Empirical Assessment of the Suitability of Visual Variables to Achieve Notarial Tasks Established from 3D Condominium Models - Innovations in 3D Geo-Information Sciences - page 197

Table 3 Comparison of the selective ability of visual variable to distinguish between two groups
of elements

Our experiment Bertin Carpendale Ware

Color (hue) Ye s Ye s Ye s Ye s

Value - Ye s Ye s Yes (-)

Texture Ye s Ye s Ye s Ye s

Yes = Achievable, “-” = no possible conclusion

• Value: Ware ( Chap. 3 ) concluded that “ the use of gray - scale colors is not a par-
ticularly good method for categorically coding data (nominal scale). Contrast
effects reduce accuracy, and the luminance channel of the visual system is fun-
damental to so much of perception (shape perception, in particular) that it is a
waste of perceptual resources to use gray - scale encoding ”. Based on this state-
ment, we determine that value may be good but limited in some cases.
• Texture: Ware ( Chap. 6 ) concluded that “ texture could be used to visualize dif-
ferent categories of information. In order to make a set of nominal coding tex-
tures distinctive, make them differ as much as possible in terms of dominant
spatial frequency and orientation components. ” From this statement, we deter-
mined that texture performed well.

The following table presents the comparison about the capacity of the visual varia-
bles (color hue, value and texture) to distinguish the isolation between two groups
of elements (Yes means selection was achievable, the symbol “-” indicated where
no conclusion can be drawn) (Table 3 ).
Overall, the comparison indicates that color (hue) and texture perform similarly
based on the four references. Value is still missing an experiment on our side in
order to suggest a strong conclusion. Nevertheless two out of four notaries appre-
ciate value for accomplishing notarial tasks. The conclusion of Ware about value
also indicated some conflict. Ware justified his decision according to the disadvan-
tages of value, including less accuracy and the limited number of combinations
compared to other visual variables. It is hard to tell if it is those two points that
influenced the performance of value in our tests. But it is an interesting point to
investigate since our results are compatible with the conclusion of Ware that value
is “not particular good” (Ware, Chap. 3 ) .
Regardless, the fact that this comparison helps us to confirm our experiments,
it enables us to explore two research questions central to this project. First, if
we compare our results with the proposals of Bertin (2D paper), Carpendale
(2D/3D screen visualisation) and Ware (2D/3D screen visualisation), our pre-
liminary results indicate that color (hue) and texture performed similarly. About
value, some disparities with Bertin and Carpendale are revealed. Nevertheless,
it is currently difficult to explain this difference and its correlation with 2D
paper and 2D/3D screen visualisation. It may be correlated to notarial tasks
itself. Additional tests are required. Consequently, except for the visual variable
value and within the limits of our experiments, we can conclude that there is no