Exploring generative design AI tools for astronomical instrumentation: a CubeSat chassis case study

Alastair Macleod; Anjali Bhatt; Anmol Goyal; Bradley Frank; David Isherwood; David Lunney; Graham Wilks; Katherine Morris; Marcell Westsik; Noah Schwartz

REVIEW 3 major objections 7 minor 18 references

Exploring generative design AI tools for astronomical instrumentation: a CubeSat chassis case study

T0 review · 3 major / 7 minor · reviewed 2026-07-31 · grok-4.5

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

Pith's one-line read Generative design AI helps brainstorm CubeSat chassis layouts fast, but only pays off on high-value parts.

desk verdict Honest Fusion 360 chassis case study with useful workflow notes; the “clear brainstorming benefit” claim is thinner than the abstract suggests because there is no head-to-head vs the conventional design. read the letter →

arxiv 2607.28217 v1 pith:2BNMU27V submitted 2026-07-30 astro-ph.IM

Younes Chahid , Tassos Aretos , Will Cochrane , Katherine Morris , David Isherwood , Zeshan Ali , Graham Wilks , Noah Schwartz

show 7 more authors

Anmol Goyal Marcell Westsik Alastair Macleod Steve Watson Anjali Bhatt David Lunney Bradley Frank

This is my paper · ORCID

classification astro-ph.IM

keywords ArtificialintelligenceGenerativeDesignAstronomyInstrumentationCubeSatAdditiveManufacturingMechanicalSpaceEarthObservation

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

Astronomical instruments packed into CubeSats must stay stiff, light, and manufacturable under tight thermal, vibration, and schedule constraints. This paper walks through applying a commercial AI-plus-finite-element generative design tool to the chassis of the ADOT deployable optical CubeSat. With preserve and obstacle geometry, loads, materials, manufacturing rules, mass and stiffness targets set, the solver produced more than eighty candidate designs in under two hours, all meeting a 100 Hz first-mode floor. The authors argue the real value is early multi-constraint brainstorming and manufacturing-path scouting, not a finished part: the geometries remain black-box, not production-ready, and the setup cost only justifies itself on critical, high-value components.

What carries the argument

Constraint-driven generative design in Autodesk Fusion 360: preserve and obstacle geometry, structural loads and fixed supports, material and process rules (additive, milling, casting), plus objectives (max stiffness, mass under 1.5 kg, first mode above 100 Hz, symmetry) that drive iterative FEA lightweighting across many cloud-solved candidates.

What would settle it

A controlled redesign of the same ADOT chassis that reports setup time, final mass, stiffness at critical interfaces, manufacturability fixes, and total cost against the published conventional chassis—and checks whether manufacturing rules (for example additive overhang) are actually met without heavy post-processing.

Watch

Extended reading notes

Core claim

An AI and FEA generative design workflow, applied to the ADOT 6U CubeSat chassis with explicit design-space, load, material, manufacturing, and objective constraints, can generate dozens of mass- and stiffness-oriented concepts in hours and is clearly useful in early brainstorming—yet black-box solvers, non-manufacturing-ready geometry, and time-consuming setup limit tangible gain to high-value mechanical components.

Load-bearing premise

The chosen preserve and obstacle layout, connector load idealizations, generous displacement bound, and single commercial solver setup are enough to judge the tool’s real value for chassis design, even without a head-to-head mass or cost comparison to the conventional baseline.

Editorial extensions

If this is right

Early instrument structure trade studies can sample many material and process combinations before locking a manufacturing path.
Generative output still needs engineer-led CAD cleanup, re-FEA, GD&T, and work-holding or support features before build.
Setup effort is only worth it for critical, high-value parts, not routine brackets.
Wider adoption needs more transparent or open solvers and better design-for-manufacture preserve geometry.

Reading between the lines

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

Optical payload interfaces may need tighter local displacement targets than the global 0.1 mm used here before generative results can replace conventional optical-bench intuition.
Case studies that publish editable constraint packs and before/after mass-stiffness numbers would do more for adoption than more qualitative workflow tours.
The same preserve-obstacle pattern could transfer to other deployable space optics where stowed volume and launch modes dominate the structure.

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.

Desk Editor's Note

read the letter

Punchline: this is a candid experience report on running Autodesk Fusion 360 generative design on the ADOT 6U chassis, not a new method. What you get is a documented constraint setup (preserve/obstacle connectors, multi-process AM/mill/cast rules, mass <1.5 kg, f1 ≥ 100 Hz) and a cloud of >80 candidates in under two hours, plus an unusually frank limitations section.

