Game Development Reference
These calculations have a very difficult constraint: their duration must not be
perceptible. Otherwise, the user will not only detect that the image is animated, but also
a jerky sequence of fixed images, which will give him a very uncomfortable experience.
When this constraint is respected, we talk about real-time rendering 1 .
Two factors need to be taken into account to reach this goal of real time. First,
the latency of devices, mainly the position and direction sensors. In fact, the dura-
tion between the user's movement and the computer' reception of the information to
change the position has to be the shortest possible. Otherwise, the time lag between
the movement and the repercussion on the image produced will make the visualisation
and thereby the entire application very uncomfortable (similar to seasickness).
Then, the minimum frequency of calculation and mainly display of images should
be approximately 25 images per second (similar to that of a cinema). The frequency
is often higher, about 50 to 80Hz, for the stability of images and thereby the user's
comfort. As a result, the virtual reality system has at the most 1/25th of second to
calculate and display an image. This extremely short duration has a direct impact
on the complexity of the simulation of the rendering and proves to be the biggest
bottleneck for virtual reality applications.
That is why a number of research projects have been undertaken to reduce the time
taken to calculate an image. Another expression of the same objective is to increase
the complexity of the simulation either geometrically (by increasing the number of
polygons processed per image) or optically (by improving the complexity of the ren-
dering model) maintaining the calculation time constant. Along with these algorithmic
research projects, the performances of VR systems have improved “simply'' by bene-
fiting from the increase in the power of processors and mainly from specialised graphic
architectures (3D card) created mainly for the video games industry.
15.1.3 Quality and perception
The quality of digital information provided to the user's brain is very difficult to assess
since it is directly related to the mechanisms involved in human perception, which
are not yet perfectly known and which vary from person to person. The same set of
rendering methods would thus be suitable for one group of users and ineffective for
another. It is thus useful to define the criteria to assess the quality of the rendering in
Computer experts were using the single criterion of (photo) realism for a long
time to qualify the quality of their digital images. It involved comparing the photo-
graph of an object or a scene with the computer-generated image of the same object;
the computer-generated image was called realistic if the human eye could not find
any difference between the two. This assessment guided the researchers in developing
methods more and more similar to and inspired from the models used by physicists.
All the radiosity algorithms, which were very much in vogue in the 90s, are a perfect
example of this connection. An advantage of this path: development of special methods
producing very “beautiful'' images; a disadvantage: by getting closer to the physical
1 Please note that this use of the term “real time'' does not have the same meaning as used in the
computer domain, for example, where the time constraints are even more strict.