Glance again at pages 40 and 42, where the cube was viewed by "looking down" and "looking up," and note that the vertical lines do not actually remain vertical in the picture but instead appear to converge downwards and upwards respectively.
Many books state arbitrarily that such lines should always appear vertical. Although contrary to the "truth" of seeing, this rule is laid down in order to simplify matters. But such simplification is helpful only in mechanical (T-square and triangle) perspective where converging verticals means complicated drafting to establish and work with distant vanishing points, and complicated procedures to determine vertical measurements.
Therefore, when working freehand (without drafting considerations) let the visual truth dictate. If you have difficulty accepting this "truth," the following will help.
Try this again, from both viewpoints, with the book held almost on a level with the central visual ray. The principle is now more dramatically demonstrated because convergence and foreshortening are almost at a maximum.
Take a book and hold it horizontally in this manner (left).
What you see (right) are lines converging to a central vanishing point at eye level. Being standard perspective drawing this is readily accepted.
Now hold the book vertically, above your head, in this manner (left), and view it at approximately the same angle.
What you see (right) is exactly the same as before, only now the convergence is upward instead of horizontal.
Therefore, the convergence and hence the picture is identical from both viewpoints. THE REASON IS SIMPLY THAT THE RELATIONSHIP BETWEEN EYES (SIGHT LINES) AND SUBJECT (BOOK) IS IDENTICAL IN BOTH CASES. (Note angle .)
 Things Seen By Looking Straight Out And Things Seen by Looking Up
But again why is upward and downward convergence so rarely used? The reason is that we usually see things by looking more or less horizontally. Not only is this attitude more natural to the anatomical structure of our neck and head, but so much of what we see exists at or near eye level.
Therefore, most of the time our central visual ray is horizontal, and consequently our imaginary picture plane is vertical (i.e., at right angles to the ground). And under these conditions, vertical elements continue to appear vertical.
When then would upward or downward convergence be appropriate? For one thing, it could be used when drama or interest was desired. But it probably makes most sense when related to the nature of the subject matter. IN OTHER WORDS, THINGS USUALLY SEEN FROM BELOW OR FROM ABOVE SHOULD BE DRAWN WITH CONVERGING VERTICALS.
Examples of things typically seen by looking up, i.e., objects usually above eye level. (Note upward convergence of vertical lines.)
Things Seen By Looking Down 
Examples of things typically seen by looking down, ue., objects usually below eye level. (Note downward conver gence of vertical lines.)
 Review: Looking Up, Straight Out, Down
So when we look up or down at an individual element, such as a single cube, each viewing angle results in a different convergence of the vertical lines. At right are the resulting pictures for each viewing angle shown at left.
This means that all the verticals would still appear vertical. (Note also that the vanishing points must be further apart than in the previous views because the observer is further away.)
Since looking straight out is so very natural and common, this viewpoint is probably the most frequently used in perspective drawing.
But — and this is very important — if we were to back away and view all the cubes simultaneously (i.e., all within one cone of vision) then the central visual ray would be approximately horizontal and our face and the picture plane approximately vertical.
Looking Straight Out
Was this article helpful?