Interactive data exploration and manipulation are often hindered by the size of the datasets. This is particularly true for 3D datasets where the problem is exacerbated by occlusion, important adjacencies, and entangled patterns. These complexities make visual interaction via common filtering techniques difficult. In this paper we describe a set of techniques aimed at performing real-time multi-dimensional data deformation with the intention of helping end-users to easily select, analyze, and eliminate specific spatialand- data patterns. Our interaction techniques allow animation between view configurations, semantic filtering and view deformation.
Any subset of the data can be selected at will at any step along the animation. Selected data can be filtered and deformed in order to remove occlusion and ease complex data selections. The technique is simple to learn and implement, flexible, and real-time interactive with large datasets (tens of millions of displayed data points). We demonstrate our technique to three domain areas: 2D image manipulation, 3D medical imaging, and astrophysical exploration.