Chines in BSpline Surfaces/Images

[Britton Ward 0]

In many of our surface descriptions we model a chine by tripling a knot vector [for a 3 Degree Surface].  For example the V knot vector for a surface here:

 

0.000000,0.000000,0.000000,0.000000,  0.166667,  0.333333,       0.5, 0.666667,  0.666667,  0.666667,  0.833333,         1,         1,         1,         1

 

Importing this surface as an IGES works as expected.

 

However, if you create an image of the surface the topology is lost and the surface appears to be refit using the number U and V resolutions you have specified.  The result is a loss of the hard corner at the chine, replaced by an approximation and unfairness.

 

As a work around I realize I can subsurface the imported surface into two surfaces and make them a surface collection.

I wonder if there would be a way in the future for these fitting transforms to detect a chine discontinuity in the knot vector and preserve this in the fitted or transformed surfaces.

 

I'm sure there may be downstream issues but thought I would point it out.

 

I can supply an example if that would make it more clear.

 

 

Britt

[Stefan Wunderlich 6]

Hi Britt,

 

Could you attach a simple sampleproject where the issue is present?

 

Cheers,

Stefan

[Mr. Claus Abt 14]

Hi Britt,

 

I believe this can only be covered by something like feature recognition functions, I am afraid and not really generalized. Detecting gradually changing and even disappearing discontinuities within one surface is really a challenge. Maybe we can develop some ideas jointly - we are certainly open to new ideas.

 

So far we used the subsurface approach that you describe above and had quite good control.

 

Best regards

Claus

[Britton Ward 0]

Claus/Stefan,

 

Back onto this quickly.  I think I discovered my problem.  I was incorrectly populating the knot vectors when I read my file in so that they were defaulting to an even distribution and losing the multiplicity. I have reviewed the code and fixed my read statements and all seems consistent now.

 

Britt

[Stefan Wunderlich 6]

Hi Britt,

 

good to hear that it works now.

 

Cheers,

Stefan