z-coordinate to Depth Buffer

It depends on your projection matrix! You should look at how your projection matrix is formed, and then follow through the math of the matrix multiplication to get the formula for computing the z and w components after projection. The depth buffer value will then be z / w.

Here is my really old routine from my really old library that I've been too lazy/afraid to replace with something optimized. This generates a projection matrix that you can multiply in at the end of your matrices. I'm sure whatever lib you're using has something similar and better but just so you can get the idea.

template<class T> void TDL3D4x4Matrix<T>::Set(T fFOVAngleY, T fAspectRatio, T fNearZ, T fFarZ)
{
   T fFOVHalfY = fFOVAngleY * (T) 0.5;
   T fSinFov = (T) sin(fFOVHalfY);
   T fCosFov = (T) cos(fFOVHalfY);

   T fHeight = fCosFov / fSinFov;
   T fWidth = fHeight / fAspectRatio;
   T fRange = fFarZ / (fFarZ-fNearZ);


   m_a.D[0][0] = fWidth;
   m_a.D[0][1] = 0.0f;
   m_a.D[0][2] = 0.0f;
   m_a.D[0][3] = 0.0f;

   m_a.D[1][0] = 0.0f;
   m_a.D[1][1] = fHeight;
   m_a.D[1][2] = 0.0f;
   m_a.D[1][3] = 0.0f;

   m_a.D[2][0] = 0.0f;
   m_a.D[2][1] = 0.0f;
   m_a.D[2][2] = fRange;
   m_a.D[2][3] = 1.0f;

   m_a.D[3][0] = 0.0f;
   m_a.D[3][1] = 0.0f;
   m_a.D[3][2] = -fRange * fNearZ;
   m_a.D[3][3] = 0.0f;
   m_pType = &DL3D_PROJECTION;
}

Thanks for your answers!

Then,

If have a vector v and the matrix WVP (WorldViewProjection)

I'm using Left Hand System so v1 = v * WVP;

And the depth buffer value will be v1.z / v1.w

Is it correct??

Yes, that's correct.

Thank you!