Graded element sizing for Delaunay refinement (MESH_MINSIZE_ENFORCEMENT_V2_AND_GRADING_PLAN_2026-09-01.md Track B). More...
#include "mesh/meshgenerator.h"#include <QRectF>#include "meshkit/sizefield.hpp"#include <QVector>Go to the source code of this file.
Classes | |
| struct | mesh::SizeFieldOptions |
| Policy for SizeField::build(). More... | |
| struct | mesh::SizeFieldOptions::Region |
Polygon overrides: inside ring the size is min'd with h. More... | |
| class | mesh::SizeField |
| Distance-to-feature field sampled as a max-area function. More... | |
Namespaces | |
| namespace | mesh |
The Qt-facing MeshKit API (meshkit::core_qt, meshkit::model_qt, meshkit::io_qt): mesh value types and cell attributes, the mesh generator, quad regions, matching and cleanup, mesh statistics, and the INP and 2DM mesh codecs. | |
Graded element sizing for Delaunay refinement (MESH_MINSIZE_ENFORCEMENT_V2_AND_GRADING_PLAN_2026-09-01.md Track B).
Graded element sizing for Delaunay refinement (MESH_MINSIZE_ENFORCEMENT_V2_AND_GRADING_PLAN_2026-09-01.md Track B).
WHY THIS EXISTS. With a uniform maximum-area cap the WHOLE domain is refined to the cap, however far a cell sits from anything that needs resolution — on a large domain most of the output vertices buy nothing. This field grades the target size with distance from constrained features. The ideal continuous model uses exact Euclidean distance:
h(x) = nearSize + gradation · d(x) A(x) = (√3/4) · h(x)² (area of the equilateral triangle)
where d(x) is the distance to the nearest constrained feature (constraint segments, hole-ring edges, tagged Steiner points). In that ideal model, h approaches nearSize at features and has Lipschitz slope gradation. The sampled approximation below is not a proof of that bound for the produced mesh. Near-feature resolution and final cell counts require verification; grading is intended to allow coarser cells away from features.
The outer domain ring is deliberately NOT a seed: it is usually a watershed clip, not a hydraulic feature, and seeding it would pin fine cells along the whole perimeter. The mesher still refines near the boundary wherever the boundary's own local feature size demands it.
Mechanics: a uniform background grid over the domain bbox holds the distance to the nearest seed at each cell centre — exact distances are stamped in a small neighbourhood around every seed, then a two-pass chamfer transform propagates them using axial weight pitch and diagonal weight sqrt(2) * pitch. The ideal point-seeded octile metric can overestimate Euclidean distance by about 8.24%; this is not a bound for segment stamping, bilinear interpolation, or the complete sizing pipeline. Overestimating d increases h and its permitted area, producing a COARSER target, not a conservative refinement guarantee. The terrain-error term and the area floor further affect the final target. Computation is serial and deterministic for fixed input.