Hi!
I tried posting this problem on Gamedev earlier, though it didn't get a lot of attention, so I hope it's not bad karma to post it here as well. The more readers the merrier :)

So I'm trying to reconstruct the world space postion from a linear depth buffer (depths between [0,1] in view space) in my deferred renderer for a bounding volume (namely a light sphere). I know this topic has been covered before in a few threads, though these have focused on the problem when using fullscreen quads, and not bounding volumes. Also, MJP has posted a nice technique in his blog here, but I don't really understand it and can't get it working. MJP proposes this method to get the view space position from the depth buffer:
 
float3 VSPositionFromDepth(float2 vTexCoord, float3 vPositionVS) {    
// Calculate the frustum ray using the view-space position.   
// farCip is the distance to the camera's far clipping plane.    
// Negating the Z component only necessary for right-handed coordinates    
float3 vFrustumRayVS = vPositionVS.xyz * (farClip/-vPositionVS.z);    
return tex2D(depthSampler, vTexCoord).x * vFrustumRayVS;
}

So I create my texcoords, which after a litte cpu-testing/debugging seems to work just fine, like this: 
input.ssPos.xy /= input.ssPos.w;
//Setup texture coordinates, transforming from [-1,1]->[1,-1] to [0,1]->[1,0]
float2 tex = (0.5f * (float2(input.ssPos.x, -input.ssPos.y) + 1)) - halfPixel;  

where ssPos is output from the vertex shader computed normally as: 
output.ssPos = mul(input.pos, WorldViewProj);   

Okay...so after having run: 

//View space position
float3 wsPos = VSPositionFromDepth(tex, input.vsPos); 
//Transform to world space
float3 wsPos = mul(wsPos, InvertView);   

the shader still produces incorrect results. I don't see why wsPos is incorrect, which is mysterious to me. Any thoughts on how to approach the problem?