FreeForm and outliers

[[Br...]]

The outliers on a freeform surface are killing me. Once in a while i get crazy numbers showing the result of a profile characteristic at like 5mm or something like that. I can see from the Graphical analysis that its just a single little outlier. Has anyone found any way to deal with outliers in a freeform surface?

[[Er...]]

I know, sometimes I almost punch some one, makes me furious 🙂

The only "work around" I have is to delete those points/segments. Thats also the reason I dont use long scanning paths. I try to keep them short, but many. Or use single points if not too time consuming. (Usually I fix this with sub sequent eval, and I don't save the changes, I also inspect the parts visual, to be sure I dont delete deviations that exist, welding burrs, scratches, etc. )

Really wish we could set out liers based on sigma.


How ever, I have Been thinking about create a work around using scipy, and mathplot. And simply export the points, filter out "out liers" and plot the surface deviations. It also gives you alot more flexibillity with the plot it self. But time to write this is not something I have. But if I ever get time, Ill post it here.