When doing full screen post effects, if the size of the texture and the size of the render target match, using linear and point filter for the texture should be the same, however, is this only true when you are sampling the exact center of the texels (like Gaussian Blur or plain averge)? How about when doing something like Poisson Blur where the random sample points are not necessarily on the center of the texel? Does linear filter make a difference in this case, even though the texel/pixel ratio is still 1:1?