Representation schemes with support to multi scale models have been explored in many scientific areas, specially in computer graphics. The most important theories involved in this process are scale spaces and multiresolution techniques. Those theories are well defined for models where we can build a regular parametrization of its domain. Unfortunately, this is a very restrictive property in most computer graphics applications. The main contribution of this work is to propose a new approach, capable of building models in different scales, based on the most important characteristics of scale spaces and multiresolution. An unique framework is defined, built as a combination of discrete and volumetric implicit representations, which is able to describe an object in different scales/resolutions. Each implicit representation in this collection is generated by a sampling procedure of the euclidean distance function, induced by a linear piecewise parametric surface, representing a Manifold. We present two approaches to build these sets, each one inspired on scale space or multiresolution theory. The first one is the intrinsic approach, where the scale/resolution control is obtained applying simplification and/or fair procedures directly on the piecewise linear parametric surface. On the other hand, the extrinsic approach deals with scale/resolution by working in the domain of the implicit representation, through the application of morphological filters