Curvature of SubD contours

SubD, for me, so far, has been producing perfect or near-perfect zebras.

When I learned recently that most of the Curves from Objects toolbar worked with SubD objects, I decided to start playing around with various contours and checking them with _CurvatureGraph.

Here are some results:

For duplicated edges (magenta in this document), very good, simple curvature:

The same is true on more complex edges:

But when I obtain a contour with _Intersect using a plane set to bisect faces directly between edges (cyan curves), things get weirder:

(oddly the smoothest curvature here is on Penta faces)

Results are identical after converting to NURBS polysurface.

Is this an expected result of how SubD math works? Or is this happening because SubD is still a work in progress?

1 Like

@dalelear I think I need to tap you in here-

1 Like

When you use Zebra on a subd, the zebra analysis is calculated from the true SubD surface. No approximations, no lies.

When you extract NURBS surfaces and curves from a subd, there are some areas where the NURBS are a close approximation to the subd surface. We already have several bugs logged for cases where that NURBS approximation has poor tangent and curvature properties.

If you can post a 3dm containing the subd, I can look at the subd and give you more precise information about what you’re seeing, why you’re seeing it, and it’s an approximation, exact fit, or a bug.

1 Like

Hope Max doesn’t mind me piggybacking on his thread. I wanted to play with a simple case and noticed G0 continuity in one direction (all the way round), and uploaded in case it’s maybe not expected and is an example that may help. Or maybe because it’s an extraordinary point, it’s just to be expected.

G1_and_G0.3dm (360.3 KB)

It’s nobody’s thread. It’s about a question, not me. The more info the better. I’ll upload when I get a free minute.