Open stla opened 3 years ago
No, I don't think so.
Handling the end conditions is just a bit more complicated and I didn't need this yet, so I didn't do it (yet?).
I'm preferring the splines.quaternion.CatmullRom
(via splines.quaternion.DeCasteljau
), because this directly yields the angular velocity. I've implemented splines.quaternion.BarryGoldman
just for comparison, and the closed one was sufficient for that.
Hello,
I'm currently working on implementing the Barry-Goldman quaternions spline in the R language. I follow your code. I am wondering why you set it closed. Is there a problem if we don't close the spline with this algorithm?