||Multiresolution modelling consists in obtaining representations of an object at different levels of detail and subsequently selecting the proper resolution according to the visualization conditions. Within this representation framework, either the generation of the simplified meshes or the management of the level of detail, are performed following certain criteria or error metrics. Classic simplification techniques are based on considering the geometric error introduced during the coarsening process as a measure of quality of the simplified mesh. However, in those cases in which the final receptors are humans, it is convenient to take into account the features that make the human visual system more sensible to certain stimuli. This circumstance encourages the transformation of the error metrics in order to incorporate the visual quality perceived by a human observer and to preserve the most outstanding regions. In this seminar, the viability of incorporating the proximity to the relevant regions in an object as a criterion to guide the simplification process will be analyzed. The inclusion of this distance measure provides more discriminant information than the mere identification of relevant features. With this approach, a general framework is proposed in which every element of a polygonal mesh has associated an importance value. This measure is understood as the proximity to perceptually relevant regions and is used in order to weight the allowed error during the simplification process. The identification of outstanding regions can be faced following different approaches, by analyzing the final 2D rendered image or by directly studying the 3D model of the object. The criteria applied to define the perceptual relevance and the final application in which the object will be used condition the technique used to compute distances, being possible to carry out the computation either in image space or in object space. The flexibility of the proposed approach allows its application independently of the nature of the extracted features and its integration into different simplification techniques.