What is actually new is the packaged case for astronomical instrumentation practice—the specific preserve/obstacle layout, manufacturing-parameter choices, and the adoption judgment that tangible gain is mostly early brainstorming on high-value parts. Algorithmically nothing is new; McClelland, Vlah, Buonamici, and this group’s own prior AM/field-driven chassis papers already cover the territory. Credit where due: the workflow figures are clear, the self-reported limits (black-box commercial solver, non-manufacturing-ready organic geometry, setup cost, AM overhang not respected) match what the results show, and the citation pattern is appropriate rather than circular. Objectives are inputs; the solver scores geometries against them. Fine for this genre.

Soft spots in proportion. The stress-test lands on the strongest claim. “Clear benefit” is supported mainly by design count and meeting mass/modal floors under gravity-plus-connector loads and a generous global displacement bound (<0.1 mm). Launch vibration spectra and thermal cases flagged in the intro are not in the load set. Manufacturing constraints were partly violated. Critically, mass/stiffness/modal numbers for the cited conventional baseline never appear on Fig. 4 or in a table, so you cannot judge improvement versus ordinary CAD intuition. That does not make the paper dishonest—it makes the benefit language a notch too strong for the evidence. Free parameters (mass target, displacement bound, tool diameters, draft, etc.) are normal user knobs, not hidden fitting.

Who it is for: mech designers in CubeSat/instrument groups deciding whether generative CAD is worth the setup tax. Not for people hunting algorithms or reproducible open solvers. Math/data are qualitative FEA screenshots and a Pareto scatter from a proprietary cloud; soundness is mid for that claim class. I would bring it only if the reading group does small-sat structures. I would not cite it unless I were writing a methods survey on generative CAD adoption. A serious editor should still send it to referees rather than desk-reject: scoped instrumentation case studies with honest limits deserve review time, ideally with a request for quantitative baseline comparison and shared constraint/geometry artifacts. Engage if you build chassis; otherwise file and move on.

Referee Report

3 major / 7 minor

Summary. This paper presents a practical case study of Autodesk Fusion 360’s AI/FEA generative design applied to the structural chassis of the ADOT 6U CubeSat. The authors define preserve/obstacle geometries, connector loads (gravity and payload weight), three aluminum alloys tied to AM/milling/casting constraints, and objectives of maximized stiffness, mass <1.5 kg, and first modal frequency ≥100 Hz. Cloud solving produced >80 candidate geometries in under two hours; all reported designs met the modal-frequency floor under the stated loads, and one AM candidate is taken through export, light post-processing, re-FEA, and re-integration into the assembly. The discussion argues that such tools are clearly useful in early brainstorming for multi-constrained astronomical structures, while remaining limited by black-box solvers, non-manufacturing-ready geometry, and setup cost that confines tangible gain to high-value parts.

Significance. Disseminated, instrument-specific case studies of generative design in astronomical mechanical engineering remain scarce, so a concrete CubeSat chassis workflow with explicit constraint settings has real community value. The paper is appropriately scoped as a qualitative evaluation rather than a new algorithm or flight-qualified design. Strengths include a clear end-to-end workflow description, multi-process manufacturing constraints, a multi-objective result cloud (Fig. 4), and an honest limitations section (black-box nature, post-processing burden, setup time). The central claim is modest and useful if supported; it does not require machine-checked proofs or open code to be publishable, but it does require enough quantitative anchoring against the conventional baseline and against the load/manufacturing cases the introduction itself flags as mission-critical.

major comments (3)

[Results §3; Fig. 4; Discussion §4] Results §3 and Fig. 4 report that generated designs meet mass <1.5 kg and f1 ≥100 Hz, and the abstract/discussion assert “clear benefits” in early brainstorming. The conventional reference chassis (Fig. 2; Refs. 7–8) is never given comparable mass, stiffness, or first-mode values on the same plot or in a table. Without that head-to-head, “benefit” reduces to rapid enumeration inside the authors’ constraint box and does not yet underwrite even a scoped brainstorming-utility claim relative to ordinary CAD practice. Add baseline metrics (and, if available, a conventional redesign effort/time note) or explicitly temper the claim to “rapid multi-constraint enumeration.”
[Introduction §1; Methodology §2.1; Results §3] Introduction §1 stresses launch loads, thermal variations, and optical-alignment stiffness as the drivers for chassis design. Methodology §2.1 applies only fixed displacements on release-spring connectors plus gravity and payload weight on remaining connectors, with a generous global displacement bound (<0.1 mm). No launch vibration spectra, quasi-static load factors, or thermal cases appear. For a CubeSat optical chassis this is a load-bearing gap: the claim that the tool helps with “multi-constrained mechanical structures” for astronomical instrumentation needs either (i) at least one representative dynamic/thermal case in the generative setup or post-check, or (ii) a clear statement that the study is limited to static gravity/payload idealizations and that launch/thermal readiness is future work.
[Results §3; Fig. 5; Fig. 6] Results §3 states that manufacturing constraints such as the AM overhang angle were not respected, attributed to preserve/obstacle arrangement, yet an AM design is still selected for the follow-up path in Fig. 6 and used to illustrate the workflow’s value. If a primary manufacturing constraint is violated, the paper should either (a) show a feasible redesign of preserve/obstacle geometry that recovers manufacturability, (b) discard non-compliant candidates from the “successful” set, or (c) quantify how often manufacturing constraints failed across the >80 designs. Leaving violated constraints inside the success narrative weakens the manufacturability half of the evaluation.

