How is the Visibility percent analyzed?

I don’t quite understand how Visibility percent is being analyzed. My understanding is that what this component can do is to see how much of the grit-segmented target object is visible from some arbitrary observation point, is this correct?
Even if I connect the components correctly, the determination is that I can see places that clearly should not be visible. I would like to hear a detailed explanation of when each grid of objects is considered visible and when it is considered not visible. Also, I believe there is a range of visibility from 0% to 100%, but when I look at the analysis results, there is only a red grid (0%) and a blue grid (100%). What does this mean?