Fillets, Kinks, and UV's?


I’ve noticed lots of filleting problems (naked edges, or fillets that show up, but won’t trim the prior geometry) and failures when a surface has either or both conditions:

A) a lot of UV’s (often resulting from FlowAlongSrf commands of a simple planar surface).
B) kinks in the surface

When I rebuild these high UV surfaces to lower UV’s, I tend to have more success filleting a resulting solid.

The questions I have are if Rhino rebuilds a surface with a lot of UV’s (in my case 26U, 55V from a previously planar surface):
A. Why are the UV’s so high? (is this a tolerance issue?)
B. Like doing the limbo, how low should one go, or how low is too low, when rebuilding UV’s?

As for kinks, well… is there any hope other than manual repair after the fact?

You can download an example of the problem rhino file here (2.2 meg)

Many thanks for any insights!


(Pascal Golay) #2

Hi Dave- FlowAlongSrf or other deformation tools will pretty much always add considerable complexity to surfaces, especially polysurfaces in the refitting process- this is unavoidable. But- why are you doing this? is it to build flat and then change the shape? I mean, the simpler object looks OK to me, why are you trying to get there with deformation?



Hi Pascal,

Thanks for confirming my suspicion about UV surface complexity after deformation.

I’m happy for some workflow advice, since there has to be a simpler method that I’m overlooking. I have 2D top views and front views of the geometry. I’m simply looking for the simplest way to get to the 3D geometry on the left, which I then wish to fillet the edges.

What I included, on the left, was a FlowAlongSrf method, using a curve from the top view to create a bent surface to which I applied the flat planar surface of the eyeglass from created from the front view. Then rebuilt UVs to a lower number, etc, to get what you’re seeing as the simpler object.

Many thanks!



While you’re pondering this, insights into why transformations like FlowAlongSrf tend to rebuild with high UV’s would also be welcomed.

Is there any way to reduce the resulting surface UV’s through things like lowering tolerance settings, or keeping control points for line geometry at a minimum?



Well they make highly dense surfaces because what they’re for are deformations that were previously impossible to do with solids at all, it has to make surfaces really dense to try to keep all the edges in relatively the same position and within tolerance while being bent. It’s not for situations where the simplest possible geometry is the goal.


Thanks, Jim. It would be nice to be able to lower the tolerances if simpler surfaces are desired. Is this possible?


I suppose but wouldn’t really advise it. I guess I had one case where I had a history chain of deformations on a surface and using FitSrf on the result removed some micro-fluctuations in it, but playing with tolerance settings is a nasty hack to be avoided. If you’re determined to try to use it, maybe look into not using it for ALL the surfaces of your object.


I hear ya’!

In an ideal world, transformations would give either a tolerance option, or a UV number option with deviation shown (like Rebuild). Result? Simpler shapes that remain watertight solids, but can still be edited more easily after a transformation.

Sometimes, while designing, “close enough” shapes are more desired than highly accurate (to 0.001 mm) shapes.