From a performance standpoint does the total number of vertices in a model, which uses blend shapes, matter when the blend is performed? Or is the overhead contained in just the delta of the vertices that move?
For example, say I have a model which has 20000 verts and 200 of them move during a blend. Would that be more expensive (to blend) than a model that has 400 vertices with 200 moving during the blend.
The reason I ask is I have a character for which, from a production pipeline pov, it would be useful if it was all one mesh. But it has a very small number of blending vertices which are currently a separate object.
If I were to attach the blending object to the main mesh, would I see a performance hit?
Thanks in advance