{"id":"75c60dec-bf28-4e78-8329-524d3de1a492","arxiv_id":"1908.01938","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The paper generates complete, continuous heterogeneous porous scaffolds inside trivariate B-spline solids by defining a threshold field and a triply periodic minimal surface in the parametric domain, plus a compact TDF file format.","lead":"This paper presents a computer method for designing porous scaffolds, the lattice structures used in tissue engineering, by placing a repeated minimal surface inside a 3D B-spline model and mapping it through that model. It also introduces a compact file format for storing such scaffolds and reports large storage savings compared with explicit mesh files.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Porosity control is asserted from parametric-domain calibration curves but never measured on the mapped physical scaffolds; Jacobian distortion breaks the threshold-to-porosity transfer, leaving the central quantitative claim unvalidated.","rationale":"The reader's weakest assumption identifies exactly the same gap: the porosity calibration was established only in the uniform parametric grid, and the paper does not validate it on the final mapped scaffolds. This is the most load-bearing concern because the method's practical value for tissue engineering rests on quantitative porosity control, and the mapping through a general TBSS with varying Jacobian should change volume fractions. The constructive claims about completeness and continuity appear sound: with positive Jacobian, the TBSS mapping is a homeomorphism, so a closed, continuous scaffold in the parameter domain maps to a closed, continuous scaffold in the physical domain. The storage-format claim is plausible but also deserves a fairer comparison; still, the porosity issue is more central because it affects the main functional promise of the method. The missing evidence is a validation experiment, not an internal contradiction, so the appropriate verdict remains conditional pending the proposed porosity measurement. I therefore keep the reader's CONDITIONAL verdict unchanged.","tokens_in":12716,"tokens_out":4406,"duration_ms":53361,"concrete_test":"Reconstruct the Ball joint P-type pore example (Table 2) from the stored TDF and TBSS. Measure the actual physical porosity of the mapped scaffold, for example by Monte Carlo sampling of Eq. (4) mapped through Eq. (2), or by a watertight volume computation on the generated mesh. Independently compute the intended porosity by sampling the fitted TDF C(u,v,w) on a 50×50×50 grid, converting each sample to a porosity using the P-pore curve in Fig. 4, and averaging with weight |J(u,v,w)| over the parameter domain. If the measured and intended porosities differ by more than about 3 percentage points, the calibration transfer is invalid and the porosity-control claim must be revised. Repeating the same comparison on a TBSS with intentionally strong Jacobian variation would confirm whether the discrepancy is driven by the non-uniform mapping.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim is that porosity is controlled by the TDF (Eq. (4)). The only evidence is the uniform-grid calibration in Figs. 4-6, which gives the volume fraction of {ψ ≥ C} in the flat parametric cube. After mapping through the TBSS (Eq. (2)), however, the physical volume fraction of the corresponding set is weighted by the local Jacobian: V_phys = ∫_{ψ≥C} |J| dV / ∫ |J| dV. Unless |J| is constant, the calibration curves no longer predict physical porosity. The paper never reports measured porosity of the final mapped scaffolds in Figs. 13-17, nor does it compare them with any target derived from the TDF; Table 2 lists only runtimes and storage sizes. Additionally, the fitted B-spline TDF (Eqs. (5)-(10)) only approximates the discrete threshold values, introducing another unquantified source of porosity error. Thus the 'porosity controlled by TDF' claim is not established quantitatively, even though the completeness and continuity claims do follow from the positive-Jacobian construction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a method for generating heterogeneous porous scaffolds inside a trivariate B-spline solid (TBSS). The key idea is to construct a threshold distribution field (TDF) C(u,v,w) over the cubic parameter domain of the TBSS, define a TPMS as the level set ψ(u,v,w) = C(u,v,w), polygonize it with marching tetrahedra, and map the resulting surface and volume structures through the TBSS to obtain a porous scaffold. The authors claim three main contributions: (1) guaranteed completeness of TPMS units and continuity between them, (2) easy porosity control via the TDF, and (3) a compact 'TDF' file format that stores the scaffold as a procedural representation rather than a mesh. The paper includes several examples (ball joint, Venus, Moai, tooth, Isis) and a comparison with two prior methods, claiming storage savings of at least 98% over STL. The completeness and continuity claims are argued from the positive-Jacobian mapping of a unitary TPMS, and the storage format is presented with an appendix. However, the paper does not quantitatively verify that the porosity of the final mapped scaffolds matches the intended TDF-derived distribution.","tokens_in":12972,"tokens_out":5065,"duration_ms":104723,"significance":"If the claims are validated, the method would address three recognized drawbacks