The simplest approach to multiresolution is explicitly providing the same object at different resolutions. Each mesh is stored independently, therefore only few levels of detail can be used. For this reason switching between two different representations causes distracting popping effects. Furthermore one single object can't be selectively refined. Viewing a large terrain model requires the whole mesh (even those parts far away from the camera) to be rendered at the same resolution that is necessary for a sufficient approximation near the viewpoint. Splitting a large object into several smaller parts is also a bad idea, although this can reduce the rendering time in some cases. But the borders between two patches at different resolutions become clearly visible, even cracks in the surface can occur.