It is quite clear, that two main requirements, which the custom graphics has to meet are connected with creating a dynamic and realistic model at the same time. But, it should be pointed out, that if we want to reach the high quality, representing the object, - the huge volume of data is required for its description.
But, firstly, the modern powerful equipment is very expensive and therefore, it is not always available for an usual customer (but, in fact, a lot of the computer games were developed for him). But, secondly, it should be said, that the appetites of the developers surpass in several times the possibilities of the today's graphical accelerator.
So that, the following question still continues to be actual: how to find the necessary description for the object, which, firstly, can provide it with some degree of the representation accuracy and, secondly, the given object also has to be as compact and convenient for processing as it is possible?
Thus, the certain knowledge in the field of custom graphics, concerning the question: how the human eye and brain accept the reality, can help to solve this problem.
The fact is that, a three-dimensional object can't belong to the field of vision entirely, because some fragments can be hidden behind the other fragments or objects. As to the back side of the object, it is not visible at all.
There is some vector, that is the set of numbers, which describe the object. The hierarchical analysis of the custom graphics is connected with comparing this vector with some basis in the vector space.
The visual perception process of the human brain has a similar hierarchical structure.
Neuron networks are the logical elements, created on the basis of the mathematical models of the nervous system of the living creature. The main elements of these networks are neurons. A neuron is a logical element, which has one or several inputs and one output. From time to time, a neuron receives the impulses from the other neurons of a network. If the summarized impulse at the input exceeds the neuron's threshold, the neuron will form a signal at the output. Otherwise, the signal will not be created. So that, each neuron is a filter of information, which works under the certain conditions. The connections between neurons are determined only by the network configuration.
So, the neuron networks have allowed researchers to present the model of the sample graphics, which shows the main processes of the human vision system.
The model uses the following abstraction levels of the custom graphics: pixels, zones, objects and labels. The certain processing steps correspond to each level. Such image features as contours and color boundaries are described at the pixels level. The next level determines the zones of pixels, their sizes, dimensional orientation, color and texture. The third level allows to process the objects in the observer's coordinate system. The fourth level processes so-called labels or prototypes, i.e. the total object characteristics, which don't depend on the position and orientation of an observer.