After examining the Three.js example found at webgl_performance, I decided to try improving performance by converting the model into a buffer geometry using the following code:
var buffer = THREE.BufferGeometryUtils.fromGeometry( geometry );
Despite my expectations of a performance boost (thanks to the buffer geometry magic), I was surprised to find that both the original and altered versions were yielding approximately the same frames per second (~20).
Links to the original and altered versions:
Original: webgl_performance
Altered: webgl_performance_buffer_geometry
This outcome sparked my curiosity - could it be that the json loader already utilizes buffer geometries, rendering my conversion unnecessary? Shouldn't I always convert to buffer geometries when it's possible for optimization purposes?