This involves creating a micro trifan of thirty-six small triangles around the hub vertex, and then using the weighted average of the trifan normals. Despite this being a brute force method, there is no appreciable difference between the weighted average normal and the envelope normal. After extensive testing on various envelopes, the hub plane angles were found to be the same in both cases. This page shows the results using an enlarged cylinder hub with a shallow depth to make a disc at each vertex. The discs accurately trace the envelope contours:

supercircle of revolution

ellipse of revolution

superellipse of revolution

ellipsoid

superellipsoid

egg of revolution

axolotyl of revolution

bell of revolution

teardrop of revolution

omphalos of revolution