While the first work on PLOD dates back to ’76 [1], most of the development has been done during the last decade. These advancements can be grouped into three areas, namely criteria, mechanism and error measure. Criteria are a set of functions that select areas from the objects that need to be drawn with a certain LOD. The mechanism is another set of functions that modify the geometry to achieve the desired LOD. They correspond to polygon simplification mechanisms that fall under four categories [5]: sampling, adaptive subdivision, vertex decimation and vertex merging. The error measure is an evaluation of the differences between the original object and the modified one, and it is used to control the mechanism. Measuring deviations from the original mesh to the modified mesh allows the quantification of the errors introduced when modifying the mesh. Common error measures in the literature include vertex-vertex, vertex-plane, vertex-surface, and surface-surface distances. Ideally, we would like these errors to be imperceptible to the user.
The most important part of a PLOD system is the set of criteria used to modulate the LOD. These criteria are related to or based on physiological and psychological aspects of human vision [2, 3]. We list below several important criteria [4]:
There are several examples of systems that make use of eye-gaze for guiding perceptually motivated simplifications including Reddy [10], Luebke [3], Williams [18] and Murphy [17]. Both [3] and [17] make use of an eye tracker to estimate the eye-gaze vector. In [3], the user‘s head is placed in a chin rest to avoid having to calculate the position of the eyes. Only [17] tracks the head and the eyes simultaneously allowing the user to move in a more natural way.