m3math.h

Go to the documentation of this file.

#ifndef LL_M3MATH_H
#define LL_M3MATH_H

#include "llerror.h"
#include "stdtypes.h"

class LLVector4;
class LLVector3;
class LLVector3d;
class LLQuaternion;

// NOTA BENE: Currently assuming a right-handed, z-up universe

//                           ji 
// LLMatrix3 = | 00 01 02 |
//                         | 10 11 12 |
//                         | 20 21 22 |

// LLMatrix3 = | fx fy fz |     forward-axis
//                         | lx ly lz | left-axis
//                         | ux uy uz | up-axis

// NOTE: The world of computer graphics uses column-vectors and matricies that 
// "operate to the left". 


static const U32 NUM_VALUES_IN_MAT3     = 3;
class LLMatrix3
{
        public:
                F32     mMatrix[NUM_VALUES_IN_MAT3][NUM_VALUES_IN_MAT3];

                LLMatrix3(void);                                                        // Initializes Matrix to identity matrix
                explicit LLMatrix3(const F32 *mat);                                     // Initializes Matrix to values in mat
                explicit LLMatrix3(const LLQuaternion &q);                      // Initializes Matrix with rotation q

                LLMatrix3(const F32 angle, const F32 x, const F32 y, const F32 z);      // Initializes Matrix with axis angle
                LLMatrix3(const F32 angle, const LLVector3 &vec);       // Initializes Matrix with axis angle
                LLMatrix3(const F32 angle, const LLVector3d &vec);      // Initializes Matrix with axis angle
                LLMatrix3(const F32 angle, const LLVector4 &vec);       // Initializes Matrix with axis angle
                LLMatrix3(const F32 roll, const F32 pitch, const F32 yaw);      // Initializes Matrix with Euler angles

                //
                // Matrix initializers - these replace any existing values in the matrix
                //

                // various useful matrix functions
                const LLMatrix3& setIdentity();                         // Load identity matrix
                const LLMatrix3& clear();                                       // Clears Matrix to zero
                const LLMatrix3& setZero();                                     // Clears Matrix to zero

                //
                // Matrix setters - set some properties without modifying others
                //

                // These functions take Rotation arguments
                const LLMatrix3& setRot(const F32 angle, const F32 x, const F32 y, const F32 z);        // Calculate rotation matrix for rotating angle radians about (x, y, z)
                const LLMatrix3& setRot(const F32 angle, const LLVector3 &vec); // Calculate rotation matrix for rotating angle radians about vec
                const LLMatrix3& setRot(const F32 roll, const F32 pitch, const F32 yaw);        // Calculate rotation matrix from Euler angles
                const LLMatrix3& setRot(const LLQuaternion &q);                 // Transform matrix by Euler angles and translating by pos

                const LLMatrix3& setRows(const LLVector3 &x_axis, const LLVector3 &y_axis, const LLVector3 &z_axis);
                const LLMatrix3& setRow( U32 rowIndex, const LLVector3& row );
                const LLMatrix3& setCol( U32 colIndex, const LLVector3& col );

                
                //
                // Get properties of a matrix
                //
                LLQuaternion quaternion() const;                // Returns quaternion from mat
                void getEulerAngles(F32 *roll, F32 *pitch, F32 *yaw) const;     // Returns Euler angles, in radians

                // Axis extraction routines
                LLVector3 getFwdRow() const;
                LLVector3 getLeftRow() const;
                LLVector3 getUpRow() const;
                F32      determinant() const;                   // Return determinant


                //
                // Operations on an existing matrix
                //
                const LLMatrix3& transpose();           // Transpose MAT4
                const LLMatrix3& orthogonalize();       // Orthogonalizes X, then Y, then Z
                void invert();                  // Invert MAT4
                const LLMatrix3& adjointTranspose();// returns transpose of matrix adjoint, for multiplying normals

