Game Development Reference
In-Depth Information
Figure 3.10 Variation of visibility and perception of texture. (Illustration:A. Fuchs, with permission)
7 Perspective:
This is certainly the most effective rule allowing the perception of a world in relief on
a monoscopic screen. This technique is used in painting, right from the Renaissance
era, to show three-dimensional space on a plane surface. It is worth noting that when
the video game designers speak of “3D'' video games, they change the plane images
(without perspective) to images with perspective on monoscopic screens. There are var-
ious types of perspectives (isometric, geometric, photographic, artistic…). Isometric
perspective is used in industrial design. It is distant from the reality than the photo-
graphic perspective but so much easier to draw by hand. There is no vanishing point,
but “receding'' parallels as shown in Figure 3.11.
Note: The reader can note that we have just used these rules which are shown in
the graph in Figure 3.12. You can close one eye but the perception of depth remains
the same! These rules are now frequently used in computer-generated images to give
a three-dimensional representation on a computer screen. For example, in Open GL
you can calculate your images in isometric perspective (orthographic projection) using
gluOrtho2D command or in photographic perspective 3 (perspective projection) using
gluPerspective command.
3.2.2.2 Convergence and retinal disparity
Let's see the phenomenon of convergence and its effects on the retinal images: By staring
at a finger held at 20 cm, you see a ghost image of any object in the background. On
the other hand, if you stare at an object placed behind your finger, your finger looks
split into two: the brain cannot merge two different images of your finger as the retinal
disparity is too large (Figure 3.13).
As the visual fields of the two eyes (partially) overlap and the optical axes converge,
we have two slightly different views of the same scene which helps us perceive the depth.
3 In other words, with a pinhole model and thus without involving optical distortions.