Hi,
>1. In argInit(...), does anyone know why the the
>second row of the 3x4 projection matrix is subtracted
>from a multiple("image height1") of the third row?
>This is equivalent to turning the image coordinate
>system upside down, I think.
Yes.
>2. In argConvGLcpara2(...), after the QR
>decomposition, when the projection matrix is converted
>for OpenGL: Can anyone point me to a reference that
>explains this operation, or tell me what and why it is
>being done? It seems to be a basic operation, I don't
>know.
In the ARToolKit, screen coordinates are represented in
PIXEL unit. For example, a image with 640x480 size has
[0639] range of value in X axis and [0479] range of value
in Y axis.
In order to use such camera parameters for openGL,
it should be convert for openGL representation.
Basically openGL uses [1,1] range of value for normalized
clipping volume representation in X,Y,Z axis.
So [0649] range and [0479] range should be convert to [1,1] range.


Hirokazu Kato
Faculty of Information Sciences
Hiroshima City University Phone: 0828301705
Email: kato@s ........................ Fax: 0828301435
URL: http://www.sys.im.hiroshimacu.ac.jp/people/kato/
