fabricated-lens experiment that actually measures performance under real manufacturing deviations.
  3. [Sec. 3.1; Supp. A; Sec. 4.2] The tolerance perturbations are defined in Sec. 3.1 and Supp. A as independent clamped Gaussians with hand-set ranges (Tab. A1), and the evaluation in Sec. 4.2 samples from exactly the same distribution used in training. Consequently, the robustness gains in Tab. 2 are in-sample with respect to the assumed error model and do not yet establish robustness under measured manufacturing statistics. The authors should either validate the tolerance model against real fabrication measurements (e.g., measured decentration/tilt distributions of produced lenses) or explicitly qualify the central robustness claim as conditional on the assumed Gaussian model.
minor comments (5)
  1. [Eq. (5)] In Eq. (5), the same symbol Pλ,f is used for both the traced ray position and the averaged centroid; please use a distinct notation such as P̄λ,f for the centroid to avoid ambiguity.
  2. [Eq. (7)] Eq. (7) is missing an operator between PSFIdeal and PSFP erb; the intended convolution or correlation should be explicitly written.
  3. [Sec. 4.1] The claim that spot-size errors relative to Zemax are '< 1µm' is not backed by a quantitative table; please report the per-field RMS spot error values so the accuracy claim is verifiable.
  4. [Sec. 4.3] Please clarify whether the off-the-shelf lens used in the real-world experiment corresponds to Lens1 or Lens2, and if not, state explicitly that the physical experiment does not use the designed lens parameters.
  5. [Abstract] The phrase 'the first end-to-end tolerance-aware optimization framework' should be qualified, since Li et al. [22] and Zheng et al. [46] already address fabrication tolerances in deep optics; if the claim is restricted to refractive lenses, this restriction should be stated explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the shared train/test tolerance sampler is an external-validity limitation, not a circular derivation.

full rationale

The paper's derivation chain does not reduce to its own inputs. The tolerance ranges in Tab. A1 are assumed, not fitted to data: Sec. 3.1 states 'theta_Delta obeys the normal distribution, theta_Delta ~ N(0, max^2/9)' and the ranges are hand-set tolerances. During training, Sec. 3.3 samples N=64 tolerance patterns per iteration; during evaluation, Supp. B.2 samples randomized tolerances multiple times using the same perturbation model. This means the simulation-based robustness numbers are in-sample with respect to the assumed Gaussian tolerance model, but that is a standard train/test setup for a robustness objective, not a case where a predicted quantity is equivalent by construction to a fitted parameter. The paper does not claim to have measured real manufacturing statistics for the designed lenses, so the realism of the tolerance model is an external-validity concern, not circularity. The real-world experiment in Sec. 4.3 uses an off-the-shelf lens and optimizes only the decoder, so it does not validate the co-designed optics; this is a scope/overclaim issue, not a circular step. The abstract's 'over 2dB improvement' is not supported by the tabulated arithmetic (Lens1: 29.61 vs 29.25 = +0.36 dB; Lens2: 28.08 vs 25.75 = +2.33 dB; average about 1.35 dB), but an inaccurate summary number is not circularity. Self-citations, including [42] for the differentiable PSF-map rendering, are implementation details and do not carry a uniqueness theorem or load-bearing premise that would force the result. No step exhibits the reduction patterns: no self-definitional fitting, no fitted-input-called-prediction, no imported uniqueness theorem, and no renaming of a known result. The central contribution is an optimization framework with novel losses; even if the evaluation is conditional on the assumed tolerance distribution, the derivation is self-contained and not circular.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on hand-chosen tolerance ranges, a Gaussian independence assumption, geometric optics approximations, and loss weights; no new physical entities are introduced. The evaluation is in-sample with respect to the tolerance model.