minor comments (7)

[Abstract; Introduction §1] Abstract and §1: “and and respecting” — duplicate “and”.
[Methodology §2.1] §2.1 Design Objectives: “Minimum firs modal frequency” — typo for “first”.
[Discussion §4] §4: “primarly”, “chosen chosen algorithm” — typos.
[Fig. 4] Fig. 4 caption and body: clarify what “opaque design values” means (subset for Fig. 5?) so the Pareto cloud is readable without the figure alone.
[Methodology §2.1] Fig. 5 / milling tool axis: text refers to “seen in 5” without “Fig.”; keep figure callouts consistent.
[Introduction §1; Discussion §4] Keywords and title use “generative design AI”; the body is almost entirely commercial topology/generative FEA in Fusion 360. A short sentence distinguishing marketing “AI” from the underlying optimization would help non-specialist readers.
[Methodology §2] Refs. 7–8 supply the conventional chassis; a one-line quantitative summary of that baseline in §2 would make the case study self-contained.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: engineering case study of a commercial optimizer, not a first-principles prediction chain

full rationale

This paper is a practical workflow evaluation of Autodesk Fusion 360 generative design on the ADOT CubeSat chassis. Design objectives (max stiffness, mass <1.5 kg, f1 ≥ 100 Hz), preserve/obstacle geometry, loads, materials, and manufacturing criteria are explicit user inputs; the solver enumerates geometries scored against those same inputs. Meeting the modal-frequency floor or mass target is therefore the ordinary behavior of a constrained optimizer, not a claimed independent prediction or derivation. Self-citations (Schwartz et al. 2022; Morris et al. 2024) supply mission context and the conventional baseline chassis figure; they do not underwrite a uniqueness theorem or force the qualitative adoption conclusions. There is no fitted parameter renamed as a forecast, no self-definitional identity presented as discovery, and no ansatz smuggled in via prior author work. Weaknesses noted by a skeptical reader (missing head-to-head metrics vs. the conventional chassis, partly violated AM overhang, simplified load set) are evidence-strength issues, not circularity of the derivation chain. Score 0 is appropriate.

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

The central adoption claim rests on standard structural FEA practice, NASA CubeSat modal guidance, author-chosen mass/displacement targets, idealized connector loads, and trust in a proprietary generative solver—not on new physical entities. Free parameters are engineering set-points that shape which designs appear ‘successful.’

free parameters (6)

mass_target = <1.5 kg
Design objective mass <1.5 kg is an author-chosen target that defines success in the generative runs.
global_displacement_bound = <0.1 mm
Additional stiffness filter global displacement <0.1 mm described as generous and stage-appropriate; directly selects the Fig. 5 family.
min_first_modal_frequency = 100 Hz
100 Hz floor taken from NASA GEVS CubeSat recommendation and imposed as hard objective.
AM_min_thickness_overhang_axis = 5 mm; 45°; Y+
Manufacturing knobs (5 mm min thickness, 45° overhang, Y+ print axis) hand-set and noted as not fully satisfied by outputs.
milling_tool_parameters = 10/40/60 mm
Tool diameter 10 mm, shoulder 40 mm, head 60 mm constrain reachable generative features.
casting_draft_thickness = 3°; 5 mm; Z
3° draft and 5 mm minimum thickness plus Z ejection direction set casting branch.

