pith. sign in

arxiv: 2604.08997 · v1 · submitted 2026-04-10 · 💻 cs.CE · math.OC

Scale-invariant projection optimization in tomographic volumetric additive manufacturing

Seungpyo Woo , Sangyup Lee , Hayden K. Taylor This is my paper

Pith reviewed 2026-05-10 17:24 UTC · model grok-4.3

classification 💻 cs.CE math.OC
keywords tomographic volumetric additive manufacturingprojection optimizationscale-invariantlinear-fractional programmingCharnes-Cooper transformationdose spillagetarget conformity3D printing
0
0 comments X

The pith

Scale-invariant projection optimization decouples shape from dose scaling to balance fidelity and spillage in TVAM.

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

This paper introduces a scale-invariant projection optimization framework for tomographic volumetric additive manufacturing that focuses on projection shape while treating absolute dose as a separate parameter. It formulates the problem as a linear-fractional program using normalized metrics for target conformity and dose spillage, then applies the Charnes-Cooper transformation to convert it into a standard linear program. Two practical cases are defined for control: minimizing spillage under material tolerance limits or maximizing conformity under inhibition constraints. A matrix-free primal-dual solver supports large-scale use, and tests confirm the method yields explicit fidelity-separation trade-offs even when accounting for 3D optical blur.

Core claim

Projection patterns for TVAM can be designed by optimizing normalized conformity and spillage ratios rather than absolute intensities, allowing the projection shape to be determined independently of overall dose level through a linear-fractional program that converts to a linear program, thereby producing controlled trade-offs between in-part accuracy and unwanted exposure outside the target.

What carries the argument

The SiPO framework, which rests on normalized conformity and spillage metrics cast as a linear-fractional program and transformed into a linear program via the Charnes-Cooper method.

If this is right

  • The framework produces explicit, tunable trade-offs between target fidelity and unwanted exposure.
  • It continues to perform under forward models that include 3D optical blurring.
  • The two deterministic cases give direct process controls for either strict material limits or hard inhibition zones.
  • Matrix-free solving makes the approach practical for high-resolution 3D volumes.

Where Pith is reading between the lines

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

  • Manufacturers could adjust print recipes by changing only dose while keeping the same projection shape, simplifying calibration.
  • The same normalized-metric approach may apply to other projection-based processes such as stereolithography variants where stray light control matters.
  • Coupling the solver with real-time sensor data from the build volume could enable adaptive correction during printing.

Load-bearing premise

The normalized conformity and spillage metrics plus the linear-fractional formulation are enough to represent the real physical needs of accurate part formation and minimal stray exposure inside actual TVAM machines.

What would settle it

Fabricate test objects with the optimized projections on physical TVAM hardware, measure the resulting part geometry and unintended cured regions, and check whether the observed fidelity-spillage pairs match the numerical trade-off curves predicted by the model.

read the original abstract

Tomographic volumetric additive manufacturing (TVAM) requires projection patterns that achieve high in-part fidelity while suppressing unintended exposure outside the target. We present a scale-invariant projection optimization framework (SiPO) that decouples projection shape from absolute dose scaling. The method formulates projection design as a linear-fractional program based on normalized conformity and spillage metrics, which is converted into a linear program via the Charnes-Cooper transformation. Two practical deterministic cases are introduced for process control: minimizing dose spillage under strict material tolerances and maximizing target conformity under hard inhibition constraints. A matrix-free primal-dual hybrid gradient solver enables large-scale implementation. Numerical results demonstrate that the framework provides a clear trade-off between target fidelity and process separation and remains effective under 3D blur-aware forward models.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a scale-invariant projection optimization (SiPO) framework for tomographic volumetric additive manufacturing (TVAM). It formulates projection pattern design as a linear-fractional program using normalized conformity and spillage metrics derived from a 3D blur-aware forward model of dose delivery. The linear-fractional program is converted to a standard linear program via the Charnes-Cooper transformation. Two deterministic cases are defined for process control: minimizing spillage subject to material tolerances and maximizing conformity subject to hard inhibition constraints. A matrix-free primal-dual hybrid gradient solver is employed for scalability. Numerical experiments are reported to illustrate trade-offs between target fidelity and process separation, with claims of effectiveness under blur-aware models.

