PointDeviation - Points from lines/polylines needed

PointDeviation - Points are only extracted at knots/control points of lines, polylines and degree 1 curves. This appears to be current intended result. Lines, polylines and degree 1 curves should be divided to obtain the requested number of points, the same as if Divide was used to obtain the test points.

I use PointDeviation extensively to check surfaces against input and reference lines and curves. Not having lines and polylines divided can be misleading without careful inspection. It is not obvious why the current behavior is considered desirable.

From Help for CurveDeviation: https://docs.mcneel.com/rhino/7/help/en-us/index.htm#commands/pointdeviation.htm

CurveDivisions (Curve-only)

Specifies how many sample points to spread along a measured curve to evaluate deviation.

  • If the curve to measure is pre-selected, this option will not appear. The previous setting is applied.
  • When a polyline is measured, polyline segment points are directly used. This option is ignored.

Simple example file: PointDeviationTest.3dm (33.9 KB)

@pascal @John_Brock Any possibility of PointDeviation being updated so that the extraction of points for lines/polylines/degree 1 curves works the same as the extraction of points from higher degree curves?

Hi David - would you ignore polyline vertices altogether?


For me ignoring polyline vertices would be okay. They are currently ignored by Divide, and in polycurves the junctions of individual curves ignored.

The reason for adding extracting points from lines, polylines, curves and polycurves in PointDeviation is to check how closely those items match other objects. I typically use 100 points or more per object. Only checking vertices for lines and polylines can be very misleading.

Hi David - they (vertices) are ignored by Divide but I am not sure that it the most relevant comparison here - it seems like a vertex will be likely to be more extreme, closer/farther than a nearby division point, is what I am thinking, so not including these seems potentially unnecessarily misleading as well. I’ll chat with the developer.

(also: PointDeviation request - add curves as input )



Why can only vertics or division points be included? Why not both?

Sometimes vertex points will be the most extreme than points between vertices, but other times they may be the closest points. My request was prompted by a lobster boat hull model I’m working on which has a straight keel. I used a line to represent the keel and then used Patch to obtain a preliminary surface. For such shapes the ends of the keel will typically be close than at lease some ares along the keel.

This differs from my response in the earlier thread due to my recent experience.


Hi David - right, that is where I was headed with my question above… if we divide the polyline, should we also include the vertices… Sounds like yes…


RH-60426 is fixed in the latest WIP