That looks really close. The workflow would require laying out the back tangents, then fixing the curve to points along those tangents (the point of curvature (PC) and point of tangency (PT) ). So being able to explicitly control the positions of those points is critical. I would also think if three points are colinear, that would force the curve to a tangent (line segment)...microelly2 wrote: ↑Sat Apr 07, 2018 2:39 pmHere a bezier and a nearly approximation with two (lightblue) circles and a liner segment, all with tangent connections.

So this eliminates the complexity of managing many tangent constraints for individual arcs. But the back tangent structure still needs to be managed... and individual constrained points which force the curve to its shape...

All that assuming we can reliably get the curve to resolve to simple arcs...

Still.. I'm getting excited!