-2

I have a simple struct "mat4", consisting of float[4][4], and a *= function to multiply 4x4 matrices. It takes a const mat4& "rhs" as follows:

this->m[0][0] = this->m[0][0] * rhs[0][0] + this->m[0][1] * rhs[1][0] + this->m[0][2] * rhs[2][0] + this->m[0][3] * rhs[3][0];
this->m[0][1] = this->m[0][0] * rhs[0][1] + this->m[0][1] * rhs[1][1] + this->m[0][2] * rhs[2][1] + this->m[0][3] * rhs[3][1];
this->m[0][2] = this->m[0][0] * rhs[0][2] + this->m[0][1] * rhs[1][2] + this->m[0][2] * rhs[2][2] + this->m[0][3] * rhs[3][2];
this->m[0][3] = this->m[0][0] * rhs[0][3] + this->m[0][1] * rhs[1][3] + this->m[0][2] * rhs[2][3] + this->m[0][3] * rhs[3][3];

this->m[1][0] = this->m[1][0] * rhs[0][0] + this->m[1][1] * rhs[1][0] + this->m[1][2] * rhs[2][0] + this->m[1][3] * rhs[3][0];
this->m[1][1] = this->m[1][0] * rhs[0][1] + this->m[1][1] * rhs[1][1] + this->m[1][2] * rhs[2][1] + this->m[1][3] * rhs[3][1];
this->m[1][2] = this->m[1][0] * rhs[0][2] + this->m[1][1] * rhs[1][2] + this->m[1][2] * rhs[2][2] + this->m[1][3] * rhs[3][2];
this->m[1][3] = this->m[1][0] * rhs[0][3] + this->m[1][1] * rhs[1][3] + this->m[1][2] * rhs[2][3] + this->m[1][3] * rhs[3][3];

this->m[2][0] = this->m[2][0] * rhs[0][0] + this->m[2][1] * rhs[1][0] + this->m[2][2] * rhs[2][0] + this->m[2][3] * rhs[3][0];
this->m[2][1] = this->m[2][0] * rhs[0][1] + this->m[2][1] * rhs[1][1] + this->m[2][2] * rhs[2][1] + this->m[2][3] * rhs[3][1];
this->m[2][2] = this->m[2][0] * rhs[0][2] + this->m[2][1] * rhs[1][2] + this->m[2][2] * rhs[2][2] + this->m[2][3] * rhs[3][2];
this->m[2][3] = this->m[2][0] * rhs[0][3] + this->m[2][1] * rhs[1][3] + this->m[2][2] * rhs[2][3] + this->m[2][3] * rhs[3][3];

this->m[3][0] = this->m[3][0] * rhs[0][0] + this->m[3][1] * rhs[1][0] + this->m[3][2] * rhs[2][0] + this->m[3][3] * rhs[3][0];
this->m[3][1] = this->m[3][0] * rhs[0][1] + this->m[3][1] * rhs[1][1] + this->m[3][2] * rhs[2][1] + this->m[3][3] * rhs[3][1];
this->m[3][2] = this->m[3][0] * rhs[0][2] + this->m[3][1] * rhs[1][2] + this->m[3][2] * rhs[2][2] + this->m[3][3] * rhs[3][2];
this->m[3][3] = this->m[3][0] * rhs[0][3] + this->m[3][1] * rhs[1][3] + this->m[3][2] * rhs[2][3] + this->m[3][3] * rhs[3][3];

I just wanted to get confirmation whether it was correct or not - when I multiply two matrices in C++ (projection * view matrices) and give the resulting matrix to the shader, I get nothing on the screen showing up.

But if I give the shader projection & view matrices separately, and multiply them in GLSL - then it all works great, results are as expected.

So there must be something wrong with the matrix multiplication function?

oglglslnoob
  • 9
  • 7
  • 2
    Don't do the multiplication in-place, it won't work right. – Sneftel Dec 02 '17 at 22:54
  • see [Understanding 4x4 homogenous transform matrices](https://stackoverflow.com/a/28084380/2521214) and look for `matrix_mul` in the code (however the matrices are encoded in 1D array OpenGL style manner). You can also exploit dot product for this (on rows and columns or on rows and transposed rows...) in place is not possible as @Sneftel suggest because you overwrite values needed for computation of other cells ... you can buffer actual row instead of whole matrix however .... – Spektre Dec 03 '17 at 10:15

1 Answers1

0

Shouldn't

this->m[0][1] = this->m[0][0] * rhs[0][1] + this->m[0][1] * rhs[1][1] + this->m[0][2] * rhs[2][1] + this->m[0][3] * rhs[3][1];

be multiplying by rhs[1][0] .. rhs[1][3]? You are not stepping through the columns of rhs as you step through the rows of this.

SoronelHaetir
  • 14,104
  • 1
  • 12
  • 23
  • Hey, thanks and yeah.. ended up with a for loop instead, less code and more reliable.. Working great now! Cheers – oglglslnoob Dec 03 '17 at 00:08