Game Development Reference
Similar tests were done with images calculated in real time (for eliminating pseu-
doscopic movements). Testers should perform a pointing task for half an hour in a
processed or unprocessed virtual environment. The ease of accommodation, range of
accommodation as well as stereoscopic acuity was measured before and after the task
to measure visual strain less subjectively. This experience showed that attenuation of
visual strain after the task is real and can be measured.
On the other hand, these processings can be surprising especially for stereoscopic
image learners, because it slightly deteriorates the clarity of images. All the qualitative
and quantitative results are given in the thesis of Jérôme Perrin (1998) as well as that
of Laure Leroy (2009).
For any virtual reality application using stereoscopic images within a professional
framework, adaptive blur filtering makes depth perception by reducing visual strain
13.1.5 Creation of images in orthoscopic vision
for a design review
In our IMAVE project for Sommer Allibert (Immersion Adapted to a Vehicle) that we
have already presented, the main objective is the visualisation of a dashboard from the
driver's seat. The application makes design review possible among individuals from
different professions. After the study of sensorimotor I 2 , it was agreed that depth
perception would possibly be the best (especially for perception of curves and shapes).
Is viewing the dashboard from the driver's seat possible in orthoscopic vision, without
excessive fatigue? Since the dashboard also has to be observed when the driver bends
over, it was decided that it would be necessary to track the driver's head as well.
Without showing all the calculations in this section, a solution is possible. We will
remember that to perceive the dimensions and correct volumes of objects better in 3D,
it is necessary that restored vision respects depth perception by the laws of perspective
and by the laws of stereoscopy. To have orthoscopic vision, you must position the
screen at the dashboard level and at an optimum distance with respect to the eyes in
such a manner that the two limits at the front and at the rear of the dashboard involve
the same horizontal parallaxes (in absolute value) that are less than 1.5 ◦ .
With the real dimensions of the dashboard and the nominal position of the driver's
head (in driving position), we get a solution for orthostereoscopic vision with the per-
ceptions of dimensions and volumes being very accurate. But the solution is technically
limited because the horizontal parallaxes are close to 1.5 ◦ (Figure 13.15).
The problem is more complicated if the observer wants to look at the dashboard
from a closer distance: the negative horizontal parallax exceeds 1.5 ◦ in absolute value.
A compromise had to be made: when the observer is closer to the dashboard and as
a result, closer to the screen, we reduce the stereoscopic effect of depth (in practice,
the two viewpoints of virtual cameras move closer and are therefore no longer in the
nominal positions of the eyes). The results for viewing the dashboard were satisfactory
for some people who could see the dashboard in 3D, in orthoscopic vision and without
significant visual strain.