Significance. If the numerical results hold under the stated assumptions, the work provides a mathematically clean, scale-invariant approach to projection optimization in TVAM that decouples shape from absolute dose scaling. The application of the Charnes-Cooper transformation and the matrix-free solver are practical strengths that could enable larger-scale implementations. The framework offers a clear, deterministic way to explore fidelity-separation trade-offs, which could be useful for process control if the normalized metrics prove predictive of physical outcomes.

major comments (2)
  1. [Abstract / Numerical results] Abstract and numerical results section: The central claim that the framework 'remains effective under 3D blur-aware forward models' and provides a 'clear trade-off between target fidelity and process separation' rests on the normalized conformity and spillage metrics. These metrics are linear ratios of integrated dose quantities; however, real TVAM involves nonlinear material responses (e.g., oxygen inhibition, threshold polymerization), which are not modeled. The manuscript does not demonstrate that optimizing these linear ratios guarantees the claimed high in-part fidelity or suppressed unintended exposure once the physical nonlinearities are present.
  2. [Formulation / Numerical results] Formulation section (linear-fractional program): The two deterministic cases (min spillage under tolerances, max conformity under constraints) inherit the same limitation as the metrics themselves. No sensitivity analysis or comparison against a nonlinear forward model is provided to show that the Charnes-Cooper transformed LP solutions remain near-optimal or feasible when the actual dose-response deviates from the linear blur model.
minor comments (2)
  1. [Introduction / Methods] Notation for the normalized metrics (conformity and spillage) should be defined explicitly with equations early in the manuscript rather than assumed from context.
  2. [Solver section] The manuscript would benefit from a brief discussion of how the matrix-free solver handles the specific structure of the TVAM projection operator (e.g., memory requirements for large voxel grids).

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for their constructive feedback on our manuscript. We address the two major comments point by point below, clarifying the scope of our linear model and acknowledging limitations where appropriate.

read point-by-point responses

  1. Referee: Abstract / Numerical results: The central claim that the framework 'remains effective under 3D blur-aware forward models' and provides a 'clear trade-off between target fidelity and process separation' rests on the normalized conformity and spillage metrics. These metrics are linear ratios of integrated dose quantities; however, real TVAM involves nonlinear material responses (e.g., oxygen inhibition, threshold polymerization), which are not modeled. The manuscript does not demonstrate that optimizing these linear ratios guarantees the claimed high in-part fidelity or suppressed unintended exposure once the physical nonlinearities are present.

    Authors: We agree that our framework is developed under a linear blur-aware forward model of dose accumulation, as detailed in the formulation and numerical sections. The normalized metrics and Charnes-Cooper transformation are defined and shown to be effective specifically within this linear setting, providing scale-invariance and the reported trade-offs. The abstract's claim of remaining effective under 3D blur-aware models refers to this linear model. We do not assert that the solutions guarantee performance under nonlinear responses such as threshold polymerization or oxygen inhibition. To address the concern, we will revise the discussion section to explicitly state this scope limitation and suggest future integration of nonlinear models (e.g., using SiPO outputs as warm starts for nonlinear optimization). revision: partial

  2. Referee: Formulation / Numerical results: The two deterministic cases (min spillage under tolerances, max conformity under constraints) inherit the same limitation as the metrics themselves. No sensitivity analysis or comparison against a nonlinear forward model is provided to show that the Charnes-Cooper transformed LP solutions remain near-optimal or feasible when the actual dose-response deviates from the linear blur model.

    Authors: The two cases are formulated directly from the linear-fractional program and solved via the matrix-free primal-dual hybrid gradient method under the linear model. No sensitivity analysis to nonlinear deviations is included, as the manuscript focuses on the mathematical properties and numerical behavior within the linear blur-aware setting. Conducting such analysis would require defining specific nonlinear dose-response functions and additional comparative experiments, which is outside the current scope. We view the linear model as a practical starting point for projection design but acknowledge that robustness under nonlinearities is unexamined here. revision: no

standing simulated objections not resolved
  • Demonstrating that the optimized projections remain effective or near-optimal under nonlinear material responses (e.g., oxygen inhibition or threshold effects) would require new modeling, sensitivity studies, and physical validation experiments not present in the manuscript.