free parameters (3)
  • Tolerance ranges (decentration, tilt, central thickness, curvature) = +/- 0.04 mm, +/- 0.05 deg, +/- 0.04 mm, +/- 0.3%
    Chosen by hand (Tab. A1) to define the Gaussian perturbation model; the entire optimization and evaluation are conducted against these ranges, so the claimed robustness is conditional on them.
  • Loss weights lambda_Spot and lambda_PSF = Not given numerically, described as dependent on lens structure
    Introduced in Eq. (4) and tuned by hand; the ablation (Tab. 4) shows they are necessary, so the result depends on these weights.
  • Number of sampled tolerance patterns N = 64 for training
    A hyperparameter chosen to balance memory and optimization; the ablation (Fig. 9) shows performance degrades with smaller N.
assumptions (3)
  • domain assumption Geometric ray tracing with Snell's law is an accurate model of the optical systems under study; diffraction and wave effects are ignored.
    The pipeline is built on the DeepLens differentiable ray tracer (Sec. 4) and validated only against Zemax geometric spot diagrams (Sec. 4.1), not against physical measurements of the designed lenses.
  • domain assumption Manufacturing and assembly tolerances are independent across lenses, independent across tolerance types, and follow a clipped Gaussian distribution with the specified ranges.
    Stated in Sec. 3.1 and the supplement (theta_delta ~ N(0, max^2/9)), with no empirical justification from measured fabrication statistics.
  • domain assumption The PSF-based spatially-variant convolution model adequately represents the imaging process for training and evaluation.
    The renderer uses 51x51 PSF kernels on an 8x8 grid (supplement B.1), which is a common approximation but limits fidelity for strongly aberrated or tolerated systems.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tolerance-Aware Deep Optics." pith.science (2026). https://pith.science/paper/J2BEEETR

@misc{pith2026250204719,
  author       = {Pith},
  title        = {Pith review of: Tolerance-Aware Deep Optics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J2BEEETR}},
  note         = {Machine review of arXiv:2502.04719}
}
read the original abstract

Deep optics has emerged as a promising approach by co-designing optical elements with deep learning algorithms. However, current research typically overlooks the analysis and optimization of manufacturing and assembly tolerances. This oversight creates a significant performance gap between designed and fabricated optical systems. To address this challenge, we present the first end-to-end tolerance-aware optimization framework that incorporates multiple tolerance types into the deep optics design pipeline. Our method combines physics-informed modelling with data-driven training to enhance optical design by accounting for and compensating for structural deviations in manufacturing and assembly. We validate our approach through computational imaging applications, demonstrating results in both simulations and real-world experiments. We further examine how our proposed solution improves the robustness of optical systems and vision algorithms against tolerances through qualitative and quantitative analyses. Code and additional visual results are available at openimaginglab.github.io/LensTolerance.

Figures

Figures reproduced from arXiv: 2502.04719 by the authors.

Figure 1
Figure 1. When subjected to random tolerances perturbations, the performance of conventional deep optics degraded severely, whereas [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The tilt, decentration and central thickness tolerances [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Tolerance-aware optimization for deep optics. We integrate tolerances into differentiable ray tracing. Every kind of tolerances are randomly sampled from its distribution N (0, ˜σ 2 i ), use ray tracing with tolerances to render perturbed spatially-variant PSF maps and simulated the imaging results by spatially-variant convolution, then noise is added to simulate sensor-captured images. These images are then passed … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Spot diagrams and RMS spot sizes produced by our [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Deblurring results comparison of deep optics with and without tolerance-aware optimization under tolerances perturbations. The [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Physical experiment results demonstrating the robustness of our tolerance-aware deep optics. Despite random tolerance pertur [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 9
Figure 9. Figure 9: Ablation study on the number of sampled tolerances [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 8
Figure 8. Figure 8: Qualitative analysis results for the decoder’s robust [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

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. Deep Learning for Optical Misalignment Diagnostics in Multi-Lens Imaging Systems

    physics.optics 2025-06 conditional novelty 6.0 of 10

    Deep learning models can infer per-lens misalignment errors from simulated optical measurements in multi-lens systems, with accuracy of about 0.03 mm and 0.01 degrees in the spot-diagram method.

