Strange problem with TriRemesh

Hi @DanielPiker

first thanks for your work on the remeshing component. It is a really great addition to the meshing tools!

I am using it to remesh LOD2.1 mesh data from .dxf mesh geometry provided to the public by official geodata services.

I need to adapt/increase the mesh density for simulation purposes.

While the component runs reliably for single mesh input, I do run into errors when I load many meshes at once. The input meshes are pretty clean, no non manifold edges, no bad objects, meshRepair does not complain about anything etc. I disjoint all meshes, but it made no difference.

The output omits certain input meshes, but when I feed these separately they do remesh just fine.
I have not yet found a single mesh input that makes the component fail.

very odd… I tried grafting input, no help though. I also tried R8 WIP, problem persists.

I am thinking about running an iterative loop via anemone, to process mesh by mesh.
But it would be great if you could have a look what could be causing this error.

cheers,
Daniel
triRemesh_problem.gh (29.3 KB)
LOD2.1__TriRemesh_problem.3dm (4.2 MB)


Ok… I built a iterative loop in order to crunch through the mesh data one by one.

the loop would stop at a problematic input mesh and it helped me identify the first culprit.

This mesh will break the component if I try to remesh with Length 5.
It works with length 2, but then creates some unusable output.
The input mesh seems valid, but I guess the (ugly) topology somehow creates a problem for the remesher.

maybe some help to make the algorithm more robust. I attached it and will upload more if I find additional showstoppers… :slight_smile:


LOD2.1__TriRemesh_problematicMesh.3dm (30.4 KB)

1 Like

I generally noticed the following issues:

inputting disjoint meshes is problematic. I had more success with splitting the disjoint meshes in rhino, and providing a clean object list to GH where I could trace back a single piece of mesh.

mesh topologies like in this image create the most trouble
Others are very thin faces (which most of them simply filtered out in GH via mesh area)

I attached a file with some more meshes that created problems.

For the sake of processing many objects (I had to deal with 10000s of objects) it would be very cool, if the component could keep processing and insert null objects or a dummy object in space (like a meshSphere or such at the location of the problematic mesh)

that it auto colours bad output meshes is pretty helpful btw.!!

LOD2.1__TriRemesh_problematicMesh_collection.3dm (58.6 KB)