of TPMS-based scaffold design: broken boundary units, discontinuities between adjacent units, and large file sizes. The parametric-domain construction with a positive-Jacobian TBSS mapping is a conceptually clean way to guarantee completeness and continuity, and the procedural TDF file format is a practical contribution that could benefit downstream fabrication and interactive design. The local TDF modification workflow adds useful flexibility. The porosity-control claim, however, is the central quantitative assertion of the paper, and it is currently supported only by calibration curves computed on a uniform parametric domain, with no measured verification on the deformed physical scaffolds. The absence of such validation leaves the primary contribution only partially established, though the issue appears addressable with additional experiments or a Jacobian-aware calibration.","major_comments":[{"comment":"The porosity-threshold calibration curves in Figs. 4-6 are computed for a uniform cubic cell in the parametric domain. After mapping through the TBSS (Eq. (2)), the physical volume fraction of the set {ψ ≥ C} is ∫_{ψ≥C} |J| du dv dw / ∫ |J| du dv dw, where |J| is the Jacobian determinant. For a TBSS with non-constant Jacobian, the calibration curves do not directly predict physical porosity. The paper never reports the measured porosity of the final scaffolds in Figs. 13-17 or compares it with the intended TDF-derived distribution, so the claim that porosity is controlled by the TDF is not quantitatively established. Please provide physical-domain porosity measurements (e.g., by voxelizing the mapped scaffold and computing the volume fraction) for the presented examples, or alternatively restrict the method to TBSSs with constant Jacobian, or provide a Jacobian-corrected calibration procedure.","section":"§3.2, Figs. 4-6, Eq. (4)"},{"comment":"The TDF is obtained by least-squares progressive-iteration approximation (LSPIA) fitting of discrete threshold values on a 50×50×50 grid with a 20×20×20 control grid. The fitting error is not quantified. Since the TPMS is defined by ψ(u,v,w) - C(u,v,w) = 0, approximation errors in C directly shift the level surface and hence change the porosity. The convergence proof cited in [28] does not provide a concrete error bound for this particular fitting problem. The paper should report the fitting error (e.g., max and RMS error between the fitted B-spline and the discrete TDF) and, if possible, analyze its impact on the resulting porosity.","section":"§3.2, Eqs. (5)-(10)"},{"comment":"The proposed TDF file format as described stores the period coefficients, the control points and knot vectors of the TDF, and the control points and knot vectors of the TBSS, but it does not store the TPMS type (P, D, G, or I-WP) or the structure type (pore, rod, or sheet). These fields are essential for reconstructing the scaffold from the file; without them the same stored data could produce different scaffolds. The format appendix should be completed by adding these fields, and the text in §3.4 should list them as part of the stored information.","section":"§3.4 and Appendix"}],"minor_comments":[{"comment":"The computation method for the porosity-threshold curves in Figs. 4-6 is not stated (e.g., voxel grid resolution, or whether the volume fraction is computed from the marching-tetrahedra mesh or by numerical integration). Please state the method for reproducibility.","section":"§3.2"},{"comment":"The notation in Eq. (8) is confusing: the index set Iαβγ is defined by the condition Ni,p(uα)Nj,q(vβ)Nk,r(wγ) ≠ 0, but the summation uses I ∈ Iαβγ and evaluates the basis functions at (uI, vI, wI). The set should be defined as the set of grid nodes in the support of the (i,j,k)-th basis function, and the sum should run over those nodes. Please clarify the notation.","section":"§3.2, Eq. (8)"},{"comment":"The closing procedure for volume structures adds 'outside triangles' on the boundary faces when all vertices satisfy ψ ≥ C. This works if the iso-surface ψ = C does not pass through grid vertices; otherwise ambiguous cases may arise. A short comment on this assumption or a note on handling degenerate cases would be useful.","section":"§3.1"},{"comment":"The statement 'The larger the value of C(u,v,w), the larger the pore size' is valid for pore structures but is reversed for rod structures. Please qualify the statement to avoid ambiguity.","section":"§4.1"},{"comment":"Several typographical errors and formatting issues need correction, including 'compeleteness' in Section 3, 'ja:math' in reference [21], and inconsistent spacing in equations. A careful proofreading pass is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper presents a genuinely useful pipeline and the completeness/continuity claims are well-founded by the parametric construction. The main weakness is the unverified transfer of parametric-domain porosity calibration to the deformed physical domain; this is a correctable but essential gap. The TDF file format is promising but requires a small specification addition. The storage comparison against STL is favorable but somewhat expected for a procedural format; the authors should be encouraged to compare with other compact representations or at least state the comparison scope more carefully. The self-citation to LSPIA [28] is appropriate and not a concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the geometric construction is sound, and the completeness and continuity benefits are real and follow from the construction. But the paper's headline quantitative claim—that porosity is controlled by the TDF—is not actually measured in the final physical scaffolds, and the calibration curves as presented don't transfer to a deformed domain.