Pith. sign in

REVIEW 4 major objections 5 minor 10 references

Poxel: Voxel Reconstruction for 3D Printing

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper proposes Poxel, a voxel-based reconstruction framework that outputs print-ready CMYKWCl voxel grids directly from photographs, avoiding the intermediate conversions that degrade color fidelity in physical 3D prints.

desk verdict A plausible pipeline idea that is under-specified and renders its central fidelity claim unverifiable. read the letter →

arxiv 2501.10474 v1 pith:LKBSAIM5 submitted 2025-01-16 cs.GR cs.CV

Ruixiang Cao , Satoshi Yagi , Satoshi Yamamori , Jun Morimoto This is my paper · ORCID
classification cs.GRcs.CV
keywords PoxelprintablevoxelCMYKWClcolormodelphotopolymerjetting3Dreconstructionradiancefieldsrenderingdiscretization
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

Poxel is a proposal for making 3D reconstructions that are ready to print, not just ready to view. Standard radiance-field methods encode scenes in view-dependent RGB colors optimized for a virtual camera, which cannot be sent directly to a multi-material printer. Poxel instead learns a grid of voxels, volume cells, where each voxel carries layered CMYKWCl ink amounts, removes view dependence, and uses anisotropic voxels so layer height and color blending match the printer's mechanics. The paper demonstrates end-to-end prints from phone and camera footage on a commercial jetting printer, reporting that printed colors stay closer to their digital counterparts than an adapted 3D-Gaussian-splatting pipeline. If the approach holds, it would let a single reconstruction feed a printer without a separate color-managed conversion stage.

What carries the argument

The central object is the Poxel, an anisotropic, printable voxel whose color is stored as layered CMYKWCl, cyan, magenta, yellow, black, white, clear, ink amounts rather than RGB. The framework learns this grid by volume rendering with view-independent colors, applies a regional color average, $\bar{C} = \frac{1}{N}\sum_{i=1}^{N} C_i$ over $N=12$ voxels in a $1 \times 6 \times 2$ block, to smooth transitions, and discretizes to the printer's ink palette during optimization. That combination is what lets the output be sliced and printed without a separate color-space conversion.

What would settle it

Print a standardized color target reconstructed by Poxel on a CMYKWCl jetting printer and measure the patches with a spectrophotometer; if the measured colors depart from the source colors by more than the printer's normal gamut error, the claimed fidelity advantage of direct CMYKWCl voxel output fails.

Watch

Extended reading notes

Core claim

The central claim is that a voxel representation designed around the printer's actual ink set, rather than around virtual rendering, removes the main source of print error: converting a view-dependent RGB model into printable ink layers after the fact. By optimizing voxel colors directly as CMYKWCl values and averaging colors over a fixed $1 \times 6 \times 2$ voxel region to smooth gradients, Poxel produces a grid that can be sliced into printer layers. In the paper's comparison, the Poxel print had smoother color transitions and better alignment with the printer's color gamut than a baseline that rendered with 3D Gaussian splatting, converted the splats to ellipsoidal volumes, and then mapped RGB to CMYKWCl, because Poxel never leaves the print color space.

Load-bearing premise

The method assumes that averaging RGBA colors over a fixed $1 \times 6 \times 2$ voxel block and then quantizing the result to CMYKWCl inks produces colors the printer reproduces faithfully; this was checked by visual comparison, not by measuring printed color patches against a colorimetric standard.

Editorial extensions

If this is right

  • Printed Poxel models should retain the smooth color gradients of the source scene because color averaging happens in the print color space rather than in RGB.
  • A single Poxel reconstruction can be sliced directly into layers for a multi-material jetting printer, eliminating the RGB-to-CMYKWCl conversion step that introduced artifacts in the comparison baseline.
  • Removing view dependence makes the representation invariant to camera angle, which is necessary for an object that is viewed in the round as a physical print.
  • Anisotropic voxel spacing lets the reconstruction allocate more resolution along the print height axis, where layer thickness limits achievable detail.

