RouteFlow: Trajectory-Aware Animated Transitions

TL;DR

RouteFlow addresses the challenge of animating trajectories by preserving both global movement trends and critical local hotspots while minimizing occlusion. It introduces a two-stage pipeline inspired by a bus-route analogy: trajectory-driven path generation via bottom-up hierarchical edge bundling with an anchor-based force model, and incremental object layout generation via circle packing guided by a hotspot DAG. The approach demonstrates superior performance in revealing global trends and hotspots across seven real datasets and through a user study, while maintaining competitive object-tracking fidelity. The work contributes a concrete formulation of sequential bus-routing and seat-allocation subproblems and provides an open-source implementation for trajectory-aware animated transitions.

Abstract

Animating objects' movements is widely used to facilitate tracking changes and observing both the global trend and local hotspots where objects converge or diverge. Existing methods, however, often obscure critical local hotspots by only considering the start and end positions of objects' trajectories. To address this gap, we propose RouteFlow, a trajectory-aware animated transition method that effectively balances the global trend and local hotspots while minimizing occlusion. RouteFlow is inspired by a real-world bus route analogy: objects are regarded as passengers traveling together, with local hotspots representing bus stops where these passengers get on and off. Based on this analogy, animation paths are generated like bus routes, with the object layout generated similarly to seat allocation according to their destinations. Compared with state-of-the-art methods, RouteFlow better facilitates identifying the global trend and locating local hotspots while performing comparably in tracking objects' movements.

Figures (16)

• Figure 1: A comparison of three animated transition methods on a bird migration example.
• Figure 2: Illustration of our analogy.
• Figure 3: The pipeline of our method.
• Figure 4: Illustration of three types of forces in our algorithm.
• Figure 5: Illustration of our bottom-up hierarchical edge bundling algorithm.