>but they are complex valued and change when you sample from them

I see, that does sound pretty different