Consider two objects of identical size, S, one at the front of the scene (and, say, closer to the eye than any other object in the scene), and the other is at the back of the scene (and, say, farther from the eye than any other object in the scene). Suppose (i) the farther object is at a distance, f, from the eye and appears to be of size Sf after projection on the display; and (ii) the closer object is at a distance, c, from the eye and appears to be of size Sc after projection on the display. One measure of the amount of perspective distortion is the ratio: Sc/Sf, which will always be greater than or equal to 1. A ratio of 1 means no perspective distortion at all.
The sizes are inversely proportional to the distances, hence we have:
Define the depth of the scene, Δ, as Δ = f - c, then:
If Δ << c, there will be very little perspective distortion, and the scene will look more like an orthogonal projection than a perspective one. If Δ >> c, then there will be very exaggerated distortion (imagine looking at a tree in the distance with a pencil right in front of your eye), and the scene could appear cartoonish.
Generally we want the relationship between Δ and c to be somewhere between these two extremes. Since Δ is out of our control (it is determined solely by how the scene is constructed), our only control over the amount of perspective distortion is c: the distance we place the eye from the front of the scene.
To summarize, note the following: