[vtkusers] using vtkImplicitFunction in two ways to clip a cylinder

[Dean Inglis]

David,

thanks for the clarification: it was the inside outside aspect I was missing 
...

Dean

----- Original Message ----- 
From: "David Gobbi" <david.gobbi at gmail.com>
To: "Dean Inglis" <dean.inglis at camris.ca>
Cc: <vtkusers at vtk.org>
Sent: Tuesday, May 03, 2011 2:13 PM
Subject: Re: [vtkusers] using vtkImplicitFunction in two ways to clip a 
cylinder


Hi Dean,

The VTK clipping/cutting functions work by sampling the implicit
function at the at the points that make up the data set.  This means
that the function should have at most one zero between any two
points that make up a cell edge.

It isn't just "closeness" that matters, it is also necessary that the
implicit value is positive on one side of the surface, and negative
on the other.  So yes, it would possible to use normals to establish
an in-side and an out-side, and it would be possible (but difficult) to
write a polydata-to-implicit-function class based on this idea.

 - David


On Tue, May 3, 2011 at 11:48 AM, Dean Inglis <dean.inglis at camris.ca> wrote:
> Hi David,
>
> with vtkPlane as an implicit function, EvaluateFunction is a continuous
> scalar function that returns the dot product of the plane normal
> with the vector defined by the test point and the plane origin: in plane
> points from an orthogonal vector to the normal and hence a 0 value is
> returned,
> otherwise the magnitude of the projection of the resulting vector on the
> normal is returned.
> With vtkImplicitData set, if all of its scalar values are a constant, say 
> 1,
> then a point on the data set, if it is a surface, will return a 1 and
> OutValue
> otherwise: hence a delta function. Either way, one knows if a test point
> is on or off the surface. In the latter case, one has no indication of
> "how close". Why do filters like vtkClipPolyData and vtkCutter need
> a measure of (for lack of a better term ) "closeness" ?
>
> Would having a modified vtkImplicitDataSet work if the data set
> were required to return a continuous value in the case when the
> the test point didnt evaluate to being on the surface (maybe using surface
> normals somehow)?
>
> thanks,
> Dean
>
>
> ----- Original Message ----- From: "David Gobbi" <david.gobbi at gmail.com>
> To: "Dean Inglis" <dean.inglis at camris.ca>
> Cc: <vtkusers at vtk.org>
> Sent: Tuesday, May 03, 2011 1:20 PM
> Subject: Re: [vtkusers] using vtkImplicitFunction in two ways to clip a
> cylinder
>
>
> Hi Dean,
>
> I get the feeling that you are looking for something similar to
> vtkImplicitSelectionLoop for polydata surfaces. No such thing exists
> in VTK, that is why there is so many people have expressed interest
> in the VTK Journal submission that clips polydata with polydata.
>
> The way vtkImplicitDataSet works is similar to vtkImplicitVolume. The
> implicit value is interpolated from the data, and everywhere outside
> the data has a value of OutValue. So clipping with vtkImplicitDataSet
> works well if the cells are volumetric, but if the cells are 2D, then you
> end up with something akin to a delta function.
>
> - David
>
>
> On Tue, May 3, 2011 at 10:48 AM, Dean Inglis <dean.inglis at camris.ca> 
> wrote:
>>
>> Im having a problem getting a proof of principle established
>> that demonstates clipping with vtkClipPolyData using vtkImplicitDataSet.
>> I would ultimately like to clip a vtkPolyData with an arbitrary
>> (non-parametric)
>> surface defined by vtkPolyData composed of 2D cells (vtkTriangle,
>> vtkTriangleStrip
>> etc.). I have coded two simple demos in tcl: one uses a vtkPlane as the
>> vtkImplicitFunction and is controlled interactively with
>> vtkImplicitPlaneWidget,
>> the other uses a vtkPolyData output from a vtkPlaneWidget. In the latter
>> case
>> I am using vtkImplicitDataSet as the function by assigning a scalar array
>> (of 1's) to the
>> polydata, but no clipping occurs. Can anyone offer suggestion on how to
>> get
>> the latter demo working?
>>
>> best,
>> Dean
>
>