Grid correction at boundary

Hi guys,

There is a problem with the grid size/shape at the boundary of the grid shell. as it is impossible to construct that type of grid at the boundary.

I cannot make a grid, like the central part of shell (blue one see attached) at the boundary beam (square). and I’m so stuck in it.
After this, I have to do an analysis on karamba 3d too.

Will wait for your help. Do anyone guide me regarding this. (18.5 KB)
grid shell2


The issue is that your base mesh has kind of a messy situation going on at its outer edges.

Instead of using Mesh (Mesh Brep) - which is not the best way to get a mesh if you rely on a clean topology -, I like to construct a clean mesh manually!
This way you retain much more control over the its topology and thus final outcome.

Since you want a filleted situation where the mesh is anchored, I’d do that part later in Kangaroo with the help of targeted anchors, and start out with a chamfered polygon instead. Meshes are edgy things and you need huge amounts of subdivisions to reproduce round geometries, however while starting out you practically always want to have a simple, low poly mesh that is still easily modifiable.

From the chamfered polygonal polyline, you can construct individual quad meshes.

These can be joined and welded to form a bigger mesh, that can in turn be subdivided further.

After carefully defining the Kangaroo goal objects, it should do the rest.
If you do dial up the subdivisions of the mesh, you may also need to adjust the Kangaroo goal values!

You could probably also have remeshed your base mesh with the QRemesh (Quad Remesh) component, but again it’s often times better to have more control.
After all, the initial quad layout hugely influences the final outcome! Not only does it dictate the overall mesh pattern, but also how the mesh is relaxed by Kangaroo.

I’ve laid the mesh out like this, but it surely is not the only way to do it! There are other ways (or patterns) to design the initial quads that in turn produce varied end results of the vaulted structure. Check this out!

Apart from quads, you could also explore tris (triangles) or even ngons (hexagonal or other). (28.0 KB)


hi @diff-arch,

It is quite clear and helpful.
New Microsoft PowerPoint Presentation.pdf (590.6 KB)
sorted out the grid on the outer edges and your script is very clear to me.

But apart from this, I’m wondering when I make the physical model the central part will also the square grid as yours on boundaries. (see attached)

because this issue will be creating difficulty on analysis on Karamba3d too.

You need to give EdgeLengths a little more leeway to prevent this, or define for instance a guide curve for each of the arches that the mesh edge should adhere to as additional goals.

1 Like

thankyou so much. it is quite helpful.

And your comment along with the script made it clearer to me.

And yeah creating mesh manually is far better than meshbrep.

1 Like

Yes, that’s the most important take away from this, not that MeshBrep is a bad component, but that if the base mesh is important topology-wise for the overall outcome, it should probably designed with that in mind!
I’d experiment with the different mesh types shown at the Blender site that I posted above and beyond, if you haven’t already. You can get vastly different results with various mesh topologies, and this is a nice opportunity to give it some individuality and take ownership of your design. In general, I think that we shouldn’t leave everything to the computer and/or algorithms.