# Help needed with finding centre line through 2 curves

Well this method will also have some tolerance issues (even if very minor) because interpolating those midpoints does not guarantee the knot vector matches with the initial curves in some way. For instance using uniform vs chord vs square root to interpolate.

The point is every method has its issues. pick what you like best cuz unless those two curves are exact and you are tweening between or averaging, then there is no 100 percent full proof way. It is some what subjective.

Ok would this tween curve be considered smooth?

No method is perfect until they are perfect circle sharing center. Then it becomes simple case of offset. Anything under the precision required, is an acceptable result.

1 Like

Depends what you need it for.

Well, this is out by quite alot will have to re think.

What methods did you use here?

I went from my rebuilt curves on to tween

Rebuild is always a very loose way to go I think. Maybe try fitcrv to different degrees. Maybe to 4 or 5 even.

What method are you actually using to measure the dimension in the middle? Because in the middle, it can never be equal unless its an offset of a curve.

@Michael_Pryor - TweenCurves works well here I tihnk but you need to align the end points - the default is crossed up.

-Pascal

I am dividing the central curve then bisecting from midpoint from the point immediately left and right, extending that line past the boundaries, trimming then measuring. This is the best i have managed but still upto 4mm off
f

As Pascal showed, the end points of curves has to be perpendicular as if it offset to curve then you can crank up the precision.

I see. Is there any difference between refit tween option and fitcrv? Do they do essentially the same thing or is refit fitting in some way which compares the input curves that will be tweened?

ok, i made the end lines perpendicular and extended curve to meet but the curve has extended in an odd way

Hi Michael - @rajaa will know best but I would think the fit option makes compatibly refit curves,so to speak, whereas `FitCrv` on its own would not necessarily make them so. (but it does raise the question… should it be an option with a multiple selection - -make the output compatible. I need to know more about how this stuff works, I’ll ask.)

-Pascal

which option did you used to extend curve?

extend curve-extend curve

I see it is what I suspect also but would be interested to know what the conditions are for refit compatibility.

type: natural or arc or line or smooth?

yep, all give similar results to the pic in my post