From: Angus Dorbie (dorbie++at++sgi.com)
Date: 04/25/2001 04:11:31
I think some concepts are getting blurred here. It's unclear whether
sheer should be on the viewing matrix or the projection matrix but since
this is a bit of a hack it's kind of a grey area. Ordinarrily I'd
suggest you try building your solution where the sheer transformation is
on the projection matrix BUT I don't see an easy way to do that in
Performer either.
I think you should just try your sheer matrix on viewing and see what
happens.
If that doesn't work then the other equivalent option is to use a pfDCS
at the head of two pfScenes, for left and right eyes, the scenes would
contain the same information except for the sheer matrix in the top
pfDCS. A viewing transformation is the inverse equivalent of a modelling
transformation and when you start operations like sheer and scale on the
viewing matrix the results can get counter intuitive and sometimes the
assumptions and optimizations in the code w.r.t. culling and deriving
the eye point etc might break. A scene pfDCS will work but a sheer in
the viewing matrix will probably work too. The statement about undefined
results is really a 'no promises' opt out clause if our optimizations
defeat your intentions.
Cheers,Angus.
Guillaume Smietanski wrote:
>
> Hello Performer Developper,
> I have a question about shearing matrix. We devellop stereo Displays
> with for one an Orhtogonal Projection. To have a stereo depiction we
> have to solution one is to translate and rotate the picture depending of
> parallaxe and the stereo effect we want. The Problem is that the picture
> is not really correct for the borders. The second one is to use a shear
> transformation with an matrix (with homogene coordinate) like:
> 1 sh 0 0
> 0 1 0 0
> 0 0 1 0
> 0 0 0 1
> This is possible with Open GL but it seams that it should not be
> possible with Performer: in the manual page of the pfChannel::setViewMat
> I can read that the "mat[rix] must be orthonormal or results are
> undefined". I am not really a mathematic specialist but if I don't make
> any mistake a shearing is not an orthonormal transformation. The pfDCS
> manual page confirm my opninon: I read about setMat that shearing is not
> allowed.
>
> Have you any experience whith shearing transformation ? If how did you
> do compute the transformation ?
> Thank you a lot for any remarks.
> Guillaume Smietanski.
-- For Performer+OpenGL tutorials http://www.dorbie.com/"Whenever there is a conflict between human rights and property rights, human rights must prevail." --Abraham Lincoln
This archive was generated by hypermail 2b29 : Wed Apr 25 2001 - 04:12:02 PDT