Reading between the lines

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

  • A natural next test, not reported in the paper, is a quantitative colorimetric evaluation between the digital reconstruction and the prints, such as CIEDE2000 color differences; the paper's fidelity improvement is currently supported by visual inspection.
  • The fixed $1 \times 6 \times 2$ averaging window is a tunable hyperparameter, and per-printer adjustment of its shape and size could further reduce color banding on devices with different native resolutions.
  • Because the ink palette is the only printer-specific input to the discretization, the same framework should transfer to other CMYKWCl jetting printers by swapping the palette lookup, which would broaden the evidence for the central claim.
  • If direct voxel output becomes accepted, printer drivers could consume reconstructed voxel grids directly, collapsing the separate slicing and color-management stages that current printing pipelines perform.
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 'Poxel,' a voxel-based reconstruction and color-conversion pipeline for full-color multi-material jetting 3D printing. The method replaces view-dependent RGB radiance fields with view-independent CMYKWCl voxels, uses anisotropic voxels and regional color averaging, and directly outputs printable voxel grids. The evaluation consists of two printed models on a Stratasys J850 and a qualitative comparison against a 3D Gaussian Splatting baseline. The core technical steps, particularly the RGB-to-CMYKWCl discretization and the optimization losses, are not specified, and the reported fidelity improvement is supported only by subjective statements.

Significance. If fully specified and validated, a direct pipeline from digital reconstruction to printer-native CMYKWCl voxel grids would be a useful contribution to color 3D printing, potentially avoiding conversion artifacts. The use of a real multi-material printer and physical test prints is a concrete strength. However, in its current form the contribution cannot be reproduced or quantitatively assessed, and the central claim of improved fidelity is not supported by the evidence presented.

major comments (4)
  1. [III-C] The RGB-to-CMYKWCl discretization is not specified. The text states that colors are 'discretized into CMYKWCl' but gives no algorithm, lookup table, or mathematical relationship between the averaged RGBA value from Eq. (1) and the printer's ink mixture. This mapping is the core of the claimed fidelity improvement, so its absence makes the pipeline unreproducible and the central result unfalsifiable.
  2. [IV / IV-B] The results are based on subjective observations of two printed models. There are no colorimetric measurements, no ΔE or other error metrics, no error bars, and no controlled baseline with a standard color-management workflow. The statement in Section IV-B that 'the Poxel model provided superior alignment with the printer's color gamut' is therefore an assertion, not a demonstrated finding. Quantitative evaluation against measured print patches is required before any fidelity claim can be accepted.
  3. [III-D] The optimization is under-specified. Equation (1) defines a regional averaging operation, but it is not a loss function, and the 'structural loss' mentioned in the same section is never defined. There is no description of the training procedure, hyperparameters, voxel grid resolution, or how the averaged colors are used in the optimization. Without these details, the method cannot be reproduced or compared.
  4. [III-B] The anisotropic voxel configuration is not described quantitatively. The paper claims that anisotropic voxels enable finer control over layer resolution and color blending, but it never states the actual voxel dimensions or how they map to the printer's 600x300 dpi planarity and 0.014 mm layer thickness. The only numeric parameter introduced later, the 1x6x2 averaging window in Section III-D, is presented without justification or sensitivity analysis.
minor comments (5)
  1. [III-A, III-B, II-A] The headings and text contain spacing artifacts ('V olume', 'V oxel', 'T raining'); these should be fixed for readability.
  2. [IV] The experimental protocol description is inconsistent: the text mentions 'two batches' of printed models, then 'first test' and 'second test', while Fig. 3 refers to a 'Third attempt'. Please clarify how many models were printed and which attempts correspond to which results.
  3. [III-D, Eq. (1)] Equation (1) averages RGBA values but does not state whether the values are in linear or gamma-encoded space; this distinction is important for color mixing and should be specified.
  4. [IV-A] The 3DGS baseline adaptation is described only qualitatively ('converting 2D splats to 3D ellipsoidal volumes, slicing along the height axis, and mapping RGB values to a CMYKWCl palette'). More detail is needed to interpret the comparison and to assess whether the baseline is a fair one.
  5. [References] References [4]-[6] are vendor web pages; archival citations or DOIs, if available, would be more durable and useful to readers.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper makes no prediction that reduces by construction to its inputs.

