I have noticed that there are two possible ways to convert 2D screen coordinates into 3D world coordinates. The problem is that each approach gives me a different result for my ray origin and direction. Here are the two aprroaches as implemented in my code.
Function 1:
void ScreenToWorld1( iX, int iY )
{
D3DVIEWPORT9 viewport;
D3DXMATRIXA16 matProj, matView, matWorld;
D3DXVECTOR3 vRayOrigin, vRayDir;
D3DXVECTOR3 vMouseNear, vMouseFar;
vMouseNear = D3DXVECTOR3((float)iX, (float)iY, 0.0f);
vMouseFar = D3DXVECTOR3((float)iX, (float)iY, 1.0f);
m_pD3DDevice->GetViewport(&viewport);
m_pD3DDevice->GetTransform( D3DTS_PROJECTION, &matProj );
m_pD3DDevice->GetTransform( D3DTS_VIEW, &matView );
D3DXMatrixIdentity(&matWorld);
D3DXVec3Unproject( &vRayOrigin, &vMouseNear, &viewport, &matProj, &matView, &matWorld );
D3DXVec3Unproject( &vRayDir, &vMouseFar, &viewport, &matProj, &matView, &matWorld );
}
Function2:
void ScreenToWorld2( int iX, int iY )
{
D3DVIEWPORT9 view;
D3DXMATRIXA16 matProj, matView, matWorld, m;
D3DXVECTOR3 vRayOrigin, vRayDir;
m_pD3DDevice->GetViewport(&view);
m_pD3DDevice->GetTransform(D3DTS_PROJECTION, &matProj);
// Compute the vector of the pick ray in screen space
D3DXVECTOR3 v;
v.x = ( ( ( 2.0f * iX ) / view.Width ) - 1 ) / matProj._11;
v.y = -( ( ( 2.0f * iY ) / view.Height ) - 1 ) / matProj._22;
v.z = 1.0f;
D3DXMatrixIdentity(&matWorld);
m_pD3DDevice->GetTransform(D3DTS_VIEW, &matView);
m = matWorld * matView;
D3DXMatrixInverse( &m, NULL, &m );
// Transform the screen space pick ray into 3D space
vRayDir.x = v.x*m._11 + v.y*m._21 + v.z*m._31;
vRayDir.y = v.x*m._12 + v.y*m._22 + v.z*m._32;
vRayDir.z = v.x*m._13 + v.y*m._23 + v.z*m._33;
vRayOrigin.x = m._41;
vRayOrigin.y = m._42;
vRayOrigin.z = m._43;
}
What is really confusing is that I need to use Function 1 when checking for ray-3D plane intersection and Function 2 for ray-mesh intersection.
If some could please clear this up for me I would be very grateful.
Thanks,
Lukas