Problems with ShortPath


(Menno Deij - van Rijswijk) #1

When I try to get a curve using ShortPath on the attached surface from one point to the other point, the result is a botched curve that is not even on the surface.

Rhino5SR12problematic_shortpath.3dm (104.3 KB)

Edit: the point selection order matters (?)


#2

its much too small

try gumball shift click on one and resize by 1000. then use fitsrf on that messy surface with 3 and 3 degrees and 0.1 value (tolerance) that should simplify the surface within a usable range, just try if thats ok for you. after that use your shortpath and shrink it down by 0.001 again if you must.

one side of the surface is open it may have done a short path along the open surface because it was far too small first. see if the file helps you. i attached the file resized to x1000. working in that small space will always cause problems in general.

problematic_shortpath-2.3dm (2.8 MB)


(Menno Deij - van Rijswijk) #3

I don’t agree with your observation that size is a problem, Rhino can work with any size object large or small. The only known size problems with render meshes far from the origin.

The problem is indeed the surface (which is not messy IMHO) and its performance in the short-path algorithm.


#4

definitely not if the cplane and the working environment are far bigger,
but also the other way around if objects are far bigger.

is the object meant to be 0.02 mm wide at all? what is it if i may ask?
looks like a part of a wing…


(Menno Deij - van Rijswijk) #5

It is in fact the tip of a wind turbine blade, scaled to a radius of 1.0.


#6

you can see here that it makes a difference that your object is far too small in relation by trying to intersect the rebuilt surface with the short path in your original scale, the intersection will strongly deviate. whilst doing the same rescaled to a better cplane object relation the curves get congruent.


(Pascal Golay) #7

Hi Menno- I guess high power output is not the primary concern for this turbine! (dimensions = 0.006, 0.027, 0.113 millimeters)
That is , overall very close to the file tolerance. I’m with Richard - scale the thing up.

-Pascal


(Menno Deij - van Rijswijk) #8

Nope, scaling by a factor of 1000 does not help. Neither does making the document tolerance more strict.


(Pascal Golay) #9

Yeah, I see - the back edge is open as well - at least when scaled up. But ShortPath just does not like this shape, even a rebuilt or refit version has the problem, so far. Points picked a little more to the fat part of the shape sends ShortPath correctly around the nose and it looks OK.

https://mcneel.myjetbrains.com/youtrack/issue/RH-40727

-Pascal


(Menno Deij - van Rijswijk) #10

Yeah, the back edge is supposed to be open. I was hoping that ShortPath would, you know, “just work” :wink: