Geography Reference
In-Depth Information
variables. The problem in doing this is that the perceived bar widths will not only vary
according to the assigned data values, but will also vary with the observer's viewing angle.
[An example occurs in a study of retail businesses in Toronto (Hernandez, 2007), in which
3D bar symbols representing businesses in Toronto are aligned with the streets on which
they are located.] In extreme circumstances, where bars symbols with minimal thickness are
viewed from the side, they may all but disappear. This problem was identified by the authors
of a study involving the visualization of ocean-bed characteristics by Schmidt
et al
. (2004)
and was resolved by using spherical 3D glyphs whose perceived sizes were independent of
their orientation.
The viewpoint dependency problem not only affects the ability of symbol dimensions to
carry data information, but it may also undermine the use of other symbol variations to
represent data. For example, where symbol shapes are varied to reflect nominal scale data,
then the analyst's viewpoint may make it difficult to perceive these shape differences across
the scene (Lind, Bingham and Forsell, 2003). There are several other reasons why the 3D
equivalents of 2D symbols may not work effectively in 3D scenes. As Kraak (1988, 1989) has
indicated, the surface shadowing applied to 3D point symbols to enhance their realism may
conflict with the visual variations in the lightness and texture applied to their surfaces by
3D rendering. Krisp (2006) illustrates a similar problem with 3D density surfaces, in which
the colours used to encode height are locally modulated by the hill-shading used to enhance
viewing realism. The general conclusion seems to be that, although it is relatively trivial to
convert the standard 2D geometrical symbols used in conventional thematic mapping into
3D equivalents for data visualization, in which a square becomes a cube, a circle becomes
a sphere, a line becomes a ribbon or wall, and a region becomes a prism, variations in the
size and shape of these symbols may not be accurately perceived in a 3D data visualization
because of viewpoint dependencies.
10.3.4 Stereo 3D: pretty useful, or just pretty?
Viewing 3D data visualizations in full stereo is assumed to provide the benefits of binocular
viewing enjoyed by human primates in their natural environments. Although stereo images
possess a certain 'wow' factor, as indicated by the entertainment value of the iMAX cinematic
experience, effective stereo data visualizations are not always easy to create, and require that
close attention be paid to known perceptual principles. In Figure 10.7, for example, which
shows the distribution of earthquakes below the Big Island of Hawaii in 2003, use is made
of the principle that variations in the lightness visual variable are better able to heighten the
viewer's appreciation of depth than variations in hue (Ware, 2004).
An even more important decision-making factor concerns the evidence from visual per-
ception research, which suggests that stereo is less effective than several other techniques for
indicating depth and scene layout to the viewer when undertaking tasks in 3D visualizations.
Indeed, stereo is only one of nine major visual depth cues and, in many circumstances, is not
the most important (Cutting and Vishton, 1995). A question worth asking is whether the
visualization in Figure 10.7 provides a better indication of the three-dimensional distribu-
tion of the earthquakes than, say, a monographic view that includes symbol shadows (as in
Figure 10.6). One of the more significant alternatives to stereo is the kinetic depth effect, in
which an awareness of the depth relations among objects in a scene is induced by the relative
motion of foreground and background objects, either as the viewer moves or as the objects
Search WWH ::
Custom Search