full rationale

The derivation chain in Poxel is not circular in any of the enumerated senses. The central method removes view dependence from NeRF/Plenoxel-style radiance fields and converts RGB voxel colors to a CMYKWCl representation for a Stratasys J850 printer. The only quantitative formula, Eq. (1), averages RGBA colors over a fixed 1x6x2 region; the result is used as the voxel color fed to the printer, not as a fitted parameter that later predicts the same averaged quantity. No parameter is fitted to a subset of the data and then used to predict that subset. The paper contains no self-citations, no imported uniqueness theorem, and no ansatz smuggled in via prior work. The fidelity claim ('the Poxel model provided superior alignment with the printer's color gamut') is an unsupported qualitative assertion based on two printed models, and the RGB-to-CMYKWCl discretization in Section III-C is not specified in enough detail to be reproduced or falsified. Those are correctness, completeness, and falsifiability flaws, not circularity: the claimed result is not equivalent to any input of the method by construction. Accordingly, the circularity score is 0.

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

The central claim depends on the choice of averaging window and voxel geometry, and on unstated assumptions about color reproduction. No free parameters are fit to data, but several are chosen by hand without validation.

free parameters (3)
  • Regional averaging window size = 1x6x2 (N=12)
    Introduced in Eq. (1) to smooth color transitions; chosen without a search or sensitivity analysis. Print fidelity depends on this window.
  • Anisotropic voxel dimensions = Not reported
    The paper states voxels are anisotropic but gives no layer thickness, lateral resolution, or aspect ratio. These dimensions directly affect print quality.
  • CMYKWCl discretization thresholds = Not reported
    The mapping from RGB to the six inks is described only verbally in Section III-C; thresholds or lookup tables are not provided.
assumptions (3)
  • domain assumption Averaging RGBA colors over a local region approximates the color perceived on a printed object.
    Used in Section III-D to justify the 1x6x2 averaging window. No perceptual or colorimetric model is given.
  • domain assumption The Stratasys J850 CMYKWCl ink system can reproduce the averaged colors from the voxel grid.
    Assumed throughout Section IV. No printer profile, gamut mapping, or calibration is described.
  • domain assumption A single view-independent color per voxel is sufficient for physical 3D printing.
    The paper removes view-dependency entirely (Section III-A) but does not discuss translucency, surface reflectance, or inter-reflection, which also affect printed appearance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Poxel: Voxel Reconstruction for 3D Printing." pith.science (2026). https://pith.science/paper/LKBSAIM5

@misc{pith2026250110474,
  author       = {Pith},
  title        = {Pith review of: Poxel: Voxel Reconstruction for 3D Printing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LKBSAIM5}},
  note         = {Machine review of arXiv:2501.10474}
}
read the original abstract

Recent advancements in 3D reconstruction, especially through neural rendering approaches like Neural Radiance Fields (NeRF) and Plenoxel, have led to high-quality 3D visualizations. However, these methods are optimized for digital environments and employ view-dependent color models (RGB) and 2D splatting techniques, which do not translate well to physical 3D printing. This paper introduces "Poxel", which stands for Printable-Voxel, a voxel-based 3D reconstruction framework optimized for photopolymer jetting 3D printing, which allows for high-resolution, full-color 3D models using a CMYKWCl color model. Our framework directly outputs printable voxel grids by removing view-dependency and converting the digital RGB color space to a physical CMYKWCl color space suitable for multi-material jetting. The proposed system achieves better fidelity and quality in printed models, aligning with the requirements of physical 3D objects.

