The design space of scatterplots consists of a number of parameters such as marker size and shape, image width and aspect ratio, and opacity. Different parameters yield different visual impressions of the scatterplot. Perceptual optimization of scatterplots means finding the best design parameters to support a given visualization task. This requires rendering thousands of design variations. We describe an image-based method for rendering scatterplots, which is tailored to this scenario: it enables quick updates of the design by re-using previously calculated intermediate results, and is independent of the data set size. Our approach outperforms the classic method of rendering scatterplots, i.e., drawing each marker individually onto an image, and can therefore dramatically speed up the perceptual optimization of scatterplots. We provide an open-source implementation and an online service for our method.

CCS Concepts: Computing methodologies --> Rendering; Human-centered computing --> Visualization design and evaluation methods; Graph drawings; Visualization toolkits

Simo Santala, Antti Oulasvirta, and Tino Weinkauf

Representation, Perception, and ML

EuroVis 2020 - Short Papers

115

119