| Re: Don't Extrapolate with GRID3 (or cut off excess) [message #71968] |
Mon, 02 August 2010 07:50 |
asdf
Messages: 7
Registered: August 2010
|
Junior Member |
|
|
On Jul 30, 11:48 am, Paulo Penteado <pp.pente...@gmail.com> wrote:
> On Jul 30, 12:33 pm, asdf <rmc...@gmail.com> wrote:
>
>> First, a little background. I have an irregular grid of 3D volume data
>> that I would like to visualize (currently using iVolume). The volume
>> is basically a wedge piece of a cylinder.
>> I've the placed data onto a regular, cube grid using GRID3, but the
>> result has a large, unrealistic value in an area outside of the
>> original irregular grid. From GRID3 documentation, it doesn't seem
>> like I can force it to ignore points outside the original grid, i.e.
>> don't extrapolate (correct?). I'm wondering if someone knows an
>> efficient way to set to NaN values in the regular grid that are
>> outside the original, irregular grid. It seems I can do a Delaunay
>> triangulation of the irregular grid, and use it somehow to find if a
>> point in the regular grid is inside the irregular grid, but I'm lost
>> on the details of how to do that.
>
> If your distribution is convex, you can get what you want with
> qgrid3(), instead of grid3(), as it allows to set a value for the
> points outside the convex hull.
>
> If it is not convex, things are trickier. If the original points all
> fall into a series of planes, you could make an IDLanROIGroup with
> those, then use the containspoints method to determine which points of
> the grid fall inside it.
Thanks for the reply, that works great:
qhull,x,y,z,tr,/delaunay
result=qgrid3(x,y,z,data,tr,missing=!values.d_nan)
Even faster than grid3 also.
|
|
|
|
| Re: Don't Extrapolate with GRID3 (or cut off excess) [message #71996 is a reply to message #71968] |
Fri, 30 July 2010 08:48  |
penteado
Messages: 866
Registered: February 2018
|
Senior Member
Administrator |
|
|
On Jul 30, 12:33 pm, asdf <rmc...@gmail.com> wrote:
> First, a little background. I have an irregular grid of 3D volume data
> that I would like to visualize (currently using iVolume). The volume
> is basically a wedge piece of a cylinder.
> I've the placed data onto a regular, cube grid using GRID3, but the
> result has a large, unrealistic value in an area outside of the
> original irregular grid. From GRID3 documentation, it doesn't seem
> like I can force it to ignore points outside the original grid, i.e.
> don't extrapolate (correct?). I'm wondering if someone knows an
> efficient way to set to NaN values in the regular grid that are
> outside the original, irregular grid. It seems I can do a Delaunay
> triangulation of the irregular grid, and use it somehow to find if a
> point in the regular grid is inside the irregular grid, but I'm lost
> on the details of how to do that.
If your distribution is convex, you can get what you want with
qgrid3(), instead of grid3(), as it allows to set a value for the
points outside the convex hull.
If it is not convex, things are trickier. If the original points all
fall into a series of planes, you could make an IDLanROIGroup with
those, then use the containspoints method to determine which points of
the grid fall inside it.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
