Item

Abstract

In Virtual Reality, subjects typically misperceive visually simulated turning angles. The literature on this topic reports inconclusive data. This may be partly due to the dierent
display devices and eld of views (FOV) used in these studies. Our study aims to
disentangle the specic in
uence of display devices, FOV, and screen curvature on the
perceived turning angle for visually simulated ego-rotations. In Experiment 1, display
devices (HMD vs. curved projection screen) and FOV were manipulated. Subjects were
seated in front of the screen and saw a star eld of limited lifetime dots on a dark background.
They were instructed to perform simulated ego-rotations between 45 and 225
using a joystick to control the rotation of the image. In a within-subject design, performance
was compared between a projection screen (FOV 86 64), a HMD (40 30),
and blinders that reduced the FOV on the screen to 40 30. Generally, all target
angles were undershot. We found gain factors of 0.74 for the projection screen, 0.71 for
the blinders, and 0.56 for the HMD. The reduction of the FOV on the screen had no
signicant eect (p=0.407), whereas the dierence between the HMD and blinders with
identical FOV was signicant (p<0.01). In Experiment 2, screen curvature was manipulated.
Subjects performed the same task as in Experiment 1, either on a
at projection
screen or on a curved screen (radius 2m, FOV 86 64 for both). Screen curvature had
a signicant eect (p<0.001): While subjects turned too far on the
at screen (gain
1.12), they did not turn far enough on the curved screen (gain 0.84). Subjects' verbal
reports indicate that rotational optic
ow on the
at screen was misperceived as
translational
ow. We conclude the following: First, display devices seem to be more
critical than FOV for simulated ego-rotations, the projection screen being superior to
the HMD. Second, screen curvature is an important parameter to be considered for
simulation of ego-motion in virtual reality.