This document summarizes an experiment testing the performance of blendshapes on 3D models with split or combined meshes. The experiment found no noticeable performance differences between the two approaches when morphing or animating models under 60k triangles. Splitting meshes did increase draw calls as expected but did not meaningfully impact performance. The conclusion is that for models under 60k triangles, the performance cost of using split versus combined meshes for blendshapes is insignificant.