BloomTree: Dynamic Coloring Techniques for Exploring Deep and Wide Tree Structures

Abstract This study introduces BloomTree, an interactive Sunburst system for visualizing massive hierarchical datasets, such as the Tree of Life. Traditional static color schemes fail to maintain perceptual distinguishability when applied to millions of nodes, and managing the full tree in memory is computationally costly. BloomTree addresses these challenges through a combined strategy, centered on dynamic, view‐dependent, topology‐aware color allocation. This technique recomputes color assignments for the visible subtree after each navigation step. This ensures closely related regions appear with coherently related hues and adjacent sectors maintain sufficient contrast. Furthermore, to preserve the user's mental map during navigation, zooming operations are smoothly animated by simultaneously interpolating sector geometry and color assignments. For computational scalability, BloomTree uses selective subtree streaming and metanode aggregation to handle massive data, storing the full tree on a backend server and streaming only a truncated visible region. We evaluate BloomTree at the system and user levels. System analysis shows that selective streaming and metanode aggregation significantly improve computational scalability, reducing communication volume to about 1/70 and rendering cost to about 1/2,000. A controlled user study confirms that dynamic coloring yields faster responses and higher accuracy for several perceptual and structural tasks, highlighting the effectiveness of the system for exploring multi‐million‐node trees.