Hi Joach,

if the deviation is somewhere in the order of E-7 or E-8 that's
probably well below the document tolerance which is usually something
like E-2 or E-4.
This is a result that fits with expectation. You'll have to compare
the deviation to a custom number yourself if you want a 'fuzzy'
planarity test.

--
David Rutten
[email protected]
Robert McNeel & Associates


On Mar 8, 11:36 pm, joach <[email protected]> wrote:
> Well, exhausting answer. I understand a theory. But I'm still a little
> bit in a stew when I compare a result from planar component (planarity
> boolean and deviation number). A deviation is app. zero (E -7^10), but
> result of planarity dont correspondent with this results. Why?
>
> link of 
> screen:http://grasshopper3d.googlegroups.com/web/planar%20curves%20-%20devia...
>
> Joach
>
> On Mar 8, 10:54 pm, David Rutten <[email protected]> wrote:
>
> > Hi Joach,
>
> > The intersection between two surfaces is usually a very complicated
> > curve. If there exists a nurbs curve which represents this
> > intersection perfectly, chances are it will need a massive degree
> > (think 30) and an enormous amount of control points. This is why we do
> > not provide those curves as results.
>
> > Internally, Rhino calculates the intersection to a much higher
> > accuracy than you request. The thousands of intersection points it
> > finds between the surfaces are fitted with a single nurbs curve. This
> > nurbs curve is not allowed to deviate from the intersection points by
> > more than the tolerance. So, even if all the intersection points are
> > on the same plane, it is possible that the fitting curve will deviate
> > from this plane by a maximum of <tolerance> units.
>
> > When we test for planarity, we also need to pick a tolerance. If you
> > have a flat curve with no z-elevation at all, then it's obviously
> > flat, but if you rotate the curve so that the plane is tilted along
> > each axis, then the very accuracy of floating point numbers will cause
> > it wobble, ever so slightly away from the ideal plane.
>
> > The planar component in Grasshopper is set to maximum accuracy. I'll
> > adjust it to only operate within current Document Absolute Tolerance.
> > Incidentally, the planar component will output the deviation from the
> > ideal plane. You could also compare this to a numeric value and so
> > decide upon your own tolerance within Grasshopper. What is the
> > deviation that you get?
>
> > --
> > David Rutten
> > [email protected]
> > Robert McNeel & Associates
>
> > On Mar 8, 10:06 pm, joach <[email protected]> wrote:
>
> > > It wasnt visually? I used planar component for fast test and there
> > > isnt possible set a tolerance. When I used VB.net definition
> > > (.isplanar) and your tip its work perfect. Thanks for you David for
> > > your quick answer!!! What exactlly is mean by tolerance?
>
> > > Thank for GH, a special with vb.net component it is so strong tool.
> > > GOOD WORK!!!!!
>
> > > Joach
>
> > > On Mar 8, 9:38 pm, David Rutten <[email protected]> wrote:
>
> > > > Hi Joach,
>
> > > > if the intersection tolerance is less than the planarity test
> > > > tolerance, curves will appear to be non-planar. Or are you getting
> > > > visually wacky results?
>
> > > > --
> > > > David Rutten
> > > > [email protected]
> > > > Robert McNeel & Associates
>
> > > > On Mar 8, 9:26 pm, joach <[email protected]> wrote:
>
> > > > > When I make a intersection of two Breps (first is from onNurbsSurface,
> > > > > second is a plane surface define from a frame of curve, function
> > > > > pointAt(w = 0) ) result of this operation is a array of onCurve. How
> > > > > is possible that curves arent planar? tolerace of intersection of brep
> > > > > is set to 0.01
>
> > > > > Any idea? Thanks
> > > > > Joach

Reply via email to