assumptions (5)

domain assumption Linear FEA with gravity and lumped payload connector loads plus fixed release-spring connectors adequately represents launch/operational structural demand for concept screening.
Loads and constraints defined in §2.1; no coupled thermo-elastic or full random-vibration spectrum in the generative loop.
domain assumption NASA GEVS-related guidance that first mode above ~100 Hz is an appropriate CubeSat qualification-oriented target for this study.
Cited as Ref. [9] and imposed in design objectives §2.1.
ad hoc to paper Autodesk Fusion 360 cloud generative design produces geometries whose FEA metrics are trustworthy enough to rank conceptual chassis options.
Entire results corpus is solver-native; discussion admits black-box commercial protection limits QC (§4).
domain assumption Preserve cylinders and deployed-payload obstacle bodies sufficiently encode interface and keep-out requirements, including need for line-of-sight between preserves.
Design space definition in §2.1 and Fig. 2; authors note arrangement limited manufacturability compliance.
ad hoc to paper Tangible project value of generative design is mostly confined to high-value parts because setup time is large relative to benefit on routine parts.
Core adoption conclusion in abstract and §4; qualitative cost-benefit judgment without timed baseline study.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Exploring generative design AI tools for astronomical instrumentation: a CubeSat chassis case study." pith.science (2026). https://pith.science/paper/2BNMU27V

@misc{pith2026260728217,
  author       = {Pith},
  title        = {Pith review of: Exploring generative design AI tools for astronomical instrumentation: a CubeSat chassis case study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2BNMU27V}},
  note         = {Machine review of arXiv:2607.28217}
}

read the original abstract

Generative design artificial intelligence (AI) tools are currently used in multiple scientific fields, yet their adoption in mechanical engineering computer-aided design (CAD) remains limited due to a lack of disseminated case studies, limited availability of accessible tools, insufficient training in CAD data, the absence of universal editable file formats and more. Mechanical design for astronomical instrumentation faces increasing complexity in thermal, vibrational, and mechanical requirements alongside tight project deadlines. This paper presents a practical evaluation of an AI and FEA based generative design tool applied to chassis design for the Active Deployable Optical Telescope (ADOT) CubeSat mission. Our analysis showcases the workflow steps including the setting of design, manufacturing and objective constraints. This study also shows the clear benefits of these types of tools, especially in the early brainstorming stages of multi-constrained mechanical structures, while also highlighting clear limitations like their black-box nature, the non-manufacturing-ready state of the results, and the time-consuming setup, limiting the tangible gain of these tools to high-value mechanical components.

Figures

Figures reproduced from arXiv: 2607.28217 by the authors.

**Figure 2.** Stowed and deployed CubeSat7, 8 configurations (Top image credit: Morris, et al,. (2024); Breakdown and simulation input geometry (down) [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** AI generative design process. – Structural Loads: gravity and weight of payload parts applied on remaining connectors also shown in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Maximum displacement versus first mode modal frequency for more than 80 generated designs. The opaque [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Example of constraints that maximise stiffness, mass target of 1.5 kg and global displacement [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: Structural loads and max displacement FEA simulation (up) applied on generated design (down). [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references

[1]

Chahid, Y., Atkins, C., Lister, G., Tuck, R., Watson, S., Morris, K., Isherwood, D., Strachan, J., Harman, J., Chartsiriwattana, P., et al., ``Additive manufacturing applications in astronomy: a review,'' Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI 13100 , 485--510 (2024)

2024
[2]

McClelland, R., ``Generative design and digital manufacturing: using ai and robots to build lightweight instrument structures,'' in [ Current Developments in Lens Design and Optical Engineering XXIII 0.1em ] , 12217 , 141--148, SPIE (2022)

2022
[3]

Vlah, D., Z avbi, R., and Vuka s inovi \'c , N., ``Evaluation of topology optimization and generative design tools as support for conceptual design,'' in [ Proceedings of the design society: DESIGN conference 0.1em ] , 1 , 451--460, Cambridge University Press (2020)

2020
[4]

and Vaerewyck, M., ``Evolved structures guide for gsfc applications rev b,'' (2025)

McClelland, R. and Vaerewyck, M., ``Evolved structures guide for gsfc applications rev b,'' (2025)

2025
[5]

Buonamici, F., Carfagni, M., Furferi, R., Volpe, Y., and Governi, L., ``Generative design: An explorative study.,'' Computer-Aided Design & Applications 18 (1) (2021)

2021
[6]

Schwartz, N., Brzozowski, W., Ali, Z., Milanova, M., Morris, K., Bond, C., Keogh, J., Harvey, D., Bissell, L., Sauvage, J.-F., et al., ``6u cubesat deployable telescope for optical earth observation and astronomical optical imaging,'' in [ Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave 0.1em ] , 12180 , 1097--1108, SPIE (2022)

2022
[7]

Morris, K., Schwartz, N., Ali, Z., Milanova, M., Miller, C., Zhou, J., Isherwood, D., Brzozowski, W., Gillespie, R., Bond, C., et al., ``Accurate deployable mirrors for small optical cubesat telescope: design and laboratory validations,'' in [ Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave 0.1em ] , 13092 , 1075--1086, S...