Figures

Figures reproduced from arXiv: 2501.10474 by the authors.

Figure 1
Figure 1. Overview of the Poxel Framework for 3D Printing. (a) Anisotropic [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The real-world experiment of Poxel. The model is printed by a [PITH_FULL_IMAGE:figures/full_fig_p001_2.png] view at source ↗
Figure 3
Figure 3. End-to-end reconstruction result of Poxel before discretization [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: End-to-end reconstruction result of Poxel (a) Input figure taken from an DJI Osmo Pocket 3 camera[7]. (b) Reconstructed voxel grid after color [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 6 canonical work pages

  1. [1]

    4" FUNCTION default.is.dash.repeated.names #1 FUNCTION default.name.format.string

    11em plus .33em minus .07em @technote 4000 4000 100 4000 4000 500 `\.=1000 = #1 #1 #1 0pt [0pt][0pt] #1 * \| ** #1 \@IEEEauthorblockNstyle \@IEEEauthorblockAstyle \@IEEEauthordefaulttextstyle \@IEEEauthorblockconfadjspace -0.25em \@IEEEauthorblockNtopspace 0.0ex \@IEEEauthorblockAtopspace 0.0ex \@IEEEauthorblockNinterlinespace 2.6ex \@IEEEauthorblockAinte...

    work page 2003

  2. [2]

    write newline

    " write newline "" initialize.prev.this.status FUNCTION begin.bib preamble empty 'skip preamble write newline if " thebibliography " longest.label * " " * write newline " [1] #1 " write newline " url@rmstyle " write newline " " write newline " [2] #2 " write newline " =0pt " write newline " " ALTinterwordstretchfactor * " " * write newline " = 2 plus " wr...

  3. [3]

    Mildenhall, P

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, ``Nerf: Representing scenes as neural radiance fields for view synthesis,'' 2020. [Online]. Available: https://arxiv.org/abs/2003.08934

    arXiv 2020

  4. [4]

    A. Yu, S. Fridovich-Keil, M. Tancik, Q. Chen, B. Recht, and A. Kanazawa, ``Plenoxels: Radiance fields without neural networks,'' 2021. [Online]. Available: https://arxiv.org/abs/2112.05131

    arXiv 2021

  5. [5]

    Kerbl, G

    B. Kerbl, G. Kopanas, T. Leimkühler, and G. Drettakis, ``3d gaussian splatting for real-time radiance field rendering,'' 2023. [Online]. Available: https://arxiv.org/abs/2308.04079

    arXiv 2023

  6. [6]

    , ``Polyjet technology overview,'' https://www.stratasys.com/en/guide-to-3d-printing/technologies-and-materials/polyjet-technology/, 2024, accessed: November 12, 2024

    Stratasys, Ltd. , ``Polyjet technology overview,'' https://www.stratasys.com/en/guide-to-3d-printing/technologies-and-materials/polyjet-technology/, 2024, accessed: November 12, 2024

    work page 2024

  7. [7]

    Sailner 3D Printers , ``Sailner 3d printing solutions,'' https://www.sailner.com/, 2024, accessed: November 12, 2024

    work page 2024

  8. [8]

    , ``3d printing product line,'' https://mimaki.com/product/3d/, 2024, accessed: November 12, 2024

    Mimaki Engineering Co., Ltd. , ``3d printing product line,'' https://mimaki.com/product/3d/, 2024, accessed: November 12, 2024

    work page 2024

Show all 10 references
  1. [9]

    DJI , ``Osmo pocket 3,'' https://www.dji.com/jp/osmo-pocket-3, 2024, accessed: November 12, 2024

    2024

  2. [10]

    [Online]

    NVIDIA Corporation , ``Nvidia geforce rtx 4090 specifications,'' 2024, 24 GB GDDR6X memory, 16384 CUDA cores, 2.52 GHz boost clock, 450 W TDP, 4th Gen Tensor Cores, and 3rd Gen RT Cores. [Online]. Available: https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.