Reference graph

Works this paper leans on

47 extracted references · 43 canonical work pages · cited by 1 Pith paper

  1. [1]

    Ntire 2017 challenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition workshops, pages 126–135, 2017. 5, 1

    work page 2017

  2. [2]

    Single-shot hyperspectral-depth imaging with learned diffractive optics

    Seung-Hwan Baek, Hayato Ikoma, Daniel S Jeon, Yuqi Li, Wolfgang Heidrich, Gordon Wetzstein, and Min H Kim. Single-shot hyperspectral-depth imaging with learned diffractive optics. In Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision, pages 2651–2660,

    work page

  3. [3]

    PhoCoLens: Photorealistic and Consistent Reconstruction in Lensless Imaging

    Xin Cai, Zhiyuan You, Hailong Zhang, Wentao Liu, Jinwei Gu, and Tianfan Xue. Phocolens: Photorealistic and con- sistent reconstruction in lensless imaging. arXiv preprint arXiv:2409.17996, 2024. 1, 4

    work page Pith review arXiv 2024

  4. [4]

    Thin on-sensor nanophotonic array cameras

    Praneeth Chakravarthula, Jipeng Sun, Xiao Li, Chenyang Lei, Gene Chou, Mario Bijelic, Johannes Froesch, Arka Ma- jumdar, and Felix Heide. Thin on-sensor nanophotonic array cameras. ACM Transactions on Graphics (TOG) , 42(6):1– 18, 2023. 3

    work page 2023

  5. [5]

    An introduction to total variation for image analysis.Theoretical foundations and nu- merical methods for sparse recovery, 9(263-340):227, 2010

    Antonin Chambolle, Vicent Caselles, Daniel Cremers, Mat- teo Novaga, Thomas Pock, et al. An introduction to total variation for image analysis.Theoretical foundations and nu- merical methods for sparse recovery, 9(263-340):227, 2010. 4

    work page 2010

  6. [6]

    Deep optics for monocu- lar depth estimation and 3d object detection

    Julie Chang and Gordon Wetzstein. Deep optics for monocu- lar depth estimation and 3d object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 10193–10202, 2019. 2, 3

    work page 2019

  7. [7]

    Extreme-quality computational imaging via degradation framework

    Shiqi Chen, Huajun Feng, Keming Gao, Zhihai Xu, and Yueting Chen. Extreme-quality computational imaging via degradation framework. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 2632– 2641, 2021. 4

    work page 2021

  8. [8]

    Optical aberrations correction in postprocess- ing using imaging simulation

    Shiqi Chen, Huajun Feng, Dexin Pan, Zhihai Xu, Qi Li, and Yueting Chen. Optical aberrations correction in postprocess- ing using imaging simulation. ACM Transactions on Graph- ics (TOG), 40(5):1–15, 2021. 2

    work page 2021

Show all 47 references
  1. [9]

    Computational optics for mobile terminals in mass production

    Shiqi Chen, Ting Lin, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Computational optics for mobile terminals in mass production. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(4):4245–4259, 2022. 2, 3

    2022

  2. [10]

    The differentiable lens: Compound lens search over glass surfaces and materials for object detection

    Geoffroi C ˆot´e, Fahim Mannan, Simon Thibault, Jean- Franc ¸ois Lalonde, and Felix Heide. The differentiable lens: Compound lens search over glass surfaces and materials for object detection. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recogniti...

