Question

I have multiple objects moving about in a 3D space and am looking for ways to, on button press, have the camera snap and follow the object chosen.

Is there a way to make use of each object’s worldMatrix? (Below is an example of one object. This one is a planet that spins & orbits)

//set up matrices for rendering
D3DXMATRIX worldMatrixMer, viewMatrixMer, projectionMatrixMer;
m_Camera->GetViewMatrix(viewMatrixMer);
m_D3D->GetWorldMatrix(worldMatrixMer);
m_D3D->GetProjectionMatrix(projectionMatrixMer);
D3DXMatrixRotationX( &matRotateX, rx/65.0f);
//Rotate about Y axis
D3DXMatrixRotationY( &matRotateY, rotation * 15.0f);
D3DXMatrixRotationZ( &matRotateZ, rz/65.0f); 
//Collate Rot Matrices
D3DXMATRIX rotMatrixMer = matRotateX * matRotateY * matRotateZ;
D3DXVECTOR3 newVecDirMer;
D3DXVec3TransformCoord(&newVecDirMer, &initVecDirMer, &rotMatrixMer);
D3DXVec3Normalize( &currentVecDirMer, &newVecDirMer );
//Create the size of the object
D3DXMATRIX matScaleMer;
D3DXMatrixScaling(&matScaleMer, 0.1f, 0.1f, 0.1f);
//Starting position of object
D3DXMatrixTranslation(&matTranslateMer, 0.0f, 0.0f, 3.5983f * 3);
//Rotate about it's own axis
D3DXMatrixRotationY(&worldMatrixMer, rotation);
worldMatrixMer *= rotMatrixMer * matScaleMer * matTranslateMer;
//'Orbit'
D3DXMatrixRotationY( &matOrbit, (-1000.0f * rotation) / 88);
worldMatrixMer *= matOrbit;

I’m really trying to find an elegant way of making this happen, so any suggestions would be greatly appreciated.

Thanks

Answer 1

Is there a way to make use of each object’s worldMatrix?

Object matrix can be represented this way:

objx.x     objx.y     objx.z  0 //m[0][0]..m[0][3] or _11, _12, _13, _14
objy.x     objy.y     objy.z  0 //m[1][0]..m[1][3] or _21, _22, _23, _24
objz.x     objz.y     objz.z  0 //m[2][0]..m[2][3] or _31, _32, _33, _34
objpos.x   objpos.y   objpos.z  1 //m[3][0]..m[3][3] or _41, _42, _43, _44

Where m[][] and _11.._44 are corresponding elements of D3DMATRIX, objpos – object position vector, objx – object x (‘local x” transformed to world space) vector, etc.

So as long as the last column (m[0..3][3]) is 0, 0, 0, 1 you can extract object position and its “x”, “y”, “z” vectors (“side”, “up”, “front” – which is which depends on application) from matrix. If last column is not “0, 0, 0, 1”, then it is projection matrix and you can’t extract object data from it this easily.

From there you could do anything you want with your camera.