[vtkusers] get visible surface

[David Doria]

On Thu, Aug 23, 2012 at 8:09 AM, Lodron, Gerald
<Gerald.Lodron at joanneum.at> wrote:
> Thats funny, i copied my stuff from that example....
>
> I only have the problem that it only works on large rendererwindows like in my case on 1240 x 960. On 640 x 480 it does not work (result is completely empty then). So if you add
>
> renderWindow->SetSize(1240,960);
>
> before
>
> renderWindowInteractor->Start();
>
> the "broken" example should work (and of course disabling antialiasing and having a monitor with such a resolution). Another thing I did is that I use the
>
> extractSelection->PreserveTopologyOn();
>
> switch, but I think it also work without that line. The rest of my code is completely identical to your link...
>
> And I tested plugging out the monitor cable before starting my application (in this case without the key press stuff), then it crashes. I am thinking to produce a DUMMY monitor cable to solve the issue in my final system... (if the offline rendering switch will not work, not tested yet)

Hm, this is what I see:

Initial rendering:
http://daviddoria.com/Uploads/vtk/1.jpg

After selection:
http://daviddoria.com/Uploads/vtk/2.jpg

Rotating the selection:
http://daviddoria.com/Uploads/vtk/3.jpg

I thought what was happening is that the very corner of these "weird"
triangles is seen from the original perspective, so it included in the
selection, but to attempt to remedy this I made the triangles much
smaller using:

  sphereSource->SetThetaResolution(20);
  sphereSource->SetPhiResolution(20);

This time from the original viewpoint again all is well:

http://daviddoria.com/Uploads/vtk/bad_1.jpg

but after rotation, the pattern is clearly not just "the edge of the
sphere from the original perspective" triangles:

http://daviddoria.com/Uploads/vtk/bad_2.jpg

Many of these triangles are on the opposite side of the sphere from
the selection perspective. It really seems like a bug or something to
me.

David