Pixel fog can be applied using any fog formula identified by members of the D3DFOGMODE enumerated type. Pixel-fog formula implementations are driver-specific, and if a driver doesn't support a complex fog formula, it should degrade to a less complex formula.

To alleviate fog-related graphic artifacts caused by uneven distribution of z-values in a depth buffer, most hardware devices use eye-relative depth instead of z-based depth values for pixel fog. Eye-relative depth is essentially the w element from a homogeneous coordinate set. Microsoft® Direct3D® takes the reciprocal of the RHW element from a device space coordinate set to reproduce true w. If a device supports eye-relative fog, it sets the D3DPRASTERCAPS_WFOG flag in the RasterCaps member of the D3DCAPS8 structure when you call the IDirect3DDevice8::GetDeviceCaps method.

Note  With the exception of the reference rasterizer, software devices always use z to calculate pixel fog effects.

When eye-relative fog is supported, the system automatically uses eye-relative depth rather than z-based depth if the provided projection matrix produces z-values in world space that are equivalent to w-values in device space. You set the projection matrix by calling the IDirect3DDevice8::SetTransform method, using the D3DTS_PROJECTION value and passing a D3DMATRIX structure that represents the desired matrix. If the projection matrix isn't compliant with this requirement, fog effects are not applied properly. For details about producing a compliant matrix, see A W-Friendly Projection Matrix. The perspective projection matrix provided in What Is the Projection Transformation? produces a compliant projection matrix.

1. Enable fog blending by setting the D3DRS_FOGENABLE render state to TRUE.
2. Set the desired fog color in the D3DRS_FOGCOLOR render state.
3. Choose the fog formula to use by setting the D3DRS_FOGTABLEMODE render state to the corresponding member of the D3DFOGMODE enumerated type.
4. Set the fog parameters as desired for the selected fog mode in the associated render states. This includes the start and end distances for linear fog, and fog density for exponential fog mode.

The following example shows what these steps might look like in code.

// For brevity, error values in this example are not checked 
// after each call. A real-world application should check 
// these values appropriately.
//
// For the purposes of this example, g_pDevice is a valid
// pointer to an IDirect3DDevice8 interface.
void SetupPixelFog(DWORD Color, DWORD Mode)
{
    float Start = 0.5f,    // For linear mode
          End   = 0.8f,
          Density = 0.66;  // For exponential modes
 
    // Enable fog blending.
    g_pDevice->SetRenderState(D3DRS_FOGENABLE, TRUE);
 
    // Set the fog color.
    g_pDevice->SetRenderState(D3DRS_FOGCOLOR, Color);
    
    // Set fog parameters.
    if(D3DFOG_LINEAR == Mode)
    {
        g_pDevice->SetRenderState(D3DRS_FOGTABLEMODE, Mode);
        g_pDevice->SetRenderState(D3DRS_FOGSTART, *(DWORD *)(&Start));
        g_pDevice->SetRenderState(D3DRS_FOGEND,   *(DWORD *)(&End));
    }
    else
    {
        g_pDevice->SetRenderState(D3DRS_FOGTABLEMODE, Mode);
        g_pDevice->SetRenderState(D3DRS_FOGDENSITY, *(DWORD *)(&Density));
    }
 
}

Note  Some fog parameters are required as floating-point values, even though the IDirect3DDevice8::SetRenderState method only accepts DWORD values in the second parameter. The preceding example provides the floating-point values to SetRenderState without data translation by casting the addresses of the floating-point variables as DWORD pointers, and then dereferencing them.