Interactive Transitions for Map Applications

GIS experts often need to relate and compare heterogeneous geographical representations of the same region. For example, existing maps are compared to recent satellite imagery to update geographic databases, like OpenStreetMap. The means to do so, are, however often limited to data agnostic techniques such as overlaying the representations with some degree of translucency or swiping between layers. These techniques do not support users effectively in their tasks in domains such as crime analysis or urban planning. This thesis aims at proposing new interactive transitions to combine those multiple representations into one, either spatially (spatial multiplexing) or temporally (temporal multiplexing).To better understand the limits of existing approaches, this thesis first contribution is an evaluation of five interactive map comparison techniques. We characterise these techniques in terms of visual interference, user attention and scanning strategy. We evaluate them by asking participants to find differences between real satellite imagery and topographic maps, that we purposefully modified introducing six kinds of differences. Results suggest that techniques that superimpose the layers are more efficient than techniques that juxtapose them and that having a more motor driven scanning strategy can be beneficial for some tasks. Drawing from the evaluation results and interviews with GIS experts, the second contribution of this thesis is MapMosaic: a novel spatial multiplexing technique to combine geographical layers. This dynamic compositing model enables users to interactively create and manipulate local composites of multiple vector and raster map layers, taking into account the semantics and attribute values of objects and fields. We evaluate MapMosaic using two approaches: first we compare MapMosaic’s interaction model to QGIS’ (a widely used desktop GIS) and MAPublisher’ (a professional cartography tool) using the ‘Cognitive Dimensions’ framework and through an analytical comparison, suggesting that MapMosaic’s model is more flexible and can support users more effectively in their tasks. Secondly, we report on feedback obtained from experts, which further confirms the potential of MapMosaic, by describing precise scenarios where it could be useful. Spatial multiplexing can be very useful when comparing different geographical layers. However, time multiplexing might be more suitable to represent dynamics, as changes can be animated. This can be particularly useful when presenting evolution across satellite images, to illustrate effects of climate change or a natural disaster’s impact. Thus, the third contribution of this thesis is Baia: a framework to create advanced animated transitions, called animation plans, between pairs of before-and-after images. Baia relies on a pixel-based transition model that gives authors much expressive power, while keeping animations for common types of changes easy to create thanks to predefined animation primitives. We describe the model and the associated animation editor. We also report on two user studies: the first one suggests that advanced animations are perceived as more realistic and better at focusing viewer’s attention than monolithic blending, and the second one gathers feedback about the usability of Baia’s animation editor prototype.