Graphics Reference
In-Depth Information
or even personality (“a lighthearted style”). Unfortunately, all of these things are
loosely related. It's hard to imagine making an image that used, say, mosaic tiles,
conveying a film-noir mood, but with a lighthearted style. A clear definition and
characterization of style at all these levels remains elusive, but for problems like
transferring style from one rendering to another, it's an essential ingredient.
At a more operational level, much expressive rendering work has concentrated
on renderings of single objects, or scenes in which objects have similar sizes.
Abstraction tends to operate on a scale of no more than an order of magnitude.
Few expressive rendering systems have a broad enough range of application to
be able to make an effective rendering of, say, Dorothy, from The Wizard of Oz,
on the yellow brick road, surrounded by hilly fields, with the Emerald City in the
distant background drawn with a few indicative strokes. Thus, scale remains an
important challenge in expressive rendering.
Coherence is a general term for the relatedness of nearby items, whether
in a single image ( spatial coherence ) or in a sequence of images ( temporal
coherence ). Spatial coherence in expressive rendering arises in multiple contexts.
For instance, if we decide to render object outlines using a wiggly line, we need
to displace adjacent points of the outline by about the same amount, as in Fig-
ure 34.5. (If we displaced them by random amounts, the result would not be a
line!) But as you can see in the figure, if we simply start making a wiggle at some
point of an outline, when we return to that point the displacements may not match,
and the failure to match manages to particularly attract the viewer's attention.
Temporal coherence is closely related. When we animate an expressive ren-
dering, the strokes or other marks (e.g., tiles in a mosaic) vary over time. If the
marks change rapidly from one frame to the next, the eye can be easily distracted.
Proof of this can be seen by watching static on a broadcast (rather than cable)
television. The average “frame” is a neutral gray, but as you watch, your eyes
will detect patterns, notice things crawling or running across the screen, etc. If the
marks in a rendering are something like stippling (a pattern of dots used to convey
darkness or lightness) or a texture composed of short strokes, then even if the stip-
ples or strokes have some temporal coherence (i.e., each stipple changes position
slowly over time, or else disappears or appears, or each short stroke's endpoints
move slowly over time), their motion can become a stronger perceptual cue than
the marks themselves. For longer strokes, like long, thin pen-and-ink lines, the
motion percept tends to be aligned perpendicular to the stroke; if the stroke cor-
responds to a contour, then such motion is consistent with contour motion, while
motion along the stroke, as might appear when a small stroke that's part of a large
contour shrinks before disappearing, is inconsistent with contour motion.
Figure 34.5: Mostly spatially
coherent wiggles on the edges of
a torus.
34.3 Marks and Strokes
Much expressive rendering is done with primitives that can be called marks or
strokes. In stippling, for instance, each mark is a pen dot; in oil painting, each
motion of the brush across the canvas is a stroke. In pen-and-ink rendering, a
mixture of marks and strokes often serves to create texture, boundaries, etc. Not
every form of expressive rendering uses marks and strokes (see Section 34.7),
but many do. Why? First, many expressive rendering approaches mimic artistic
techniques, and the use of strokes probably originated when some primitive human
first picked up a stick and drew a shape in the dirt. So the simplest reason for marks
and strokes is the ease with which we can create them. More important, though,
is their power at triggering a response in the visual system. When we draw a
 
 
Search WWH ::




Custom Search