btMatrix3x3::setFromOpenGLSubMatrix transpose the matrix

[beaelp]

Hi,

I was getting the transform to set the position of already created bodies with the function btTrasform::setFromOpenGLMatrix().
This function calls m_basis.setFromOpenGLSubMatrix(m) and m_origin.setValue(m[12],m[13],m[14]) which is OK.

The problem comes in m_basis.setFromOpenGLSubMatrix(m): It takes each ROW of the final body matrix that are the vectors m_el and assign them the COLUMNS of the m matrix. So it results in a transpose matrix that when the object is created with a rotation it gives completely opposite results.

//This is the original function
void setFromOpenGLSubMatrix(const btScalar *m) {
m_el[0].setValue(m[0],m[4],m[8]);
m_el[1].setValue(m[1],m[5],m[9]);
m_el[2].setValue(m[2],m[6],m[10]);
}

//This is the function I think it's correct
void setFromOpenGLSubMatrix(const btScalar *m) {
m_el[0].setValue(m[0],m[1],m[2]);
m_el[1].setValue(m[4],m[5],m[6]);
m_el[2].setValue(m[8],m[9],m[10]);
}

If this is true, the corresponding function to get the opengl matrix btMatrix3x3::getOpenGLSubMatrix needs to be modified too.

//Original function:
void getOpenGLSubMatrix(btScalar *m) const
{
m[0] = btScalar(m_el[0].x());
m[1] = btScalar(m_el[1].x());
m[2] = btScalar(m_el[2].x());
m[3] = btScalar(0.0);
m[4] = btScalar(m_el[0].y());
m[5] = btScalar(m_el[1].y());
m[6] = btScalar(m_el[2].y());
m[7] = btScalar(0.0);
m[8] = btScalar(m_el[0].z());
m[9] = btScalar(m_el[1].z());
m[10] = btScalar(m_el[2].z());
m[11] = btScalar(0.0);
}

//Function I think is correct
void getOpenGLSubMatrix(btScalar *m) const
{
m[0] = btScalar(m_el[0].x());
m[1] = btScalar(m_el[0].y());
m[2] = btScalar(m_el[0].z());
m[3] = btScalar(0.0);
m[4] = btScalar(m_el[1].x());
m[5] = btScalar(m_el[1].y());
m[6] = btScalar(m_el[1].z());
m[7] = btScalar(0.0);
m[8] = btScalar(m_el[2].x());
m[9] = btScalar(m_el[2].y());
m[10] = btScalar(m_el[2].z());
m[11] = btScalar(0.0);
}

I hope this make sense and can be fixed.
Bea.

[Dominik]

Actually, the "transposed" version is correct, since the OpenGL convention of storing matrices is row-major, as opposed to the common column-major.

[loefje]

Actually, the "transposed" version is correct, since the OpenGL convention of storing matrices is row-major, as opposed to the common column-major.

sorry, but this is not true, see the OpenGL FAQ on
http://www.opengl.org/resources/faq/tec ... tions.htm
I (accidentally) started a similar topic here:http://continuousphysics.com/Bullet/php ... 1b1a7d0566

but I'm still confused.

to me, it seems beaelp is right in saying that the
getOpenGLSubMatrix gives back the transpose of an
OpenGL matrix expected by ao glLoadMatrix

J

[Dominik]

sorry, mixed up row- and clumn-major... Still, its correct that opengl uses the "transposed" notation as seen from bullet (and most other toolkits)
an OpenGL

Code: Select all

| a b c d |
| e f g h |
| i j k l|
| m n o p |

is stored as mat[16]={ a e i m b f j n c g k o d h l p }
while most other use the row-major one