#include <revolution/mtx.h>u32 C_MTXInverse ( const Mtx src, Mtx inv );
u32 PSMTXInverse ( const Mtx src, Mtx inv );#define MTXInverse C_MTXInverse // for debug build
#define MTXInverse PSMTXInverse // for nondebug buildsrc |
Matrix to be inverted. src may be either an Mtx or an MtxPtr. |
inv |
Resultant inverse of source matrix. dst may be either an Mtx or an MtxPtr.src can be the same as inv. |
0 if matrix is not invertible. In this case inv is unchanged.
1 on success.
Computes a fast inverse of a 3x4 matrix.
The algorithm used is as follows for a 3x4 matrix M:
M = [ A C ]
(A = upper 3x3 subMatrix; C = 3x1 column vector)
Minv = [ Ainv ( Ainv ) * ( -C ) ]
This function should be used whenever the inverse of a general 3x4 matrix is required. The cost of the function, in opertions, is 48 multiplies, 1 division, 2 additions, 25 subtractions. While MTXInverse will work for any invertible 3x4 matrix, for reasons of efficiency it should not be used when the inverse of a matrix can be computed directly (as in single axis rotation, scale, translation).
During a debug build, MTXInverse is interpreted as C_MTXInverse, which is a normal C language version. During a nondebug build for a Broadway processor, the paired-singles equivalent of this function, PSMTXInverse, will be automatically substituted. You can also call C_MTXInverse or PSMTXInverse explicitly.
The C version of this function becomes more efficient if src is not equal to inv as it saves a copy operation from a temporary matrix.
03/01/2006 Initial version.