The Role of Map Animation for Geographic Visualization - Geographic Visualization: Concepts, Tools and Applications - page 58

also change over time (e.g. light type, position, distance, angles, colour attributes). The speed
with which an animation is viewed is crucial (Griffin et al ., 2006) and whether the viewing
order of the frames in an animation is predefined (i.e. without any interaction mechanisms)
(Hegarty, Kriz and Cate, 2003). The effectiveness of a small multiple displays on the other
hand depends mostly on the appropriate number and choice of the small multiples, that
is, how many and which of the key events (macro steps) are selected to discretely represent
the continuous, dynamic process. Well-designed small multiple displays depict the most
thematically relevant (pre-selected) key events, and unlike non-interactive animations allow
viewers to inspect the display in their own pace and viewing order.
Figure 4.3 shows a test participant's eye movement patterns overlaid on an identical small
multiple map display, but during two different data exploration tasks. The graduated circles
show eye fixation durations (the larger the circle, the longer the fixation) and the connecting
lines represent rapid eye movements between fixations, called saccades . In Figure 4.3(a) the
task is to gain an overall impression of the small multiple display, and to verbally describe the
patterns that are discovered during its visual exploration. When animating the gaze tracks,
one can see that the viewer is not exploring the display in the implied sequence of the small
multiple arrangement, but going back and forth between the maps several times or jumping
between different rows of maps.
In Figure 4.3(b) the task was to specifically compare two maps within the display. The
significantly different viewing behaviour from Figure 4.3(a) suggests that small multiple
displays will never be informationally equivalent to non-interactive animations as Tversky,
Morrison and Betrancourt (2002) are implying. When map use contexts require a user
to compare items in a time-series (across time, space or attribute) the non-interactive
animations (locking a viewer into a pre-defined sequence) will always add cognitive load,
as the viewer will have to wait and remember the relevant information until the respective
comparative displays comes into view (see Figure 4.4).
The small multiple map display on the other hand allows the user to freely interact with
the data in the viewing sequence they deem necessary for the task. In other words, the
experimental data (Figure 4.3) suggests that non-interactive animations should be made
interactive to be informationally equivalent to small multiple map displays.
Two animation types depicted in Figure 4.4 dynamically depict the same information
shown in the static small multiple map display (see Figure 4.3). Figure 4.4(a) represents a
frame of a non-interactive animation, only containing a start button, while the interactive
animation depicted in Figure 4.4(b) allows a viewer to start and stop the animation at any
time and change its display speed and movement direction.
The great power of carefully designed animations, however, are their ability to display
micro steps in complex systems that might be missed in small multiple displays (Slocum
et al ., 2001; Jones et al ., 2004). The apprehension of micro steps is directly related to the
perception of apparent motion problem. As Griffin et al. (2006) note, this is still an unsolved
issue in animation research. The perception of apparent motion (i.e. being able to visually
interpolate fluid motion from discrete jumps in position between images) depends on (1)
the relationship between the duration of a frame in the animation being displayed, (2) the
frame rate in the animation (how many frames are displayed per unit time) and (3) the
distance of an object moving across the screen. Rensink, O'Regan and Clark (1997) have
demonstrated that observers have great difficulty noticing even large changes between two
successive scenes in an animation when blank images are shown in between scenes (e.g.
simulating a flicker). This change-blindness effect operates even when viewers know that