Analyzing mesh for sharp discontinuities

I've been trying to develop a program that would compare each mesh face (using a quad mesh) with all of the faces adjacent to it and highlight or color code the mesh edge between them based on the angle between the faces' normals. I want to look at an entire mesh and see where the "discontinuity" of the mesh is greatest and least.

The purpose of this is to analyze a mesh which has been relaxed over various objects to help users determine the locations of seams and darts in order to minimize strain when flattening the mesh. I believe that adjacent mesh faces with higher angular differences, or sharp points in the mesh, will be the ideal locations for seams and darts.

The approach I've taken is to find the mesh faces adjacent to each face using wbAdjacentFaces. This creates a list/tree from which I can measure the difference between the face normals for each face and its neighboring faces. That's about as far as I got.

Any ideas?

Replies to This Discussion

Permalink Reply by Vicente Soler on March 8, 2014 at 5:22pm

Something like this?

All edges should be shared by two faces.

Attachments:

edgeangles.gh, 11 KB

Permalink Reply by Brendan on March 8, 2014 at 5:28pm

Vicente, EXCELLENT! It's similar to what I'm looking for. I'm actually looking to color map the edges of the faces, but something like this might work just as well. It's really just a tool to help the designer choose seam lines and dart locations so as to minimize strain when I flatten the mesh.

I'm also hoping to develop a mesh flattening algorithm, with the ability to break the mesh or add slits where I choose to minimize strain, but I'm wondering if this is something best left up to some of the software packages already available for Rhino.

Permalink Reply by Vicente Soler on March 8, 2014 at 5:49pm

If you only want to color the edges is more simple.

Attachments:

angleonlyedges.gh, 12 KB

Permalink Reply by Brendan on March 10, 2014 at 6:46am

I tried both of these this morning and get the error:

"Index was outside the bounds of the array. (Line 0)"

I updated to version .0072 but it still won't run correctly. Any ideas?

Permalink Reply by Vicente Soler on March 10, 2014 at 6:48am

I think there's a problem when internalizing the mesh. Try creating yourself a mesh and referencing it.

Permalink Reply by Brendan on March 10, 2014 at 7:37am

Nope, I had swapped out the reference but it still fails.

Permalink Reply by Vicente Soler on March 10, 2014 at 7:50am

It happens when a mesh doesn't have its face normals computed. You can either use the rhino command _rebuildmesh to fix the mesh or add the line:

m.FaceNormals.ComputeFaceNormals();

to the beginning of the script.

Permalink Reply by Brendan on March 10, 2014 at 8:28am

I haven't done any Grasshopper scripting. Did I put it in the right place?

Permalink Reply by Brendan on March 10, 2014 at 9:26am

I can't seem to make it run correctly by using either of these methods. I've attached the file I'm working on. It's an open mesh, I don't know if that matters or not.

Attachments:

OPEN MESH.3dm, 67 KB

Permalink Reply by Vicente Soler on March 10, 2014 at 10:03am

As I said on the first post, the script only worked if all edges share two faces. I was assuming you had a closed mesh. The following code will check if you have naked or non-manifold edges and assign them an angle of 0.

    List<double> angles = new List<double>();
    List<Line> edges = new List<Line>();

    m.FaceNormals.ComputeFaceNormals();

    for (int i = 0;i < m.TopologyEdges.Count;i++) {
      double angle = 0;
      int[] faces = m.TopologyEdges.GetConnectedFaces(i);
      if (faces.Length == 2)
        angle = Math.Acos(m.FaceNormals[faces[0]].X * m.FaceNormals[faces[1]].X + m.FaceNormals[faces[0]].Y * m.FaceNormals[faces[1]].Y + m.FaceNormals[faces[0]].Z * m.FaceNormals[faces[1]].Z);

      angles.Add(angle);
      edges.Add(m.TopologyEdges.EdgeLine(i));
    }
    A = angles;
    B = edges;