\n\nWhat's new: defining the TPMS in the parametric domain of a trivariate B-spline solid, with a B-spline-fitted threshold distribution field, is a clean way to avoid broken TPMS units at boundaries and discontinuities between tiled units. The local modification workflow and the TDF file format are practical contributions. Storing control points and knot vectors instead of a giant STL mesh is a legitimate compact representation, and the regeneration times in Table 2 look acceptable.\n\nSoft spots, in order of size. First, the porosity control claim. Figures 4-6 are calibration curves computed on a uniform grid in the flat parametric cube. Once the TPMS is mapped through a non-affine TBSS, physical volume fractions are weighted by the local Jacobian. Unless |J| is constant or the calibration is redone in physical space, those curves don't predict the porosity of the mapped scaffold. The paper never reports the porosity of the final scaffolds in Figs. 13-17, so the central quantitative claim is not established. The B-spline fit of the TDF adds another unquantified error. Second, the storage comparison is apples-to-oranges: STL is a boundary mesh, while TDF is a procedural representation that must be re-meshed. That is a fair trade-off if stated, but Table 2 presents them as equal formats. Third, there is no code or data, which makes replication harder, though the pipeline is clearly described.\n\nThe paper is for researchers in CAD/geometric modeling and TPMS-based scaffold design. It deserves a serious referee, and I would send it out, but with a request for porosity validation on at least one mapped scaffold—using Jacobian-weighted volume fractions—and a more honest storage comparison.","headline":"The construction is sound and the completeness/continuity claims are real, but the porosity-control claim is not measured in the mapped scaffolds and needs validation before publication.","tokens_in":13475,"tokens_out":4158,"would_cite":true,"duration_ms":40989,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A trivariate B-spline solid can carry a heterogeneous porous scaffold by mapping a triply periodic minimal surface from the parametric domain, yielding complete, continuous pore units with porosity set by a threshold distribution field.","keywords":["heterogeneous porous scaffold","trivariate B-spline solid","triply periodic minimal surface","threshold distribution field","parametric domain","porosity control","TDF file format","completeness and continuity"],"falsifier":"Measure the actual porosity of a mapped scaffold, for example by voxelizing the final triangular mesh and counting the solid fraction, and compare it with the porosity predicted from the TDF calibration curves; if the mismatch exceeds a practical tolerance across a strongly deformed TBSS region, the porosity-control claim fails.","tokens_in":12547,"feed_emoji":"🦴","tokens_out":4510,"duration_ms":42682,"temperature":0.7,"pith_summary":"The paper develops a method for building heterogeneous porous scaffolds inside a trivariate B-spline solid (TBSS). Instead of embedding a TPMS directly in the physical shape or mapping one unit into each hexahedron of a mesh, it constructs a threshold distribution field (TDF) in the cubic parameter domain, generates a single connected TPMS there, and maps it through the TBSS function into the physical solid. Because the whole scaffold comes from one unitary parametric surface, the resulting pore units are complete at the boundary and continuously stitched between neighbours, and the TDF lets a user control pore size locally. If correct, this removes two known defects of prior TPMS scaffold methods and introduces a compact parametric file format that stores a scaffold in about one megabyte instead of hundreds.","feed_headline":"Porous scaffolds inside B-spline solids, complete and continuous","feed_subtitle":"A threshold field in the parametric domain controls pore size, and a compact file format cuts storage by 98 percent.","key_machinery":"The threshold distribution field (TDF), a trivariate B-spline function $C(u,v,w)$ fitted by least-squares progressive-iteration approximation (LSPIA) on a parametric grid, is the control mechanism: it converts the constant threshold in the TPMS equation into a spatially varying one, so the iso-surface $\\psi=C$ becomes a heterogeneous surface whose local pore size follows the field. The second element is the mapping through the TBSS function $P(u,v,w)$, which carries the parametric TPMS into the physical solid; a positive Jacobian of the TBSS ensures the map is fold-free, which is what preserves completeness and continuity in the final scaffold.","core_discovery":"The central claim is that a heterogeneous porous scaffold with complete, continuous TPMS units is produced by defining the TPMS in the parameter domain of a TBSS and mapping it to the physical domain. The TPMS is given implicitly by $f(u,v,w)=\\psi(u,v,w)-C(u,v,w)=0$, where $\\psi$ is a nodal approximation of a TPMS and $C$ is a trivariate B-spline TDF whose values set local pore size; the scaffold is the image of