🔒 Locked

View Transformation + Projection Transformation calculation method

Started by lucky6969b Aug 7, 2012 at 5:49 AM 1 replies 1.2k views

Original Post

Aug 7, 2012 at 5:49 AM


 virtual void Render()
 {
 CObjects::Render();
 g_pEffect->SetTechnique(g_pEffect->GetTechniqueByName("basic_with_shader"));
 D3DXMATRIXA16 compMat, invMat;
 g_pEffect->SetMatrixTranspose("matView", &g_Cam[iCurCam].GetViewMatrix());
 D3DXMatrixMultiply(&compMat, &m_matWorld, &g_Cam[iCurCam].GetViewMatrix());
 g_pEffect->SetMatrixTranspose("matWorldView", &compMat);
 D3DXMatrixInverse(&invMat, NULL, &compMat);
 D3DXMatrixTranspose(&invMat, &invMat);
 g_pEffect->SetMatrixTranspose("matWorldViewIT", &invMat);
 D3DXMatrixInverse(&invMat, NULL, &g_Cam[iCurCam].GetViewMatrix());
 D3DXMatrixTranspose(&invMat, &invMat);
 g_pEffect->SetMatrixTranspose("matViewIT", &invMat);
 D3DXMatrixMultiply(&compMat, &compMat, &g_Cam[iCurCam].GetProjMatrix());
 g_pEffect->SetMatrixTranspose("matWorldViewProj", &compMat);
 g_pEffect->SetMatrixTranspose("matProj", &g_Cam[iCurCam].GetProjMatrix());
 m_Mesh->UpdateFrame(NULL, &m_matWorld);
 m_Mesh->RenderTransparent();
 }


 vNormal = normalize(vNormal);
 Out.Pos = mul(matWorldViewProj, vPosition);
 float3 P = mul(matWorldView, vPosition); //position in view space
 float3 N = mul((float3x3)matWorldViewIT, vNormal); //normal in view space
 float3 V = -normalize(P); //viewer

*********** I strongly believe there is a problem with matViewProj Transform combining with world transform to get the positions.
What is the correct way of calculating the world transform but taking care of view and projection?

In Summary, I don't know how to define the view transformation.....

Thanks
Jack

lucky6969b

Aug 7, 2012 at 6:28 AM

http://msdn.microsoft.com/en-us/library/windows/desktop/ee418867%28v=vs.85%29.aspx

I think the AMD shader is combining the view and projection transform into one, and directly transforming world to clip space.

Adaline

Aug 7, 2012 at 10:20 AM

Hello

D3DXMATRIXA16 compMat, invMat; g_pEffect->SetMatrixTranspose("matView", &g_Cam[iCurCam].GetViewMatrix()); D3DXMatrixMultiply(&compMat, &m_matWorld, &g_Cam[iCurCam].GetViewMatrix()); g_pEffect->SetMatrixTranspose("matWorldView", &compMat); D3DXMatrixInverse(&invMat, NULL, &compMat); D3DXMatrixTranspose(&invMat, &invMat); g_pEffect->SetMatrixTranspose("matWorldViewIT", &invMat); D3DXMatrixInverse(&invMat, NULL, &g_Cam[iCurCam].GetViewMatrix()); D3DXMatrixTranspose(&invMat, &invMat); g_pEffect->SetMatrixTranspose("matViewIT", &invMat);

VS
float3 N = mul((float3x3)matWorldViewIT, vNormal); //normal in view space

I'm not sure about how matWorldViewIT is defined and used :
From the main code one can see : matWorldViewIT=(World*View)^-1 (I don't take into account transpositions)
This matrix transforms something in view space into object (local) space but the normal (at least at the end of the rendering pipeline) is expressed in world space.
It should be something like that :


 float3 N = mul((float3x3)matWorld, vNormal); //normal in world space

If you really want to express your normal in view space (this is totally possible, for example when one wants to pack normals as 2-components vector in a G Buffer) :


 float3 N = mul((float3x3)matWorldView, vNormal); //normal in view space

Another little thing :


 D3DXMatrixTranspose(&invMat, &invMat);
 g_pEffect->SetMatrixTranspose("matWorldViewIT", &invMat);

is similar to :


 g_pEffect->SetMatrix("matWorldViewIT", &invMat);

Hoping that might help a little

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.