Game Development Reference
The functional interest of the object; again for a fighter pilot, an aeroplane is of
a much greater interest than a tree. This technique, called “technique of areas of
interest'', can be applied to the entire 3D database;
The linear velocity of the object with respect to the observer: the greater the speed
of movement of the object, the less visible are its details;
Angular velocity of the model: the greater the speed of rotation of the object, the
less visible are its details;
Position of the object on the screen: the eye is more accurate at its centre than
at its border, the objects located in the axis of the eyes thus require more details.
Besides, the field of vision tends to contract with speed. This argument can be
qualified by the fact that today there only some prototypes of devices to track the
movements of the eyes;
Visibility in the environment: objects in the fog disappear faster;
Finally, the load of the image generator: we can choose to eliminate the objects (or
use lower levels of detail) when the graphic machine cannot generate a sufficient
number of images.
These criteria must of course be adjusted to other applications, but they give three basic
ideas. The details are to be reduced depending on the distance between the observer and
the object, the load on the machine and the functional criteria related to the domains
of the application.
Once the decision of transition is taken, it needs to be applied to the model.
The most natural solution is purely and simply replacing a model of the object with
another. This technique is naturally not very expensive, but can lead to disturbing
visual artefacts (blinking of far away objects coming closer) during the change and
disrupt the user's attention. On the other hand, the blending technique consists of
displaying two models at the same time by making one of them more and more trans-
parent when the other becomes more and more opaque. The artefacts disappear, but
at the expense of the cost of calculation: It not only requires rendering two models,
but also requires calculating the transparency. 3D interpolation techniques are thus
a good compromise between switching and blending: These techniques involve deter-
mining the relations between the vertices of the levels of detail off-line and carrying out
an interpolation between these vertices during the simulation.
18.104.22.168 Generating the levels of detail
The general method to calculate the levels of detail can be divided into two phases:
1 Calculation of the importance: this involves calculating the use of a vertex in the
recognition of the model to which it belongs;
2 Construction: once the vertices are sorted as per their importance, the new level
of detail related to a single importance is generated, for example by eliminating
the vertices with low importance. Refining approach has also been mentioned
in the literature (approach of multi-resolution by wavelets (Eck et al., 1995)) where
the complete model is constructed from a simple shape by gradually adding the
vertices of the most refined model.