NURBS versus polygons. An equal number of surface divisions can allow for control of the surfaces differently. Adding resolution to a polygon surface allows you to achieve smoother bends; adding resolution to NURBS helps to make sharper bends.

These two arm models have approximately the same shape and number of vertices. If your supervisor asks you to adjust the shape of the deformed elbow, which do you think would be the easier to deal with?

Let’s look at an example of an arm built out of NURBS and an arm built out of polygons. We will place a three-joint chain within the arm’s geometry and Smooth Skin each piece of geometry. Polygons and NURBS differ in many ways when they are being deformed. You’ll see the difference between these two types of surfaces when the arm is bent. If you want a nice smooth bend, you may need more points in a poly mesh in order to get a smoother transition. On a NURBS surface, you might need to take away resolution to achieve a smoother transition, as you can see in the arm on the left.

Resolution
On a polygonal surface, a bend occurs only at the location of the vertices. If you bend 90 degrees between two vertices or one cross section, the result is much sharper than if you bend 90 degrees between four vertices. Four vertices allow each of the three cross sections to rotate 30 degrees. This propagates the rotation across the surface, allowing for a less sharp angle at the bend.

If a hard break is what you’re after, your NURBS surface will need increased resolution in the area around the bend to allow that area to conform closer to the hulls (see note that follows). You must keep in mind that deformations only happen on the CVs that control the surface, not on the edit points that lie on the surface.

Distribution
As shown previously, the amount of resolution in any type of model affects how it will deform. Like with resolution, to create a surface that deforms appropriately, it is very important that your cross sections (polygons) or isoparms (NURBS) are properly distributed. It’s also important to make sure that cross sections and isoparms are properly oriented to the joints that will deform the surface. As a general rule of thumb, you look for the geometry to be created with cross sections that lie perpendicular to the joint chains that you’ll build when the geometry is complete. Why should you base the creation of your surface on the joints you haven’t built yet? Because most deformations are a result of a change in a joint’s transformation. If the relationship between the geometry and joints is set up in an organized perpendicular fashion, you’re way ahead of the game in terms of creating a character that deforms properly. Be sure to keep in mind that more resolution isn’t always the answer; what you really want is the simplest model possible that will deform appropriately. More resolution means more points to weight, sculpt, cluster and lattice into their correct positions, and in the end, this means you’ll need more time to finish the job.