2024
[8]

Morris, K., Atkins, C., Chahid, Y., Ali, Z., and Milanova, M., ``Field driven design of additively manufactured cubesat chassis,'' in [ Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI 0.1em ] , 13100 , 1508--1518, SPIE (2024)

2024

Show all 18 references

[9]

NASA Goddard Space Flight Center , ``General environmental verification standard (gevs) for gsfc flight programs and projects,'' Tech. Rep. GSFC-STD-7000B, NASA Goddard Space Flight Center, Greenbelt, MD (Apr. 2021). Approved 04-28-2021, supersedes GSFC-STD-7000A

2021
[10]

Current Developments in Lens Design and Optical Engineering XXIII , volume=

Generative design and digital manufacturing: using AI and robots to build lightweight instrument structures , author=. Current Developments in Lens Design and Optical Engineering XXIII , volume=. 2022 , organization=

2022
[11]

Proceedings of the design society: DESIGN conference , volume=

Evaluation of topology optimization and generative design tools as support for conceptual design , author=. Proceedings of the design society: DESIGN conference , volume=. 2020 , organization=

2020
[12]

Evolved Structures Guide for GSFC Applications Rev B , author=
[13]

, author=

Generative design: An explorative study. , author=. Computer-Aided Design & Applications , volume=
[14]

Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave , volume=

6U CubeSat deployable telescope for optical Earth observation and astronomical optical imaging , author=. Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave , volume=. 2022 , organization=

2022
[15]

Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave , volume=

Accurate deployable mirrors for small optical CubeSat telescope: design and laboratory validations , author=. Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave , volume=. 2024 , organization=

2024
[16]

Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI , volume=

Field driven design of additively manufactured CubeSat chassis , author=. Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI , volume=. 2024 , organization=

2024
[17]

2021 , month = apr, address =

General Environmental Verification Standard (GEVS) for GSFC Flight Programs and Projects , author =. 2021 , month = apr, address =

2021
[18]

Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI , volume=

Additive manufacturing applications in astronomy: a review , author=. Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI , volume=. 2024 , publisher=

2024

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.

[1] [1]

Chahid, Y., Atkins, C., Lister, G., Tuck, R., Watson, S., Morris, K., Isherwood, D., Strachan, J., Harman, J., Chartsiriwattana, P., et al., ``Additive manufacturing applications in astronomy: a review,'' Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI 13100 , 485--510 (2024)

2024

[2] [2]

McClelland, R., ``Generative design and digital manufacturing: using ai and robots to build lightweight instrument structures,'' in [ Current Developments in Lens Design and Optical Engineering XXIII 0.1em ] , 12217 , 141--148, SPIE (2022)

2022

[3] [3]

Vlah, D., Z avbi, R., and Vuka s inovi \'c , N., ``Evaluation of topology optimization and generative design tools as support for conceptual design,'' in [ Proceedings of the design society: DESIGN conference 0.1em ] , 1 , 451--460, Cambridge University Press (2020)

2020

[4] [4]

and Vaerewyck, M., ``Evolved structures guide for gsfc applications rev b,'' (2025)

McClelland, R. and Vaerewyck, M., ``Evolved structures guide for gsfc applications rev b,'' (2025)

2025

[5] [5]

Buonamici, F., Carfagni, M., Furferi, R., Volpe, Y., and Governi, L., ``Generative design: An explorative study.,'' Computer-Aided Design & Applications 18 (1) (2021)

2021

[6] [6]

Schwartz, N., Brzozowski, W., Ali, Z., Milanova, M., Morris, K., Bond, C., Keogh, J., Harvey, D., Bissell, L., Sauvage, J.-F., et al., ``6u cubesat deployable telescope for optical earth observation and astronomical optical imaging,'' in [ Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave 0.1em ] , 12180 , 1097--1108, SPIE (2022)

2022

[7] [7]

Morris, K., Schwartz, N., Ali, Z., Milanova, M., Miller, C., Zhou, J., Isherwood, D., Brzozowski, W., Gillespie, R., Bond, C., et al., ``Accurate deployable mirrors for small optical cubesat telescope: design and laboratory validations,'' in [ Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave 0.1em ] , 13092 , 1075--1086, S...

2024

[8] [8]

Morris, K., Atkins, C., Chahid, Y., Ali, Z., and Milanova, M., ``Field driven design of additively manufactured cubesat chassis,'' in [ Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation VI 0.1em ] , 13100 , 1508--1518, SPIE (2024)

2024