Something I just noticed :
// Gives INCORRECT Z value
GLdouble l_dZ;
glReadPixels(a_iX, l_iY, 1, 1, GL_DEPTH_COMPONENT, GL_FLOAT, &l_dZ);
// Gives CORRECT Z value
GLfloat l_fZ;
glReadPixels(a_iX, l_iY, 1, 1, GL_DEPTH_COMPONENT, GL_FLOAT, &l_fZ);
So always use GLfloat for reading the depth buffer if you want correct values.
That should be pretty obvious. Of course you can't pass a pointer to a double and tell it to pass a float and expect to get a correct result.
Exactly, try specifying a GL_DOUBLE for a double type not a GL_FLOAT. That pointer is a void so there's no concept of intrinsic type other than what you specify in the type token, like GL_FLOAT. That type is the destination type not the source type, OpenGL already knows the source type. There is no function overloading here, OpenGL is a C interface, not C++ and it matters in this respect.
I just wanted to emphasise this problem.
This mistake is made in several examples I found on the web.
This could save some people a lot of debugging / posting time.
NOTE : GL_DOUBLE is an invalid enumerant for the glReadPixels function, for reading the Z buffer only GL_FLOAT is valid !
Ofcourse it would be coolto have a 64bit Z buffer !
If everything was in 64 float, then it would make sense, but the graphics card does a significant number of operations in 32 bit float, and the rest is integer or bytes.
------------------------------
Sig: http://glhlib.sourceforge.net
an open source GLU replacement library. Much more modern than GLU.
float matrix[16], inverse_matrix[16];
glhLoadIdentityf2(matrix);
glhTranslatef2(matrix, 0.0, 0.0, 5.0);
glhRotateAboutXf2(matrix, angleInRadians);
glhScalef2(matrix, 1.0, 1.0, -1.0);
glhQuickInvertMatrixf2(matrix, inverse_matrix);
glUniformMatrix4fv(uniformLocation1, 1, FALSE, matrix);
glUniformMatrix4fv(uniformLocation2, 1, FALSE, inverse_matrix);
Whoa, a 64bit Z buffer would ROCK!
Posting Permissions
