# [WISH] SubD second degree or changing degree of a SubD?

Hello,

I wish a second-degree SubD (with a radius 0.707 weight point) G0 or G1, not G2

So I can make a spherical sphere and cylindrical tubs. In my design, I use only a second-degree curve.
If Rhino builds such a tool will be the first and only one.

• The interior of every smooth edge is G1.
• Every quad face is G1

Thx,

1 Like

Moved to Serengeti/SubD category

1 Like

Hi, Alan,

Rhino SubD objects are standard Catmull Clark subdivision surfaces with support for creases, darts and corners. Away from exceptional points, these surfaces are nonrational bicubics.

One drawback of this choice is that Rhino SubD objects cannot have exact spherical regions, so you either have to approximate the spherical region with a subd or have a separate spherical NURBS object adjacent to your subd. The SubDSphere command is a way to explore a variety of ways to get good approximations to a spherical region.

Hi Dale,

Would it follow that there would be no equivalent subd result for cylindrical or revolved objects?

Comparing a cylindrical extrusion with a diameter of 100 mm to a SubD cylinder with 8 segments, the edge midpoints deviate 0.058 mm.

You conclusion is correct. It is mathematically impossible for a non-rational polynomial to have the shape of an exact circle. More generally, any type of ellipse or hyperbola requires a rational polynomial. Parabolic regions can be exactly represented by non-rational quadratics, and hence by non-rational cubics.

So, a standard Catmull Clark subd surface cannot have exact spherical and cylindrical regions. However, typically cubic approximations get pretty close without adding too many control points. If you need exact spherical / cylindrical regions in your model, those regions need to be modeled with Rhino polysurfaces.

1 Like

Yes, the purpose is having a simple flexible manipulator for NURBS. But a precise one.
having 4 elements (with example 4 points of control of 0.707)

Staregy
@John_Brock extending SubD flexibility and improving precision:
For controlling good meshes, NURBS is the best approach (Rhino strategy). For controlling NURBS surfaces, the SubD user interface approach is a good choice. And can be the main UI component of a new tool that control NURBs as a manipulator. So that the user is comfortable.

Yes, but is not possible to maintain that deviation across the model.
Until now, SubD is nice for making potatoes faster

It is more for undemaining bio-design. The Lotus Elise was done using this technology. But we know that is not appropriate for hard surfaces modelling.

That’s why video game car meshes look melted compare to a real car surface converted to mesh.