Geography Reference
In-Depth Information
of the East End of London in the late 1880s are displayed on a conventional 2D map, using
data from the manuscript notebooks of Charles Booth's poverty enquiry (Shepherd, 2000).
Unfortunately, because of the widespread incidence of dwelling multi-occupancy in this area,
many of the point symbols visible on this map conceal others beneath them. A solution to
this problem (Shepherd, 2002) is to use the third display dimension in a 3D visualization
to display stacks of point symbols at each location, as shown in Figure 10.1(b). Because of
the widespread occurrence of locationally coincident phenomena, 2D point symbol maps
should be used extremely carefully, and 3D stacked symbol visualizations should be used
wherever appropriate.
10.3 Some problems with 3D views
Although 3D data visualization has undoubted benefits when compared with some
2D approaches, a considerable body of user experience and experimental research reveals
significant problems in using the third dimension to create effective data visualizations.
A number of these are explored below. However, space prevents detailed consideration of
the highly significant problem of user interaction within 3D virtual worlds, which are the
subject of a separate study (currently under preparation).
10.3.1 Scale variation across 3D scenes
A perspective view tends to be adopted for most 2.5D and 3D visualizations of spatial data.
Unfortunately, because of the foreshortening effect in such views, which increases with
distance from the observer, it is difficult to make accurate visual comparisons of objects
within a 3D scene. This problem was recognized early on in the history of 3D digital mapping
in an evaluation of the problems of interpreting graphical output from the SYMVU software
(Phillips and Noyes, 1978). More recently, evidence from perception research reveals that
not only is depth consistently underestimated (Swan
et al
., 2007), but that the human
visual system perceives relationships in each of the three directions separately, and that
the relationship between physical and perceived space is non-Euclidean (Todd, Tittle and
Norman, 1995). A number of solutions have been proposed to assist users in making effective
distance and measurement estimates in 3D worlds, four of which are briefly reviewed here.
It should be noted, however, that no single technique is entirely effective across the range of
tasks that need to be performed in 3D scenes.
Reference frames
Most 3D histograms and bar charts generated from non-spatial data provide a set of 3D axes
as an integral part of the display, and this assists across-scene position and size estimation by
the viewer. However, when spatially referenced data are visualized in 3D, a standard reference
framework of this kind is usually absent. In such cases, a user-defined bounding box may be
drawn around some or all of the objects within the scene in order to provide some sense of
scale. This may be a simple wireframe box, or a more elaborate set of calibrated axes, and the
geographical extent of the reference framework may be chosen by the software and/or the
Search WWH ::
Custom Search