Advice on Mesh modelling – converting to MESH from NURBS

Hi all,

I’m looking for some advice on getting better Mesh results from the NURBS conversion. The context is that I am converting a model I’ve created into a mesh for use in Unity. I know that this probably isn’t the most efficient workflow but my model is already in Rhino NURBS and it’s been years since I’ve used a mesh modeller. Having to re-learn/ learn a mesh modeller is a bit of an interruption to my workflow just now.

I’m working on creating spokes for a wheel.

I initially modelled a relatively detailed spoke and after meshing it, realised it would probably be best to start with a low-poly NURBS rather than a detailed one, then convert to a mesh. The results on the rear wheel worked pretty well as you can see in the MESH on layer A.

So, I duplicated the original low-poly NURBS to use on the front wheel and here is where I’ve started to run into problems. After duplicating, I rotated the spoke, then selected the end of the spoke (nearest the wheel rim) and repositioned it. I then did a polar array to get the right side, Then mirrored those to get the left side. Lastly, I again selected the ends of the spokes to reposition them (as the spokes are asymmetrical)

When I then selected all the spokes and did a MESH from NURBS object I ended up with spokes with a rage of face numbers (31-555), despite all the spokes starting with the same number of faces (as they were all copies of one original). The original (layer D) has 26 surfaces, but when I MESH from NURBS, (as the lowest count, on the simple settings), that same spoke ends up with 186 faces (layer E).

Obviously, starting with a NURBS spoke (layer B) and running MESH from NURBS Object (layer C) is ridiculous for the resulting face count, especially since reducing the face count to an acceptable number results in exceptional distortion that can be negated by doing a low-poly model in the first place.

My questions (regarding the above) are:

Why, when doing a MESH from NURBS Object on a low poly object, does the face count increase (by quite a lot? I could’ve thought at if you had a low-poly object to begin with that the MESH from NURBS would be unable to generate more faces than are already present.

Why, when doing a MESH from NURBS Object on multiple selections simultaneously, do the faces counts differ so dramatically when the original objects have the same face count to start with?

*note, in double checking this, I’ve found that running MESH from NURBS Object on the same group of spokes seems to produce a difference face count each time. I’ve had a smaller deviation in resultant face counts. Curious as to why, if that’s known.

Also just discovered the following link I’ll check out and try to see if it help. An initial perusal is that it might not answer my question, but might help the work. And, does this just pertain to display or does this also apply to the MESH from NURBS Object?


I’ve been working on the tyre as well. Again I started with the existing NURBS object (a simple Revolve) on layer T1. I then did a MESH from NURBS Object on it and reduced the face count (layer T2). *except I’ve just noticed that the NURBS object I’ve included wasn’t the original, I’d flattened the inside of the tyre to simplify the mesh, but it’s only there as a visual aid.

I’m comparing the Rhino MESH conversion to a very primitive model that I believe was created in a MESH modeller (thought I don’t know which, why, or how). . I tried to get the polygon counts about the same for the tyres, (extracting a best guess face count for the spokes) and I think I’m close enough. I’m intrigued by the difference in the face patterns. In terms of modelling/ display the magic of 3D – is there any difference between the face pattern/ orientation between Rhinos calculations and a MESH modeller – other than “whichever looks right in Unity and has the lowest polygon count/ is most efficient?”

Thanks for any insight anyone might be able to provide.

Fingers crossed I can attach the project

Spoke Help.3dm (2.4 MB)

Hello- if you are shooting for a simple mesh, in this particular case you can cheat you might be able to get away with ‘jagged seams’

That will leave some nakeds in the mesh but it may not matter in the rendering. Or you can cheat by:

  1. Turn off isocurcves for your spoke
  2. ExtractWireframe
    3.MeshFromLines the result.

You can FillMeshHoles to close the ends if needed.


Brilliant, thanks Pascal! And thanks for the quick reply, too

Both techniques worked great but the 2nd one had the benefit of allowing me to modify one of the end shapes a little bit. Probably won’t be noticeable in the end but it makes me feel better. the 2nd one was actually easier in the end too as it gave me a minimal number of faces without the need for futer reducing, as the 1st technique split the longs sides into many quads.