Do the methods in Jitter3DUtils like transformPoint assume column vectors multiplied by a matrix on the left or row vectors with a matrix on the right ? It's all going into opengl in the end so I'm assuming column vectors but I thought I'd check.