Legible3D Streamtube Visualizations From Shading, Shadow, And Halo

Three-Dimensional (3D) streamtubes are popular for visualizing tensor and vector fields. However, dense streamtubes suffer from visual cluttering, which impedes identification of spatial structures from regions of interests. Accordingly, effort have been make to improve structure illu-mination and rendering by designing powerful non-photorealistic techniques especially to enhance depth perceptions. While these methods generate visually compelling results, it is unclear how much they aid data analysis in real-world task contexts.Based on Bertin’s semiology of graphics, We focus on popular depth cues for dense stream-tube visualizations:distance-encoded shadow and halo, and combined them with three representa-tive rendering techniques(Phong shading, Tone shading and Hatching) on a set of tube visualization tasks for diffusion tensor MRI studies. Our result shows that both different shadings and distance-encoded shadow have significant impact on accuracy and task completion time. Tone shading lead to the most accurate answer with least completion time. Phong follows tone and hatching performs the worst. Distance-encoded shadow has significant bad impact on both accuracy and completion time. Finally, halo shows no significant impact but produces a slightly more accurate answers.