Approximation nodes
 
 
 

Approximation nodes give you more precise control over all of mental ray for Maya approximation features, some of which Maya cannot access. For example, you can use Approximation nodes to specify separate tessellation settings for surfaces, trim curves, and displacement maps.

As an example, consider a simple flat NURBS surface with a complex trim curve. For such a case, you would specify a low-quality surface approximation in conjunction with a high-quality trim curve approximation. mental ray would then ensure that the surface is approximated with only a few triangles except around the trim curves, where many triangles would be used to ensure a smooth edge.

The same analogy applies to a simple surface with a complex displacement map. For that case, you might apply a low-quality regular surface approximation in conjunction with a high-quality displacement approximation, to ensure that triangles are added only to areas where they are needed to capture the complexity of the displacement map.

This node type... Does this... To only this type of geometry...
Surface Approximation Determines how NURBS surfaces are tessellated into triangles for rendering. NURBS surfaces
Trim Curve Approximation Controls the tessellation of trim curves on NURBS surfaces. NURBS surfaces with trim curves
Displacement Approximations Controls the tessellation of displacement maps on a surface. Whereas ordinary Surface Approximations only tessellate based on the underlying surface, Displacement Approximations additionally take into account features of the displacement map when tessellating. NURBS or polygonal surfaces with displacement maps
Subdivision Approximation Control render-time smoothing of polymesh surfaces. polygonal surfaces
Note
  • Subdivision surfaces are supported by mental ray versions prior to 3.2, when the mental matter library libmisubdiv.so is linked in. mental ray 3.2 and later includes subdivision surface rendering (but not modeling) support, and do not require an external library.
  • For best performance, when using subdivision approximation, use triangles and quads or a combination of both.