When computing vertex normals, if normals of vertices around a single point are less than or equal to this angle (in degrees) apart from each other, they will be averaged together, weighted by the vertex angle in each polygon.

When set to By Vertex Angle, care is taken when computing normals to try to avoid issues introduced by sliver polygons and to weight contributions to normals by angle, so that triangulation won’t change normals significantly. When set to Each Vertex Equally, a faster, but less accurate, approach similar to the approach used by the Facet SOP, is used to compute normals, where each vertex is given equal weight around a point, and less checks are done for ensuring stability. When set to By Face Area, the behavior is similar to __Each Vertex Equally_, except that vertices on faces with more area will be given more weight when computing point normals. This is often used to fake the appearance of a smooth bevel for geometry with only a simple chamfer.

The point or primitive group to which you apply softening.

Facets the geometry in certain ways before applying the cage offset (Ray Start Offset). A soft cage isn’t faceted and ensures a seamless result across sharp edges, but can cause projection errors and skewed detail as the normals are averaged. An unsoftened cage (which is faceted) can fix this projection skewing, but can cause missing detail capture on sharp edges.

The method used to cut up the geometry.

The edge group by which to split the geometry.

How far the duplicate mesh peaks out from the normals of the original low-resolution mesh. This is the ray start position or the 'Cage'.

The attribute value by which to scale the Ray Start Offset.

How far the duplicate mesh peaks in from the normals of the original low-resolution mesh. This is the ray end position. This is turned into the raylength output attribute by calculating the distance difference between the Ray Start Offset and the Ray End Offset.

The attribute value by which to scale the Ray End Offset.