Circularity Check

0 steps flagged

No significant circularity; standard transformation applied to new objective

full rationale

The paper formulates a new linear-fractional program using normalized conformity and spillage metrics derived from the forward model (dose = projection convolved with blur), then applies the standard Charnes-Cooper transformation to obtain an equivalent linear program. Two deterministic cases (min spillage under tolerances, max conformity under constraints) are solved numerically with a matrix-free solver. These steps are independent: the metrics are explicitly defined from the physical forward model rather than from the optimization output, the transformation is a known equivalence, and numerical trade-offs are direct consequences of solving the stated program. No self-definitional loops, fitted parameters renamed as predictions, or load-bearing self-citations appear in the derivation. The framework is self-contained against external benchmarks of optimization theory and TVAM forward modeling.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that normalized conformity and spillage metrics are faithful proxies for physical exposure and that the Charnes-Cooper transformation preserves the original optimization intent. No free parameters or invented physical entities are introduced in the abstract.

axioms (1)
  • standard math Linear-fractional programs can be converted to equivalent linear programs via the Charnes-Cooper transformation
    Invoked to make the projection design problem computationally tractable.

pith-pipeline@v0.9.0 · 5428 in / 1229 out tokens · 51170 ms · 2026-05-10T17:24:11.251980+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [1]

    Framework overview The overall framework of SiPO is illustrated in Fig

    Methods 2.1. Framework overview The overall framework of SiPO is illustrated in Fig. 1. The procedure consists of four sequential stages: spatial configuration, response prescription, shape optimization, and physical scaling. First, the spatial domain is partitioned based on the target geometry to explicitly handle the intended structure and artifact -pro...

    work page

  2. [2]

    UCBerkeley,

    Results and Discussion To systematically evaluate the proposed scale- invariant linear programming (LP) framework, we consider a series of experiments spanning binary and grayscale targets, different band-region configurations, and both 2D and 3D forward models. The goal of this section is twofold: (i) to characterize the behavior of the proposed LP formu...

    work page

  3. [3]

    This one -sided pattern is consistent with the hard inhibition structure of Case 2: because the normalized solution is bounded below by the prescribed Fig

    The realized ratios in the gelation region were 0.998 ≤ 𝐦𝐦∗ 𝐦𝐦𝑇𝑇 ≤ 1.131 and 0.999 ≤ 𝐟𝐟∗ 𝐟𝐟𝑇𝑇 ≤ 1.053, indicating that the solution remained tightly anchored to the lower bound while allowing a moderate upper -side overshoot. This one -sided pattern is consistent with the hard inhibition structure of Case 2: because the normalized solution is bounded belo...

    work page

  4. [4]

    to improve target conformity while preserving essentially the same separation level. In this sense, the first stage determines what process condition is safely achievable, whereas the second stage determines how well the target can be reproduced under that fixed condition. Such a sequential strategy is attractive because it is directly motivated by the ob...

    work page

  5. [5]

    Therefore, rather than relying on computationally expensive parameter sweeps, the proposed workflow suggests a practical route for process-aware LP design in which feasible process bounds are first identified and then exploited to recover fidelity in a controlled second stage. 3.3. Grayscale target with spatially varying prescription Fig. 7 presents the t...

    work page

  6. [6]

    Conclusion This work presents a SiPO framework for TV AM. By formulating the problem in a normalized space, the proposed approach enables explicit control of target fidelity and out -of-bounds exposure for both binary and grayscale targets. The general formulation achieves near-ideal target conformity, while the target-bounded (Case 1) and dose-shaping (C...

    work page

  7. [7]

    Volumetric additive manufacturing via tomographic recon- struction

    B. E. Kelly, I. Bhattacharya, H. Heidari, M. Shusteff, C. M. Spadaccini, and H. K. Taylor, “ V olumetric additive manufacturing via tomographic reconstruction, ” Science, vol. 363, no. 6431, pp. 1075–1079, Mar. 2019, doi: 10.1126/science.aau7114

    work page doi:10.1126/science.aau7114 2019

  8. [8]

