Pattern from Features
From Features
could generate quad mesh generation with given singularities.
Boundary Point Feature
If a four-sided patch has a corner labeled as ’feature point’, then the diagonal corresponding to this corner is added, to split the four-sided patch into two three-sided patches
Inner Point Featrue
Inner point features modify the Delaunay triangulation, and the resulting set of patches, mainly by adding three-sided patches.
Curve features
The curve features are discretised and the resulting points added to the set of points used for the Delaunay triangulation, as for the boundaries.
Ex 1: British Museum
Here are four different ways to build the British Museum using different features.
1. Pattern
1a. Create pattern
1b. Modify pattern
Once the initial quad mesh has been generated, the pattern can be relaxed. Relaxation of the Pattern is not a requirement for successfully finding horizontal and vertical equilibrium. However, relaxing the Pattern during early stages of the RV2 workflow will ensure that the pattern is smooth, which will also make the eventual ThrustDiagram has smooth force
2 + 3 + 4. Define boundary conditions, create form and force diagrams
We will use all of the boundary vertices as supports, but not the singularities (the point features on the surface that were fixed in the previous step).
5. Horizontal equilibrium
6. Vertical equilibrium
bm_vertical_equilibrium.zipbm_vertical_equilibrium.zip - 612KB
Exercise 2: Armadillo Vault from Features
Method 2: Relax the Pattern
After relaxation, unfix all the vertices of the pattern by following the instructions in the previous step Unfixing of the vertices can be done (or not even necessary, if they are intended to be used as anchors) during any point before the creation of the FormDiagram
. However, for clarity of the procedures of the main workflow steps, all vertices are explicitly unfixed in this step.
Comparing pattern refinements
Left - Pattern from feature with target length 1, without smoothing or relaxation
Middle - Pattern from feature with target length 1, and 30 iterations of constrained Laplacian smoothing (notice the more evenly distributed vertices along the boundaries)
Right - Pattern From Feature with target length 1, 30 iterations of constrained Laplacian smoothing and relaxation with all boundary vertices fixed (notice the much smoother continuous edges).
Last updated