In those figures it looks like they’re using a similar approach to what we’ve discussed before here
I agree though, that if what you are after is that general look and shape for a large number of cells, then treating each cell as a chain of many capsules or spheres and doing collision might not be the most efficient way to go. Making a distribution of points for the cell centers and using some meshing or Voronoi type division could make more sense.
Let’s try and clarify what exactly it is you are looking for here though (and in your other recent post, which I presume is about the same project).
Can you define it?
In both these posts I’m seeing a 2d arrangement of non overlapping cells of roughly similar but non identical areas (or possibly with areas changing smoothly over some gradient), with rounded polygonal shapes so they fill the space with few gaps. There is also some anisotropy going on here (in parts the cells are elongated, following directions of some smooth field).
I suggested mapping (using Sporph or similar) in the other thread because that’s one easy way to turn an isotropic pattern into an anisotropic one.
One thing that complicates using a simple Voronoi approach is the distinction between the anisotropy of the distribution of points or cell centers and the anisotropy of the cells themselves.
Here’s the difference between starting from a uniform distribution and doing a non-uniform scaling of a then taking the Voronoi, vs taking the Voronoi then doing the non-uniform scaling of that:
