Hi. I've been wanting to implement SSAO without having to store the vertex normal in addition to the normal from the normal map, a brief look on the internet came up with being able to calculate this using view space position and the ddx/ddy functions. I've tried this but the result looks incorrect as below:

[attachment=5440:onormals.jpg]

I store linear view-space depth and reconstruct the position calculating the view-ray and using the stored depth from the g-buffer. Could anyone point me in the right direction to get this working?