Spatial Cues in 3D Visualization - Ambient Intelligence for Scientific Discovery - page 121

motion enables parallax, which is instrumental in stereo viewing. In our three
experiments, we incorporated several types of motion: (1) controlled versus un-
controlled motion (in the object matching experiment); (2) moving the viewed
objects themselves (in the positioning tasks of both the 'virtual world' and 'vir-
tual room' experiments); and (3) moving, or rotating, the viewed scenes (in both
the positioning and resizing 'virtual room' experimental tasks). To summarize
our observations about the effects of motion on these spatial task performances:

1. Controlling the motion of the rotating object in the mental rotation exper-
iment enabled more accurate object recognition and object matching per-
formances. However, subjects took longer judging whether the objects were
the same or different when they controlled this motion. However, we can
conclude that motion improved the quality of their object recognition and
matching judgments.
2. Moving the objects (especially in the presence of a cast shadow) in both
positioning tasks of the 'virtual world' and 'virtual room' experiments was
instrumental in providing distance (i.e. depth) information about the target
object, especially in relation to the other (stationary) objects in each scene.
3. The ability to left- and right-rotate the 'virtual room' scenes was used to
a much greater degree to extract depth information when other available
depth cues were diminished.

Stating this last point differently, 'virtual room' scenes, in both the posi-
tioning and resizing tasks, were always rotated more when there was less depth
information available from the other cues present. That is, the scenes were ro-
tated more when viewed: (1) in mono compared to stereo; (2) with no shadow,
compared to one shadow, present; and (3) when the rooms were viewed against
the more simple flat surface shape or solid texture characteristics, compared to
the more complex 'zig-zag' surface shape or checkerboard textures. We observe
that the rotational distance depth cue was used to supplement, or directly sub-
stitute, depth information as it became less available from other existing cues.

4.2 Stereo Viewing

Stereo viewing was a dominant and powerful depth cue for promoting task per-
formances in all three experiments. While viewing scenes in stereo, in comparison
to mono viewing, subjects were consistently more accurate and faster: (1) recog-
nizing and matching objects shapes in the first experiment; and (2) positioning
and resizing objects in both the second and third experiments. Furthermore, in
the third experiment, subjects relied less on the rotational distance depth cue in
the presence of stereo viewing. In short, stereo viewing consistently supported
every aspect of the spatial task performance measures in these three controlled
experiments.
We conclude that stereo viewing is a dominant depth cue for spatial task
performances of these kinds. Consequently, it may be more useful to test stereo
viewing conditions as a baseline against which to measure the relative ecacy
of other perceptual cues, rather than testing the effects of stereo viewing per se.