Dimensionality Reduction (DR) scatterplot layouts have become a ubiquitous visualization tool for analyzing multidimensional data items with presence in different areas. Despite its popularity, scatterplots suffer from occlusion, especially when markers convey information, making it troublesome for users to estimate items' groups' sizes and, more importantly, potentially obfuscating critical items for the analysis under execution. Different strategies have been devised to address this issue, either producing overlap-free layouts, lacking the powerful capabilities of contemporary DR techniques in uncover interesting data patterns, or eliminating overlaps as a post-processing strategy. Despite the good results of post-processing techniques, the best methods typically expand or distort the scatterplot area, thus reducing markers' size (sometimes) to unreadable dimensions, defeating the purpose of removing overlaps. This paper presents a novel post-processing strategy to remove DR layouts' overlaps that faithfully preserves the original layout's characteristics and markers' sizes. We show that the proposed strategy surpasses the state-of-the-art in overlap removal through an extensive comparative evaluation considering multiple different metrics while it is 2 or 3 orders of magnitude faster for large datasets.