the volume bounded by this surface under the TBSS mapping. Completeness follows because the parametric TPMS is a single surface covering the whole domain, so no unit is cut off at the boundary, and continuity follows because adjacent 'units' are just portions of that same surface. The paper further claims the TDF file format—storing period coefficients, TDF control points and knots, and TBSS control points and knots—reproduces the scaffold at any resolution and cuts storage by roughly two orders of magnitude compared with STL.","pith_inferences":["If the porosity calibration curves are not re-verified in the deformed physical domain, the quantitative porosity-control claim may fail under strong TBSS distortion; a natural extension is to measure porosity of the mapped scaffold and adjust the TDF-to-threshold mapping accordingly.","The parametric storage suggests a design-optimization loop in which the TDF control points and period coefficients act as free parameters, letting an optimizer tune pore-size distribution without ever exporting a mesh.","The same 'generate in parameter space, map to solid' trick could work for other implicit structures, such as Voronoi-like level sets or lattice functions, not only TPMSs, whenever the solid is given as a spline volume."],"forward_implications":["Any TBSS with positive Jacobian can be turned into a heterogeneous scaffold whose TPMS units are complete and continuous, eliminating the two reported defects of hexahedral-mesh and T-spline embedding methods.","Pore size becomes a design variable: editing the TDF in the parameter domain, for example by local LSPIA updates, rebuilds the scaffold locally without redoing the whole generation.","The TDF file format stores the scaffold parametrically, so the same file can generate meshes at any prescribed precision, and storage drops from hundreds of megabytes to about one megabyte.","The same pipeline extends to pore, rod, and sheet structure types and to P, D, G, and I-WP TPMSs, provided the threshold stays in the valid range for that surface."],"supporting_citations":[{"why":"Baseline hexahedral-mesh TPMS mapping method that the paper argues leaves discontinuities between adjacent units.","marker":"[9]"},{"why":"Baseline solid T-spline and TPMS embedding method that the paper argues produces incomplete boundary units.","marker":"[8]"},{"why":"Supplies the nodal approximations of P, D, G, and I-WP TPMSs and their valid threshold ranges used to define the TPMS and TDF.","marker":"[4]"},{"why":"Provides the LSPIA fitting algorithm used to turn the discrete threshold field into the trivariate B-spline TDF.","marker":"[28]"},{"why":"Provides the marching tetrahedra algorithm used to polygonize the parametric TPMS iso-surface.","marker":"[23]"}],"fun_headline_variants":["Parametric TPMS yields complete porous scaffolds in B-spline solids","Continuous TPMS scaffolds from trivariate B-spline solids","Porous scaffold file format cuts storage by 98 percent","Complete porous scaffolds via TPMS threshold fields in B-spline solids","Heterogeneous porous scaffolds: TPMS defined in parametric space"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The porosity-versus-threshold curves measured on the uniform parametric grid are assumed to still give the correct porosity after the TPMS is deformed through the trivariate B-spline solid into the physical shape, but the paper never measures the porosity of the final mapped scaffolds to check.","fun_headline_variants_meta":{"raw":{"variants":["Parametric TPMS yields complete porous scaffolds in B-spline solids","Continuous TPMS scaffolds from trivariate B-spline solids","Porous scaffold file format cuts storage by 98 percent","Complete porous scaffolds via TPMS threshold fields in B-spline solids","Heterogeneous porous scaffolds: TPMS defined in parametric space"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000642,"raw_usage":{"total_tokens":2978,"prompt_tokens":995,"completion_tokens":1983,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":611,"completion_tokens_details":{"reasoning_tokens":1897}},"tokens_in":611,"tokens_out":1983,"duration_ms":13850,"temperature":1.0,"reasoning_tokens":1897,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:58:42.958487+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual porosity of a mapped scaffold, for example by voxelizing the final triangular mesh and counting the solid fraction, and compare it with the porosity predicted from the TDF calibration curves; if the mismatch exceeds a practical tolerance across a strongly deformed TBSS region, the porosity-control claim fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Baseline hexahedral-mesh TPMS mapping method that the paper argues leaves discontinuities between adjacent units."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Baseline solid T-spline and TPMS embedding method that the paper argues produces incomplete boundary units."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the nodal approximations of P, D, G, and I-WP TPMSs and their valid threshold ranges used to define the TPMS and TDF."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the LSPIA fitting algorithm used to turn the discrete threshold field into the trivariate B-spline TDF."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the marching tetrahedra algorithm used to polygonize the parametric TPMS iso-surface."}],"review_version":1}