Establishing major axis of an elipse type closed

Anyone know the best way of creating an approx long axis for a closed curve which has a very close resemblance to an elipse but was not drawn using an elipse tool, its come in from photogrammetry so is a bit rough though I have cleaned it up with Fair. I am convinced the original shape is an elipse.
I have a custom CPlane for it but the shape is not horizontal.

If I can establish the axis that would be a good start and create an elipse to replace it.

Maybe its just a case of trial and error.

Google has just equations for establishing axes for actual elipses.


If it’s oriented horizontally or vertically, then you can use the Quadrant Osnap to find the max/min points on the ends of the curve. A line drawn between them would be the axis. I suppose you could use the area centroid command to see how close it is to the mid point of the line.

An additional clarification…
A curve drawn with the Rhino ellipse tool is not a special ellipse object. It’s just another NURBS curve, not a special unique entity type like it would be in say AutoCAD.

Hi John,
trouble is its a trace of a photogrammetry shape so has no actual Quads as yes Quads are the way.

Imagine someone takes an elipse, distorts it a tad, then simply retraces it with a curve tool with only Near Osnap on.

I see I have had no reply to a post entitled elipse through points…no command ?

that would do nicely here. I wonder that if there is circle through points, then why not elipse through points.


You’re going to have to get creative then.
I would get an area centroid point, then draw a line from that point perpendicular to the curve. That will be very close to half of the major/minor axis of your almost ellipse.

Guess so, or my elipse through point command wish :slight_smile: as edited into my post reply.

One for V6 maybe.

unless someone has code.


The major axis of an ellipse will intersect the ellipse at its points of maximum curvature. So if the curve includes both ends of the ellipse you could use the curvature graph to estimate the points of maximum curvature and draw the major axis between them. Or use John’s suggestion of the area centroid plus a point of maximum curvature.

Now thats a clever idea, never thought of that.

Wonder if someone has code for elipse through Pts.