Is “torsion continuity” a misnomer for G³ geometric continuity

continuitycurvesplane-curvesspline

I've recently been researching parametric vs geometric continuity of splines (piecewise polynomials) in 2D space.

The most common terms for each level of geometric continuity are:

  • $G^0$ is positional continuity
  • $G^1$ is tangential continuity
  • $G^2$ is curvature continuity
  • $G^3$ is torsion continuity

These all make perfect sense to me – except "torsion continuity" for $G^3$

One paper I've been reading, generalizes $G^n$ geometric continuity using so called $\beta$-parameters, so that any level of $G^n$ continuity can be achieved, regardless of the number of spatial dimensions of the spline

I have also implemented $G^3$ for 2D bézier splines, and it can be visualized using curvature combs. Note how the curvature comb itself has a tangent continuous join for $G^3$, unlike in $G^2$

Now, the concept of torsion of a curve is stated as follows:

the torsion of a curve measures how sharply it is twisting out of the osculating plane

However, as far as I can tell, the torsion of a 2D curve is always $0$ – this seems directly in contradiction of the fact that constraining curves by "torsion
continuity" ($G^3$) still works fine in 2D

Is calling $G^3$ "torsion continuity" a misnomer?

Best Answer

I think it's a misnomer. I'd say that G3 means that the derivatives of both curvature and torsion (with respect to arclength) are continuous at the junction. Continuity of torsion alone is a much weaker condition.

In design applications where people care about G3, they're usually working with planar curves. They will check that the curvature combs join smoothly (as you did), or they will check that surface reflections are G2. Obviously, in this scenario, torsion is irrelevant, since the torsion of a planar curve is always zero.