vtkCamera issues

[David Gobbi]

Hi Ken,

I hadn't realized that the purpose of setting the VPN different
from the camera->focal point direction was for creating oblique
projections.  There were no regression tests that used oblique
projections (except for the two that broke, and they weren't
supposed to).

A method SetObliqueAngles(phi,theta) might be a better way to
create an oblique projection than directly setting the VPN --
it would make it easier for users to immediately grasp what
is going on.

I'll leave the vtkCamera code as-is until we resolve these issues.

 - David

--
  David Gobbi, MSc                    dgobbi at irus.rri.on.ca
  Advanced Imaging Research Group
  Robarts Research Institute, University of Western Ontario

On Thu, 20 Apr 2000, Ken Martin wrote:

> At 09:34 PM 4/19/00 -0400, David Gobbi wrote:
> >Hi All,
> >
> >There were a couple regression tests, 'AllWriters.tcl' and
> >'byuReader.tcl,' which were failing because they didn't call
> >ComputeViewPlaneNormal() after setting up the camera.
> >
> >These examples used to produce some weird shear projection because
> >vtkCamera used to compute the View matrix based on a combination of
> >Position-FocalPoint and ViewPlaneNormal.  My recent vtkCamera
> >changes compute the View matrix using just Position-FocalPoint
> >(and ViewUp, of course).
> >
> >So my question is this:  how would people feel if I deprecated
> >SetViewPlaneNormal() and caused the ViewPlaneNormal to be
> >automatically calculated every time SetPosition() or
> >SetFocalPoint() was called?
> 
> What are you saying here David? Are you saying you are going to remove the 
> ability to do off axis projections from VTK ? I don't think that is a good 
> option if that is what you are suggesting. The reason we have View plane 
> normal is because it isn't the same as Direction of Projection (VPN is not 
> necc == DOP) The vector from the position to the focal point is the DOP. 
> The VPN is the normal of the viewing plane.
> 
> >If this was done, paranoid code like this (straight from
> >the VTK regression tests) could hopefully be avoided:
> 
> The code below has an extra line anyhow. There is no need to compute the 
> view plane normal and set it. Just one or the other. So you are talking 
> about eliminating one line of code.
> 
> >vtkCamera aCam
> >     aCam SetFocalPoint 0.00657892 0 2.41026
> >     aCam SetPosition -1.94838 -47.1275 39.4607
> >     aCam ComputeViewPlaneNormal
> >     aCam SetViewPlaneNormal -0.0325936 -0.785725 0.617717
> >     aCam SetViewUp 0.00653193 0.617865 0.786257
> >     aCam SetClippingRange 1 100
> >
> >and replaced with more concise code:
> >
> >vtkCamera aCam
> >     aCam SetFocalPoint 0.00657892 0 2.41026
> >     aCam SetPosition -1.94838 -47.1275 39.4607
> >     aCam SetViewUp 0.00653193 0.617865 0.786257
> >     aCam SetClippingRange 1 100
> >
> >Any thoughts?
> >
> >    - David
> >
> >--
> >   David Gobbi, MSc                    dgobbi at irus.rri.on.ca
> >   Advanced Imaging Research Group
> >   Robarts Research Institute, University of Western Ontario
> >
>