Does computational refocusing involve focusing after an image has already been taken? If so how does it differ from simply focusing while taking the image? Can it be modeled so that these two are the same, which would be extremely helpful.