The subdivision is done in PoseRay. POV-Ray does not have subdivision. PoseRay subdivides the model before sending it to POV-Ray for rendering.

The subdivision will keep sharp edges (from normals) and can keep existing mesh edges unmodified while changing the rest of the mesh. Also it can be applied to a particular group or material without gaps between the subdivided and not subdivided surfaces. These will be all user selectable options.

The algorithm implemented in PoseRay is the modified Loop's algorithm to account for edges and corners. The UV subdivision is also using the modified Loop algorithm and will keep the UV edges on the mesh.

From the renders I have seen here I beleive Poser 5 uses an interpolating smoothing routine. Meaning that the subdivided mesh goes through the original mesh points. Loop's method is approximating (like a bezier and its control points) Interpolating routines, if it does not check edges, will give meshes an inflated appeareance.

Using Loop's (or any other approximating method) as the number of subdivisions is increased the mesh will resemble a NURBS surface.

later,

FlyerX