6 replies to this topic

[Medo3337]

D3DXMATRIX matRotation;

How can I get Pitch, Yaw and Roll values from matRotation?

Edited by Medo3337, 19 May 2012 - 04:50 PM.

[Endurion]

This routine served me well (the one time I used it), always be careful with these three angles, usually it ends up in tears and gimbal locks:

void ToEulerAngles( float& fAngleX, float& fAngleY, float& fAngleZ )
	{
	  static float  fRadiansToDegree = 180.0f / 3.1415926f;
	  fAngleY	 = asinf( M[8] );		// Calculate Y-axis angle
	  float fCY   =  cosf( fAngleY );
	  fAngleY	 *= fRadiansToDegree;
	  if ( fabsf( fCY ) > 0.005f )	  
	  {
		float ftrx	  =  M[10] / fCY;	// No, so get X-axis angle
		float ftry	  = -M[9]  / fCY;
		fAngleX  = atan2f( ftry, ftrx ) * fRadiansToDegree;
		ftrx	  =  M[0] / fCY;			// Get Z-axis angle
		ftry	  = -M[4] / fCY;
		fAngleZ  = atan2f( ftry, ftrx ) * fRadiansToDegree;
	  }
	  else
	  {
		// Gimbal lock?
		fAngleX  = 0;					  // Set X-axis angle to zero
		float ftrx	  =  M[5];		   // And calculate Z-axis angle
		float ftry	  =  M[1];
		fAngleZ  = atan2f( ftry, ftrx ) * fRadiansToDegree;
	  }
	  if ( fAngleX < 0 )
	  {
		fAngleX += 360;
	  }
	  if ( fAngleY < 0 )
	  {
		fAngleY += 360;
	  }
	  if ( fAngleZ < 0 )
	  {
		fAngleZ += 360;
	  }
	}

The routine is a member of a 4x4 matrix with these members:

    struct tStruct
    {
	  float	    _11, _12, _13, _14;
	  float	    _21, _22, _23, _24;
	  float	    _31, _32, _33, _34;
	  float	    _41, _42, _43, _44;
    };

	union
	{
	  tStruct   ms;
	  float	 m[4][4];
	  float	 M[16];
	};

Edited by Endurion, 19 May 2012 - 10:55 PM.

[Medo3337]

How do you use this method with

matRotation?

The method I want is to be able to pass matRotation which is "D3DXMATRIX" and get the rotation X, Y and Z.

Remember, I want the EXACT values of Pitch, Yaw and Roll from

D3DXMATRIX.

Edited by Medo3337, 20 May 2012 - 03:09 AM.

[Endurion]

A D3DXMATRIX has these members:

typedef struct _D3DMATRIX {
union {
struct {
float _11, _12, _13, _14;
float _21, _22, _23, _24;
float _31, _32, _33, _34;
float _41, _42, _43, _44;

};
float m[4][4];
};
} D3DMATRIX;


You can map all the M[x] accesses to m[x][y], you just need to use the right order of columns/rows.

[Rattenhirn]

Here's what I use:
http://www.euclideanspace.com/maths/geometry/rotations/conversions/matrixToEuler/

[Medo3337]

I see some issues stated about converting matrix to Euler, I want to guarantee that I will receive VALID values and EXACTLY the same rotation values from a D3DXMATRIX every time I convert D3DXMATRIX to Euler without any problem, is that possible?

[Rattenhirn]

I see some issues stated about converting matrix to Euler, I want to guarantee that I will receive VALID values and EXACTLY the same rotation values from a D3DXMATRIX every time I convert D3DXMATRIX to Euler without any problem, is that possible?

Barring rounding errors, you'll get the exact same rotation in euler angles. However, euler angles do not represent orientations uniquely, meaning there's more than one set of euler angles for any given orientation.

Of course, as long you'll use the same algorithm for conversion, matrix M will always give the same angles X, Y and Z. But, for instance, if your rotate M by 45 degrees around the Y axis, you won't necessarily get X, Y+45 and Z. Just another set of angles that represent that orientation.

So, if you want to keep the exact euler values, e.g. as entered, you have to save them...