Vis 2008: InfoVis session

In a display of the quirkiness of having two somewhat disjoint events, I’m now sitting at the InfoVis section of the Vis part of VisWeek.

(I’ve written about this paper before.) Bachthaler et al. in “Continuous Scatterplots” consider the problem of drawing a scatterplot for very dense multidimensional dataset: in fact, infinitely densely ones. The basic idea of their work is that when you move a chunk of the spatial domain into the data domain (which is really what the scatterplots do), you have to be careful with “stretching” or “squishing” the chunk. This simple observation leads to a very nice mathematical formalism and an explicit formula for the density in the data domain over any dimension.

Their formulation is very similar to what we did for our paper. While I think we have more discussions about the implications of the histogram formulas, their visual results are absolutely impressive, and their formulation is, in principle, more general than what we have.

Jorik Blaas (in joint work with Charl Botha and Fritz Post) is now presenting his extensions of parallel coordinates for visualizing scientific data. I like the way his plots look. They’re almost continuous, which suggests the obvious juxtaposition: can we derive a coarea-based formulation for drawing parallel coordinates? In fact, Jorik just mentioned he uses blending: I’m now almost sure we could do nice, continuous parallel coordinates plots through the co-area formula. In fact, just drawing quads with additive blending using something similar to Bachthaler’s technique would probably work. however, I haven’t read Blaas’ paper yet, so there might be other issues. Ah, he just mentioned continuous parallel coordinates as a future extension. That would rock.

I was happy to hear that his code will be available. You too should be doing this, if only for selfish reasons: it makes it very easy for people to cite you. In fact, the easier you make your software to use, the harder it is for people not compare their work against yours. Everybody wins!