    High-resolution tomographic volumetric additive manufacturing,

    D. Loterie, P. Delrot, and C. Moser, “ High-resolution tomographic volumetric additive manufacturing, ” Nat. Commun., vol. 11, no. 1, p. 852, Feb. 2020, doi: 10.1038/s41467-020-14630-4

    work page doi:10.1038/s41467-020-14630-4 2020

  9. [9]

    Volumetric Bioprinting of Complex Living-Tissue Constructs within Seconds

    P. N. Bernal et al., “V olumetric Bioprinting of Complex Living‐Tissue Constructs within Seconds,” Adv. Mater., vol. 31, no. 42, p. 1904209, Oct. 2019, doi: 10.1002/adma.201904209

    work page doi:10.1002/adma.201904209 2019

  10. [10]

    Use of volumetric additive manufacturing as an in-space manufacturing technology,

    T. Waddell et al., “Use of volumetric additive manufacturing as an in-space manufacturing technology,” Acta Astronaut., vol. 211, pp. 474–482, Oct. 2023, doi: 10.1016/j.actaastro.2023.06.048

    work page doi:10.1016/j.actaastro.2023.06.048 2023

  11. [11]

    A Thermally Expandable Res in for the V olumetric Additive Manufacturing of Shape ‐Memory Functional Foams,

    S. Tisato, G. Vera, Z. Hosneolfat, and D. Helmer, “ A Thermally Expandable Res in for the V olumetric Additive Manufacturing of Shape ‐Memory Functional Foams, ” Adv. Mater. Technol., vol. 11, no. 1, p. e01355, Jan. 2026, doi: 10.1002/admt.202501355

    work page doi:10.1002/admt.202501355 2026

  12. [12]

    Tomographic color Schlieren refractive inde x mapping for computed axial lithography,

    C. Chung Li, J. Toombs, and H. Taylor, “ Tomographic color Schlieren refractive inde x mapping for computed axial lithography,” in Symposium on Computational Fabrication , Virtual Event USA: ACM, Nov. 2020, pp. 1–7. doi: 10.1145/3424630.3425421

    work page doi:10.1145/3424630.3425421 2020

  13. [13]

    Computational optimization and the role of optical metrology in tomographic additive manufacturing,

    C. C. Li et al., “Computational optimization and the role of optical metrology in tomographic additive manufacturing,” in Advanced Fabrication Technologies for Micro/Nano Optics and Photonics XV , G. V on Freymann, E. Blasco, and D. Chanda, Eds., San Francisco, United States: SPIE, Mar. 2022, p. 31. doi: 10.1117/12.2610558

    work page doi:10.1117/12.2610558 2022

  14. [14]

    Latent image volumetric additive manufacturing,

    C. M. Rackson et al. , “Latent image volumetric additive manufacturing, ” Opt. Lett., vol. 47, no. 5, p. 1279, Mar. 2022, doi: 10.1364/OL.449220

    work page doi:10.1364/ol.449220 2022

  15. [15]

    Numerical modeling of tomographic volumetric additive manufacturing based on energy threshold method,

    R. Salajeghe et al. , “Numerical modeling of tomographic volumetric additive manufacturing based on energy threshold method,” Addit. Manuf., vol. 96, p. 104552, Sep. 2024, doi: 10.1016/j.addma.2024.104552

    work page doi:10.1016/j.addma.2024.104552 2024

  16. [16]

    High fidelity volumetric additive manufacturing,

    I. Bhattacharya, J. Toombs, and H. Taylor, “High fidelity volumetric additive manufacturing,” Addit. Manuf., vol. 47, p. 102299, Nov. 2021, doi: 10.1016/j.addma.2021.102299

    work page doi:10.1016/j.addma.2021.102299 2021

  17. [17]

    Object-space optimization of tomographic reconstructions for additive manufacturing

    C. M. Rackson et al. , “Object-space optimization of tomographic reconstructions for additive manufacturing, ” Addit. Manuf., vol. 48, p. 102367, Dec. 2021, doi: 10.1016/j.addma.2021.102367

    work page doi:10.1016/j.addma.2021.102367 2021

  18. [18]

    Tomographic projection optimization for volumetric additive manufacturing with general band constraint Lp- norm minimization,

