I’m about to create a character with a joint facial rig for the first time and I decided that I will want to use the rivet MEL technique.
In my travels, I found a post on djx’s blog where he wrote his own version of the rivet MEL.
To refresh everyone’s memories, the classic rivet script uses two edges of a mesh to create two curves, which then are lofted together, the resulting NURB is read with a pointOnSurfaceInfo and the parameter UV coords from said node define the position of a locator. He uses the pointOnSurfaceInfo node, along with a modified aim constraint, to achieve proper orientation for the locator so that it follows the orientation of the mesh face.
In contrast, djx’s rivet uses a Maya follicle node to constrain two objects together. One object contains the follicle and the other is driven by said follicle.
I don’t need the advantages of djx’s rivet over the classic rivet so it really comes down to speed. Which of these two systems, if used hundreds of times, would evaluate faster? Would either of them make a noticeable difference?
