So we just take the area in the scene to be equivalent to one pixel?

I think the decision of how big of an area should be one pixel is arbitrary