Converting a ray into object space

Converting a ray into object space

• carl64

  Thanks, I did try that before, but must of not implemented it correctly.

  After looking at the pick example I am still unsure of where it is that the ray is comverted into object space, as I can only see it converted to the inverse of the view matrix:

  // Compute the vector of the pick ray in screen space
  D3DXVECTOR3 v;
  v.x = ( ( ( 2.0f * ptCursor.x ) / m_d3dsdBackBuffer.Width ) - 1 ) / matProj._11;
  v.y = -( ( ( 2.0f * ptCursor.y ) / m_d3dsdBackBuffer.Height ) - 1 ) / matProj._22;
  v.z = 1.0f;

  // Get the inverse view matrix
  D3DXMATRIXA16 matView, m;
  m_pd3dDevice->GetTransform( D3DTS_VIEW, &matView );
  D3DXMatrixInverse( &m, NULL, &matView );

  // Transform the screen space pick ray into 3D space
  vPickRayDir.x = v.x*m._11 + v.y*m._21 + v.z*m._31;
  vPickRayDir.y = v.x*m._12 + v.y*m._22 + v.z*m._32;
  vPickRayDir.z = v.x*m._13 + v.y*m._23 + v.z*m._33;
  vPickRayOrig.x = m._41;
  vPickRayOrig.y = m._42;
  vPickRayOrig.z = m._43;

  This is all the code I can see that is used for transforming the ray, is this what converts it to object space, as I would of thought that it would need to use the objects mesh, the ray direction and ray origin then are passed to the D3DXIntersect function.

  Note: I am currently using an old version of the directx 9 SDK ( around 2002, bad internet connection doesnt really allow for an update), that could be a possible problem.

  Thanks again for any help given.

  Ryan.

  
  

• chsk

  Thanks for your reply, so your saying that I have to transform the ray and then transform it again by the inverse of the meshes matrix. The problem I have with that is that what do I transform the ray with in the first transformation

  Do I have to transform it just like in the pick example with the inverse view, setting the position as the inverse of the view matrix, and multiplying the direction with the inverse view

  Or do I have to do the same calculations but instead of with the view matrix, but with the (inverse) meshes matrix (where both the rotations and world transformations are stored)

  Or do I have to use a different Matrix Or the meshes rotation matrix

  Or am I just missing something all together

  To clarify my main question, I would like to know what transformations would have to be performed on the ray before transforming it with the meshes world matrix.

  Thanks for any replies.

  Ryan.

  
  

• data_art

  Your mouse point x,y is in windows screen coordinates 0<x<width, 0<y<height

  Convert this to screen space -1<x'<1, -1<y'<1using x' = (x/(width/2))-1 (I think you have to flip Y as this point since 0,0 is bottom left in screen space)

  Then construct a screen space ray (x', y', 0) - (x',y',1) which is a 3d ray going into the screen at that point.

  Since

  ObjectSpace * WorldTransform = WorldSpace
  WorldSpace * ViewTransform = CameraSpace
  CameraSpace * ProjectionTransform = ScreenSpace

  Then you convert back by dividing byt the correct transforms, or multiplying by the inverse.

  So if you want object space coordiantes to do a mesh.intersect then

  objectray = screenray * projectionInverse * ViewInverse * worldInverse

  
  
  

• merwinp

  The example from a book I'm going through at the moment is kind of like this:

  
  // Get the project, view, and inversed view matrices
  m_Device->GetTransform(D3DTS_PROJECTION, &matProj);
  m_Device->GetTransform(D3DTS_VIEW, &matView);
  D3DXMatrixInverse(&matInv, NULL, &matView);

  // Compute the vector of the pick ray in screen space
  m_Device->GetViewport(&vp);
  vecTemp.x = (((2.0f * XPos) / vp.Width) - 1.0f) / matProj._11;
  vecTemp.y = -(((2.0f * YPos) / vp.Height) - 1.0f) / matProj._22;
  // vecTemp.y = -(((2.0f * YPos) / vp.Height) + 1.0f) / matProj._22;
  vecTemp.z = 1.0f;

  // Transform the screen space ray
  vecRay.x = matInv._41;
  vecRay.y = matInv._42;
  vecRay.z = matInv._43;
  vecDir.x = vecTemp.x * matInv._11 +
  vecTemp.y * matInv._21 +
  vecTemp.z * matInv._31;
  vecDir.y = vecTemp.x * matInv._12 +
  vecTemp.y * matInv._22 +
  vecTemp.z * matInv._32;
  vecDir.z = vecTemp.x * matInv._13 +
  vecTemp.y * matInv._23 +
  vecTemp.z * matInv._33;

  // Transform ray and direction by object's world transformation matrix
  D3DXMatrixInverse(&matInv, NULL, m_Mesh->GetMatrix());
  D3DXVec3TransformCoord(&vecMeshRay, &vecRay, &matInv);
  D3DXVec3TransformNormal(&vecMeshDir, &vecDir, &matInv);
  // D3DXVec3Normalize(&vecMeshDir, &vecMeshDir);

  // Check for intersection
  D3DXIntersect(m_Mesh->GetMesh(), &vecMeshRay, &vecMeshDir, &Hit, &FaceIndex, &u, &v, &Dist, NULL, NULL);

  
  

• PeterZ

  Thanks for your replies, I have been able to come up with some code that actually does work to some degree (it does detect the collision all the time, just some areas near the mesh, the collision is a bit off), but has a few errors within it, I will get back to you if I have any difficulties with finishing it off.

  Thanks again.

  Ryan

  
  

• IanTaite

  Compare your code to the 'pick' example in the SDK, that does exactly what you are tryng to do.
  
  

• gregaug

  From what I gather, after you've transformed the ray you then need to transform the ray and its direction by the inverse of the mesh's world transformation matrix:

  // Transform ray and direction by object's world transformation matrix
  D3DXMatrixInverse(&matInv, NULL, matMeshWorld);
  D3DXVec3TransformCoord(&vecMeshRay, &vecRay, &matInv);
  D3DXVec3TransformNormal(&vecMeshDir, &vecDir, &matInv);

  // Check for intersection
  D3DXIntersect(m_Mesh, &vecMeshRay, &vecMeshDir, &Hit, &FaceIndex, &u, &v, &Dist, NULL, NULL);

  
  
• Converting a ray into object space