Quadro FX 540 and glReadPixels...

[kaloskagatos]

Hello ,

I work on a program which runs on different platforms (HP, SUN, Linux, IRIX). I use it though an XServer emulator (Exceed3D) from a Windows2000 PC.

All worked fine untill today when I used a Windows PC with Exceed3D and a Quadro FX 540. Now the glReadPixels function returns strange values such as shown by the following code:

glReadBuffer( GL_FRONT );
glReadPixels ( xmouse-1 , ymouse-1, 3, 3, GL_DEPTH_COMPONENT, GL_FLOAT, z_picking );

whitch return on a well working station :

z_picking = (0.6561477, 0.6561477, 0.6561477, 0.6561477, 0.6561399, 0.6561399, 0.6561399, 0.6561399, 0.6561399)

and on the nvidia PC :

z_picking = (-1.222404e+37, -1.222404e+37, -1.222404e+37, -1.222404e+37, -5.576056e+27, -5.576056e+27, -1.222404e+37, -5.576056e+27, -5.576056e+27)

The consequence is that I can't locate a point in my openGl scene using the gluUnProject function, and I musn't change it...

What am I doing wrong, or is there a problem with the driver? Thanks a lot for the time you'll spend answering me, and forgive my poor english

Greets,

Kalos

[kaloskagatos]

Up

[marc]

The call looks ok, but the result suggests that the size of 'GL_FLOAT' is incorrectly
set/interpreted somewhere along the line, because the numbers look like 32-bit floats in 64-bit double spaces...
I normally only do GLReadPixels with GL_UNSIGNED_BYTE as type (which has
kept me out of trouble so far)

[kaloskagatos]

Hi, you should be right but I've found it could be a bug from the driver or the openglLib...


http://www.rage3d.net/board/printthread.php?t=33777945

http://www.rage3d.com/board/showthre...3733550&page=7

[kaloskagatos]

I'm surprised this problem still exists... Does any one have a solution?

Marc, I've tried different combinaisons of type, nothing changes...

[kaloskagatos]

Any idea?