Game Development Reference
In-Depth Information
This representation of the BRDF, and on a more general level the functions of
illumination of the scene, make it possible to better use the raw performances of
graphics processors, to divide the processing among the central processor and the
different stages of the graphics pipeline and thereby maintain the calculation time to
the minimum while ensuring an image quality corresponding to the requirements of
the application.
The different models that we saw are initially based on modelling and reproduction
of the physical phenomenon of generating an image and its acquisition by an ideal
sensor. However, one must not forget that in the scope of virtual reality, human eye
is the sensor in most of the cases 4 . Therefore, if we want to develop realistic real
time rendering in virtual reality, it becomes necessary to examine the efficiency of the
models developed with respect to their perception. Since the assessment of models in
virtual reality is not done in physical but in perceptual terms, it is important to study
the possible couplings between the models of lighting, perception and restitution in
order to calculate only that part which will be perceived and calculate it effectively.
The main objective of producing computer-generated images is their visualisation on
a suitable device. It is thus natural to wish to take into account the human visual
system, and mainly its limitations, when we want to calculate one or more computer-
generated images. In this section, we will study two contexts in which perception can
be taken into account: Use of vision models for the calculation of rendering and the
reproduction of tones.
15.3.1 Vision models and rendering calculations Vision models
The term “vision model'' refers to an algorithm that takes an image as its input and
provides the interpretation of this model done by the human visual system at the cortex
level. It is an algorithm that simulates the processing performed by the retina and the
nerve tracks.
Very few models of this type are mentioned in the literature. We will give a brief
description of three such models, which we find to be the most representative of what
is desirable in computer graphics and virtual reality.
The “Visual Differences Predictor'' by Daly
This achromatic model (Daly, 1993) has three steps:
Simulation of the luminance adaptation of the retinal cells using an experimental
curve on the perception thresholds;
4 Virtual reality is sometimes used to drive infrared sensor-based remote sensing systems. In that
case, the software environment is slightly adjusted to take into account the spectral ranges and
the sensitivity of the sensor, but the overall models remain unchanged.
Search Nedrilad ::

Custom Search