    2023

  3. [11]

    Learned rotationally symmetric diffractive achromat for full-spectrum computa- tional imaging

    Xiong Dun, Hayato Ikoma, Gordon Wetzstein, Zhanshan Wang, Xinbin Cheng, and Yifan Peng. Learned rotationally symmetric diffractive achromat for full-spectrum computa- tional imaging. Optica, 7(8):913–922, 2020. 3

    2020

  4. [12]

    A formal evaluation of psnr as quality measurement parameter for image segmen- tation algorithms

    Fernando A Fardo, Victor H Conforto, Francisco C de Oliveira, and Paulo S Rodrigues. A formal evaluation of psnr as quality measurement parameter for image segmen- tation algorithms. arXiv preprint arXiv:1605.07116 , 2016. 6

    2016 arXiv

  5. [13]

    Statistical tolerancing for optics

    David P Forse. Statistical tolerancing for optics. In Specifi- cation, Production, and Testing of Optical Components and Systems, pages 18–27. SPIE, 1996. 2

    1996

  6. [14]

    Mamba: Linear-time sequence modeling with selective state spaces

    Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752, 2023. 2

    2023 arXiv

  7. [15]

    Depth estimation from a single image using deep learned phase coded mask.IEEE Transactions on Computational Imaging, 4(3):298–310, 2018

    Harel Haim, Shay Elmalem, Raja Giryes, Alex M Bronstein, and Emanuel Marom. Depth estimation from a single image using deep learned phase coded mask.IEEE Transactions on Computational Imaging, 4(3):298–310, 2018. 3

    2018

  8. [16]

    Design and tolerance of a free- form optical system for an optical see-through multi-focal- plane display

    Xinda Hu and Hong Hua. Design and tolerance of a free- form optical system for an optical see-through multi-focal- plane display. Applied Optics, 54(33):9990–9999, 2015. 2

    2015

  9. [17]

    Depth from defocus with learned optics for imaging and occlusion-aware depth esti- mation

    Hayato Ikoma, Cindy M Nguyen, Christopher A Metzler, Yi- fan Peng, and Gordon Wetzstein. Depth from defocus with learned optics for imaging and occlusion-aware depth esti- mation. In 2021 IEEE International Conference on Compu- tational Photography (ICCP), pages 1–12. IEEE, 2021. 2

    2021

  10. [18]

    Percep- tual losses for real-time style transfer and super-resolution

    Justin Johnson, Alexandre Alahi, and Li Fei-Fei. Percep- tual losses for real-time style transfer and super-resolution. In Computer Vision–ECCV 2016: 14th European Confer- ence, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part II 14, pages 694–711. Springer, 2016. 4

    2016

  11. [19]

    Minimalist vision with freeform pixels

    Jeremy Klotz and Shree K Nayar. Minimalist vision with freeform pixels. In European Conference on Computer Vi- sion, pages 329–346. Springer, 2025. 1

    2025

  12. [20]

    A realistic camera model for computer graphics

    Craig Kolb, Don Mitchell, and Pat Hanrahan. A realistic camera model for computer graphics. In Proceedings of the 22nd annual conference on Computer graphics and interac- tive techniques, pages 317–324, 1995. 4

    1995

  13. [21]

    Lens design

    Milton Laikin. Lens design. Crc Press, 2018. 2

    2018

  14. [22]

    Quantization-aware deep optics for diffractive snapshot hyperspectral imaging

    Lingen Li, Lizhi Wang, Weitao Song, Lei Zhang, Zhiwei Xiong, and Hua Huang. Quantization-aware deep optics for diffractive snapshot hyperspectral imaging. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 19780–19789, 2022. 2, 3

    2022

  15. [23]

    End-to-end learned single lens design using fast differentiable ray tracing

    Zongling Li, Qingyu Hou, Zhipeng Wang, Fanjiao Tan, Jin Liu, and Wei Zhang. End-to-end learned single lens design using fast differentiable ray tracing. Optics Letters, 46(21): 5453–5456, 2021. 1, 2

    2021

  16. [24]

    Optical design and tolerancing of freeform surfaces using anisotropic radial basis functions

    Milan Maksimovic. Optical design and tolerancing of freeform surfaces using anisotropic radial basis functions. Optical Engineering, 55(7):071203–071203, 2016. 2