    C. C. Li, J. Toombs, H. K. Taylor, and T. J. Wallin, “ Tomographic projection optimization for volumetric additive manufacturing with general band constraint Lp- norm minimization,” Addit. Manuf., vol. 94, p. 104447, Aug. 2024, doi: 10.1016/j.addma.2024.104447

    work page doi:10.1016/j.addma.2024.104447 2024

  19. [19]

    Pellizzon, R

    N. Pellizzon, R. Huber, J. Spangenberg, and J. S . Jørgensen, “ Systematic Analysis of Penalty -Optimised Illumination Design for Tomographic V olumetric Additive Manufacturing via the Extendable Framework TV AM AID Using the Core Imaging Library,” Feb. 12, 2026, arXiv: arXiv:2602.12178. doi: 10.48550/arXiv.2602.12178

    work page doi:10.48550/arxiv.2602.12178 2026

  20. [20]

    Deconvolution vol- umetric additive manufacturing

    A. Orth et al., “Deconvolution volumetric additive manufacturing,” Nat. Commun., vol. 14, no. 1, p. 4412, Jul. 2023, doi: 10.1038/s41467-023-39886-4

    work page doi:10.1038/s41467-023-39886-4 2023

  21. [21]

    Inverse Rendering for Tomographic Volumetric Additive Manufacturing

    B. Nicolet, F. Wechsler, J. Madrid -Wolff, C. Moser, and W. Jakob, “ Inverse Rendering for Tomographic V olumetric Additive Manufacturing,” ACM Trans. Graph., vol. 43, no. 6, pp. 1–17, Dec. 2024, doi: 10.1145/3687924

    work page doi:10.1145/3687924 2024

  22. [22]

    A novel linear programming approach to fluence map optimization for intensity modulated radiation therapy treatment planning,

    H. E. Romeijn, R. K. Ahuja, J. F. Dempsey, A. Kumar, and J. G. Li, “A novel linear programming approach to fluence map optimization for intensity modulated radiation therapy treatment planning, ” Phys. Med. Biol., vol. 48, no. 21, pp. 3521– 3542, Nov. 2003, doi: 10.1088/0031-9155/48/21/005

    work page doi:10.1088/0031-9155/48/21/005 2003

  23. [23]

    A first-order primal-dual algorithm for convex prob- lems with applications to imaging.J

    A. Chambolle and T. Pock, “A First-Order Primal-Dual Algorithm for Convex Problems with Applications to Imaging,” J. Math. Imaging Vis., vol. 40, no. 1, pp. 120–145, May 2011, doi: 10.1007/s10851-010-0251-1. 20

    work page doi:10.1007/s10851-010-0251-1 2011

  24. [24]

    A Flexible Growth Function for Empirical Use,

    F. J. Richards, “ A Flexible Growth Function for Empirical Use, ” J. Exp. Bot. , vol. 10, no. 2, pp. 290–301, 1959, doi: 10.1093/jxb/10.2.290

    work page doi:10.1093/jxb/10.2.290 1959

  25. [25]

    On Nonlinear Fractional Programming,

    W. Dinkelbach, “On Nonlinear Fractional Programming, ” Manag. Sci. , vol. 13, no. 7, pp. 492–498, Mar. 1967, doi: 10.1287/mnsc.13.7.492

    work page doi:10.1287/mnsc.13.7.492 1967

  26. [26]

    Programming with linear fractional functionals,

    A. Charnes and A. A. Cooper, “Programming with linear fractional functionals,” Nav. Res. Logist. Q., vol. 9, pp. 181–186, Sep. 1962, doi: DOI: 10.1002/nav.3800090303. 21

    work page doi:10.1002/nav.3800090303 1962

  27. [27]

    Karush-Kuhn-Tucker (KKT) Conditions Since the generalized scale -invariant formulation from Eq

    Supplementary 6.1. Karush-Kuhn-Tucker (KKT) Conditions Since the generalized scale -invariant formulation from Eq. ( 19) to Eq. ( 24) is a linear program, it is a convex optimization problem. Therefore, unde r standard feasibility assumptions, the KKT conditions are both necessary and sufficient for identifying the global optimum. Deriving these condition...

    work page