Information Technology Reference
In-Depth Information
Design.
The positioning and resizing virtual room experiment again used a
within-subjects design, manipulating the independent variables:
viewing mode
(stereo, mono);
number of shadows cast by objects
(none, one);
background scene
surface shape
(flat, zig-zag); and
background scene surface texture
(solid, checker-
board). The dependent variable performance measures included: accuracy, mea-
sured as
error magnitude
; completion time, measured as
response time
; and scene
rotational distance
, measured as the total number of degrees that a subject ro-
tated the scene in either a left or right direction. Distance error magnitude for
the positioning task was again defined as the Euclidean summation of the three
directional errors in the x, y, and z dimensions (i.e. distance error magnitude =
(
)
1
/
2
). Radius error magnitude for the resizing task was defined as
the absolute value of the difference between the radius length of the resized tar-
get sphere compared to the radius length of the correctly-sized referent sphere.
Trial completion time was the period of time, measured in milliseconds, from
when the scene first appeared until the subject pushed a button on the space-
ball, causing the next scene to appear. The positioning and resizing task trials
consisted of 144 unique virtual room scenes presented in random order to the
subjects.
The volunteer subject population consisted of 16 male and 14 female profes-
sional employees of GSFC. For each subject, one half of all trials were viewed in
stereo, with the remaining trials viewed in mono. One half of the trials randomly
presented scenes with no shadow, and the remaining trials had one shadow cast
by each object. In all scenes utilizing cast shadows, the shadow-casting lighting
position was fixed and stationary, directly overhead at the azimuth position. One
half of all trials randomly presented a 'flat' background scene surface, charac-
terized as a three-walled 'room' with a flat floor and walls (see Figs. 8 or 9).
Similarly, one half of the trials randomly presented a 'zig-zag' background scene
surface, characterized as a room with a 'corrugated' floor and walls (see Fig.
10). Also, one half of the scenes were presented with a solid background sur-
face texture (see Figs. 8 or 9), whereas the remaining scenes appeared with a
checkerboard texture (see Fig. 10).
e
2
x
+
e
2
y
+
e
2
z
Results.
The data was again analyzed by fitting a repeated measures MANOVA
model to the performance measures. The MANOVA model tested each of the
four main effects (viewing mode, presence of cast shadows, background scene
surface shape, and background scene surface texture) on the error magnitude,
response time and rotational distance dependent variable performance measures.
In the positioning task performance data, there were significant differences (at
the 95% confidence level) in the mean values of the dependent variables (distance
error magnitude, response time and rotational distance) as a function of all
four main effects. In the resizing task performance data, there were significant
differences (at the 95% confidence level) in the mean values of the dependent
variables (distance error magnitude, response time and rotational distance) as a
function of three of the four main effects. However, objects casting shadows did
not significantly influence subjects' resizing performance measures. Complete
quantitative results for both the positioning and resizing tasks will be provided
by requesting same from the lead author.
Search WWH ::
Custom Search