    2016

  17. [25]

    Designing easily manufactured lenses us- ing a global method

    James P McGuire. Designing easily manufactured lenses us- ing a global method. In International Optical Design Con- ference, page TuA6. Optica Publishing Group, 2006. 4

    2006

  18. [26]

    Deep optics for single-shot high-dynamic- range imaging

    Christopher A Metzler, Hayato Ikoma, Yifan Peng, and Gor- don Wetzstein. Deep optics for single-shot high-dynamic- range imaging. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 1375– 1385, 2020. 2, 3 9

    2020

  19. [27]

    Description and tolerance analysis of freeform sur- face figure error using specific-probability-distributed gaus- sian radial basis functions

    Junhao Ni, Tong Yang, Yue Liu, Dewen Cheng, and Yongtian Wang. Description and tolerance analysis of freeform sur- face figure error using specific-probability-distributed gaus- sian radial basis functions. Optics express, 27(22):31820– 31839, 2019. 2

    2019

  20. [28]

    A new approach to the simulation of optical manufacturing processes

    Don Oinen. A new approach to the simulation of optical manufacturing processes. In Optical Fabrication and Test- ing, page JWC3. Optica Publishing Group, 1990. 2

    1990

  21. [29]

    Learned large field-of-view imaging with thin-plate optics

    Yifan Peng, Qilin Sun, Xiong Dun, Gordon Wetzstein, Wolf- gang Heidrich, and Felix Heide. Learned large field-of-view imaging with thin-plate optics. ACM Trans. Graph., 38(6): 219–1, 2019. 1, 2

    2019

  22. [30]

    End-to-end optimization of optics and image processing for achromatic extended depth of field and super- resolution imaging

    Vincent Sitzmann, Steven Diamond, Yifan Peng, Xiong Dun, Stephen Boyd, Wolfgang Heidrich, Felix Heide, and Gor- don Wetzstein. End-to-end optimization of optics and image processing for achromatic extended depth of field and super- resolution imaging. ACM Transactions on Graph...

    2018

  23. [31]

    End-to-end optimization of optics and image processing for achromatic extended depth of field and super- resolution imaging

    Vincent Sitzmann, Steven Diamond, Yifan Peng, Xiong Dun, Stephen Boyd, Wolfgang Heidrich, Felix Heide, and Gor- don Wetzstein. End-to-end optimization of optics and image processing for achromatic extended depth of field and super- resolution imaging. ACM Transactions on Graph...

    2018

  24. [32]

    Learning rank-1 diffractive optics for single-shot high dynamic range imaging

    Qilin Sun, Ethan Tseng, Qiang Fu, Wolfgang Heidrich, and Felix Heide. Learning rank-1 diffractive optics for single-shot high dynamic range imaging. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1386–1396, 2020. 1, 2, 3

    2020

  25. [33]

    End-to-end complex lens design with differen- tiable ray tracing

    Qilin Sun, Congli Wang, Fu Qiang, Dun Xiong, and Heidrich Wolfgang. End-to-end complex lens design with differen- tiable ray tracing. ACM Trans. Graph, 40(4):1–13, 2021. 1, 3

    2021

  26. [34]

    Optical design tolerancing, 2015

    Synopsys. Optical design tolerancing, 2015. Avail- able: https : / / www . synopsys . com / content / dam / synopsys / optical - solutions / documents / whitepapers / optical - design - tolerancing.pdf. 2

    2015

  27. [35]

    Neural nano-optics for high-quality thin lens imag- ing

    Ethan Tseng, Shane Colburn, James Whitehead, Luocheng Huang, Seung-Hwan Baek, Arka Majumdar, and Felix Heide. Neural nano-optics for high-quality thin lens imag- ing. Nature communications, 12(1):6493, 2021. 3

    2021

  28. [36]

    Differentiable compound optics and processing pipeline op- timization for end-to-end camera design

    Ethan Tseng, Ali Mosleh, Fahim Mannan, Karl St-Arnaud, Avinash Sharma, Yifan Peng, Alexander Braun, Derek Nowrouzezahrai, Jean-Francois Lalonde, and Felix Heide. Differentiable compound optics and processing pipeline op- timization for end-to-end camera design. ACM Transaction...

