Mercurial > fife-parpg
changeset 58:9ec572a5a944
reworked matrix vector multiplication, patch by yonibear
| author | spq@33b003aa-7bff-0310-803a-e67f0ece8222 |
|---|---|
| date | Mon, 14 Jul 2008 16:39:27 +0000 |
| parents | 47979f9dce55 |
| children | 129d3dafd3a5 |
| files | engine/core/util/math/matrix.h |
| diffstat | 1 files changed, 5 insertions(+), 19 deletions(-) [+] |
line wrap: on
line diff
--- a/engine/core/util/math/matrix.h Mon Jul 14 14:41:12 2008 +0000 +++ b/engine/core/util/math/matrix.h Mon Jul 14 16:39:27 2008 +0000 @@ -210,25 +210,11 @@ /** Transform given point using this matrix */ inline PointType3D<T> operator* (const PointType3D<T>& vec) { - PointType3D<T> ret(T(0)); - for (register unsigned j = 0; j < 3; ++j) - for (register unsigned i = 0; i < 3; ++i) - ret.val[j] += vec.val[i]*m[j+i*4]; //scale and rotate disregarding w scaling - - for (register unsigned i = 0; i < 3; ++i) - ret.val[i] += m[i+3*4]; //translate - - //do w scaling - T w = m[15]; - for (register unsigned i = 0; i < 3; ++i) - w += vec.val[i]*m[3+i*4]; - - register T resip = 1/w; - - for (register unsigned i = 0; i < 3; ++i) - ret[i] *= resip; - - return ret; + return PointType3D<T> ( + vec.x * m0 + vec.y * m4 + vec.z * m8 + m12, + vec.x * m1 + vec.y * m5 + vec.z * m9 + m13, + vec.x * m2 + vec.y * m6 + vec.z * m10 + m14 + ); } /** Direct access to the matrix elements, just remember they are in column major format!!