we didn’t get to imprioving the _SplitMeshWithCurve command in V7. However, I am not aware of changes in the 5 to 7 timeframe.
We are improving _MeshSplit itself, which in version 7 SR2 is capable of splitting that mesh, if the curve is not projected on the mesh first. The problem when it’s projected first, is that there are some minimal changes in the way the projector and the intersector define the two respective operations, and that creates tiny fractures along the split. These are similar places where the old MeshSplitWithCurves was failing. However, if you don’t project first, V7 SR2 MeshSplit splits the mesh easily. I am attaching an example. I regenerated the curve by simply rebuiling (_Rebuild) it in degree 3.
Thanks for your answer! I did the upgrade to the SR2 version. Then tried what you say here: rebuild the curve. The result is much better (the mesh is split!) but some issues remain. Here is a similar test case:
I will answer myself here: After having forgotten about the 3rd dimension, I tried removing the extension. I did the opposite: shorten the cutting curve removing parts that were not directly over the mesh.
The result is very good this time. If I get some others issues about that function, I’ll post them here.
So, as you probably realized, MeshSplit extrudes/uses parallel projection to split. SplitMeshWithCurve, on the other hand, pulls to the mesh.
Which methods would you need? MeshSplit or SplitMeshWithCurve?
I reported RH-62090 but there’s Mesh.CreateFromCurveExtrusion in V7 that allows to create a splitter mesh from a curve.
Does it mean that your question got answered? It’s fine and probably better if you start new threads if the new issue is tangential.
Thanks,
Giulio
–
Giulio Piacentino
for Robert McNeel & Associates giulio@mcneel.com
I run the test code above, modifying the extension length. After some tests, I realized the results depend heavily on the length used. Here are what I’ve got:
Extended 0.1 → fail
Extended 0.2 → fail
Extended 0.4 → fail
Extended 0.5 → good result
Extended 0.6 → Split, but small artefacts on sides
Extended 0.7 → fail
Extended 1.0 → Split, with artefacts
Extended 100 (as shown above) → Split w artefacts
So that where I am at the moment. I’ve got it to work in testing, but in production there will be nobody to review the result. I am afraid its not robust enough to use for our use case if we have to guess the correct extension length
I have not tried MeshSplit yet; that is my next step. I guess that design (I mean extrusion/parallel projection) should be more stable? Is that what you would recommend for that job?
Yes, I think that pulling is not a great idea for stability, because, as you noticed, it’s hard to determine the amount of extension that might be needed, or what to do in general toward the end of the curve.
Ideally, you’d find an independent plane or mesh to use for splitting.