    2021

  29. [37]

    do: A differ- entiable engine for deep lens design of computational imag- ing systems

    Congli Wang, Ni Chen, and Wolfgang Heidrich. do: A differ- entiable engine for deep lens design of computational imag- ing systems. IEEE Transactions on Computational Imaging, 8:905–916, 2022. 2, 3, 4, 5

    2022

  30. [38]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Si- moncelli. Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing, 13(4):600–612, 2004. 6

    2004

  31. [39]

    Inference in artifi- cial intelligence with deep optics and photonics

    Gordon Wetzstein, Aydogan Ozcan, Sylvain Gigan, Shan- hui Fan, Dirk Englund, Marin Solja ˇci´c, Cornelia Denz, David AB Miller, and Demetri Psaltis. Inference in artifi- cial intelligence with deep optics and photonics. Nature, 588 (7836):39–47, 2020. 3

    2020

  32. [40]

    Phase- cam3d—learning phase masks for passive single view depth estimation

    Yicheng Wu, Vivek Boominathan, Huaijin Chen, Aswin Sankaranarayanan, and Ashok Veeraraghavan. Phase- cam3d—learning phase masks for passive single view depth estimation. In 2019 IEEE International Conference on Com- putational Photography (ICCP), pages 1–12. IEEE, 2019. 3

    2019

  33. [41]

    Aberration-aware depth-from-focus

    Xinge Yang, Qiang Fu, Mohamed Elhoseiny, and Wolfgang Heidrich. Aberration-aware depth-from-focus. IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2023. 2, 3

    2023

  34. [42]

    Curriculum learning for ab initio deep learned refractive optics

    Xinge Yang, Qiang Fu, and Wolfgang Heidrich. Curriculum learning for ab initio deep learned refractive optics. Nature communications, 15(1):6572, 2024. 1, 2, 3, 5

    2024

  35. [43]

    Opticstudio, 2013

    Zemax. Opticstudio, 2013. Available: https://www. zemax.com/. 2, 5, 6

    2013

  36. [44]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 6

    2018

  37. [45]

    Large depth-of-field ultra-compact micro- scope by progressive optimization and deep learning.Nature Communications, 14(1):4118, 2023

    Yuanlong Zhang, Xiaofei Song, Jiachen Xie, Jing Hu, Jiawei Chen, Xiang Li, Haiyu Zhang, Qiqun Zhou, Lekang Yuan, Chui Kong, et al. Large depth-of-field ultra-compact micro- scope by progressive optimization and deep learning.Nature Communications, 14(1):4118, 2023. 2

    2023

  38. [46]

    Neu- ral lithography: Close the design-to-manufacturing gap in computational optics with a’real2sim’learned photolithogra- phy simulator

    Cheng Zheng, Guangyuan Zhao, and Peter TC So. Neu- ral lithography: Close the design-to-manufacturing gap in computational optics with a’real2sim’learned photolithogra- phy simulator. arXiv preprint arXiv:2309.17343, 2023. 3

    2023 arXiv

  39. [47]

    Optical degradation correction of manufacturing-perturbed glass-plastic hybrid lens systems via a joint hardware- software optimization framework

    Jingwen Zhou, Bingkun Chen, Jiapu Yan, Zheng Ren, Wen- guan Zhang, Huajun Feng, Yueting Chen, and Meijuan Bian. Optical degradation correction of manufacturing-perturbed glass-plastic hybrid lens systems via a joint hardware- software optimization framework. Optics Express, 32...

    2024

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.