                
                // Rotate existing matrix  
                // Note: the two lines below are equivalent:
                //      foo.rotate(bar) 
                //      foo = foo * bar
                // That is, foo.rotate(bar) multiplies foo by bar FROM THE RIGHT
                const LLMatrix3& rotate(const F32 angle, const F32 x, const F32 y, const F32 z);        // Rotate matrix by rotating angle radians about (x, y, z)
                const LLMatrix3& rotate(const F32 angle, const LLVector3 &vec);                                         // Rotate matrix by rotating angle radians about vec
                const LLMatrix3& rotate(const F32 roll, const F32 pitch, const F32 yaw);                        // Rotate matrix by roll (about x), pitch (about y), and yaw (about z)
                const LLMatrix3& rotate(const LLQuaternion &q);                 // Transform matrix by Euler angles and translating by pos

                void add(const LLMatrix3& other_matrix);        // add other_matrix to this one

// This operator is misleading as to operation direction
//              friend LLVector3 operator*(const LLMatrix3 &a, const LLVector3 &b);                     // Apply rotation a to vector b

                friend LLVector3 operator*(const LLVector3 &a, const LLMatrix3 &b);                     // Apply rotation b to vector a
                friend LLVector3d operator*(const LLVector3d &a, const LLMatrix3 &b);                   // Apply rotation b to vector a
                friend LLMatrix3 operator*(const LLMatrix3 &a, const LLMatrix3 &b);                     // Return a * b

                friend bool operator==(const LLMatrix3 &a, const LLMatrix3 &b);                         // Return a == b
                friend bool operator!=(const LLMatrix3 &a, const LLMatrix3 &b);                         // Return a != b

                friend const LLMatrix3& operator*=(LLMatrix3 &a, const LLMatrix3 &b);                           // Return a * b
                friend const LLMatrix3& operator*=(LLMatrix3 &a, F32 scalar );                                          // Return a * scalar

                friend std::ostream&     operator<<(std::ostream& s, const LLMatrix3 &a);       // Stream a
};

inline LLMatrix3::LLMatrix3(void)
{
        mMatrix[0][0] = 1.f;
        mMatrix[0][1] = 0.f;
        mMatrix[0][2] = 0.f;

        mMatrix[1][0] = 0.f;
        mMatrix[1][1] = 1.f;
        mMatrix[1][2] = 0.f;

        mMatrix[2][0] = 0.f;
        mMatrix[2][1] = 0.f;
        mMatrix[2][2] = 1.f;
}

inline LLMatrix3::LLMatrix3(const F32 *mat)
{
        mMatrix[0][0] = mat[0];
        mMatrix[0][1] = mat[1];
        mMatrix[0][2] = mat[2];

        mMatrix[1][0] = mat[3];
        mMatrix[1][1] = mat[4];
        mMatrix[1][2] = mat[5];

        mMatrix[2][0] = mat[6];
        mMatrix[2][1] = mat[7];
        mMatrix[2][2] = mat[8];
}


#endif


// Rotation matrix hints...

// Inverse of Rotation Matrices
// ----------------------------
// If R is a rotation matrix that rotate vectors from Frame-A to Frame-B,
// then the transpose of R will rotate vectors from Frame-B to Frame-A.


// Creating Rotation Matricies From Object Axes
// --------------------------------------------
// Suppose you know the three axes of some object in some "absolute-frame".
// If you take those three vectors and throw them into the rows of 
// a rotation matrix what do you get?
//
// R = | X0  X1  X2 |
//     | Y0  Y1  Y2 |
//     | Z0  Z1  Z2 |
//
// Yeah, but what does it mean?
//
// Transpose the matrix and have it operate on a vector...
//
// V * R_transpose = [ V0  V1  V2 ] * | X0  Y0  Z0 | 
//                                    | X1  Y1  Z1 |                       
//                                    | X2  Y2  Z2 |
// 
//                 = [ V*X  V*Y  V*Z ] 
//
//                 = components of V that are parallel to the three object axes
//
//                 = transformation of V into object frame
//
// Since the transformation of a rotation matrix is its inverse, then
// R must rotate vectors from the object-frame into the absolute-frame.

---