esides direction, it is interesting to consider other aspects of the "eye flow as water flow" analogy. The stream of perception generated by a contrast relationship is, like contrast itself, defined by both the magnitude of difference between the contrasting objects and by the number of aspects in which they are opposed (let's call the latter parameter the dimensionality of contrast).

For example, if two squares are visually interrelated and linked by a common color and other attributes, then the velocity of the visual flow between them will depend on the difference in their sizes. If the squares are equal (symmetry) or almost equal in sizes, then the rate of perception for this couple of elements is low, and the above discussed directionality is not obvious - the viewer's eyes will move between the linked objects relatively easily in both directions. If, however, the squares are of more contrasting sizes, then the "waterfall" between their levels becomes more noticeable.

So far any motions we discussed lay in the plane of the page. However, the 3D aspect of the graphics you use also has some dynamic implications. If one of two objects is made to look "closer" or "higher" above the page (e.g. by adding a drop shadow behind it), then the eye flows that lead to that object get an additional impulse.

If we continue the analogy by visualizing the "flow" as having certain breadth in addition to velocity, then it is logical to connect this parameter to the dimensionality of the contrast link. Thus, if we make the couple of squares different not only by their size but by their color as well, then the imaginary stream between them becomes wider and, somehow, calmer. By adding more dimensions to the contrast, we're making the flow still wider and deeper, until some moment when the objects lose any visual connection and the flow between them dilutes into nil.

It is also not unreasonable to visualize eye flows as having certain inertia. If the end point of the flow attracts the eye with considerable force, we may easily run out beyond it, thus involving other objects and the surrounding space into the relationships generated by the contrasting pair. If an element belongs to more than one contrast link (which is most often the case with real world compositions), eye inertia will turn the path that we travel from a sequence of linear fragments into a smooth curve.

Now it's time to take a general look at the entire page to see if separate eye flows will combine into any discernible pattern. Such a "perceptual map" of the page, with eye flows visualized by lines, does not represent anyone's actual eye movements, but only shows the popular pathways that the viewers are most likely to use when wandering across the page. Such an overall picture of eye flows on a page may be an instructive aspect of analysis that you're recommended to perform on everything you create.

Try to think of your composition not as a static whole, but a dynamic "movie" that unveils in the viewer's perception. Even if this movie spans a fraction of a second, it has its own plot and heroes, its own inception and closing. Try to analyze various perception scenarios: scan the page in various directions, try to identify areas where eye flows feel uncomfortable due to running into an obstacle or having to turn at too sharp edges.

Indeed there exist professional compositions that look great to a considerate eye, but do not quite work for transient perception simply because of a jarring arrangement of eye flows. Dynamic analysis does require some getting used to, because for the designer who's working long hours on a page, his own creation may look all but dynamic. But the smooth flow of perception that's achieved by a thorough dynamic analysis is well worth the additional trouble.