From Theory to Practice: Design, Vision and Visualization


Diane's comments

Diane was unable to make it to the conference. However, she sent these comments based on the workshop papers. These comments are based on the images in the papers.

General concepts

1) Genre. Look at all of the examples out of context. This might seem unfair, but it tells us something about how people can discern information in a split second.

In addition, if any of these visuals were considered to be in the field of visual analytics, they would need to be readable by non-specialists or non-experts (refer to Chapter 5 of Illuminating the Path).

Genres allow us to instantly make associations with familiar visuals, such as pie charts and scatter plots.

2) The greyscale test. Do all of the figures work when converted to a greyscale? Though not always accurate, this rule of thumb can tell us a lot about whether the colors work. Remember too that some users still print out (in black & white) what is designed for the computer screen.

The first Tableau figure, for example, works in Black & White, even though the information is complex.

The color in the Daisy Map example, however, is too saturated. This is probably because the information visuals are reduced to small slivers of color that have to fight with a white luminous background. Where possible, use a 30% to 50% grey background. This will dampen down the amount of light that is emitted and will enable you to use more colors that don't fight with the background.

Using an either black or white background often causes flare from the sheer amount of emitted light.

3) Hierarchy This is one of the most important aspects of visual legibility. Most people not trained in graphic design try to give emphasis to everything. This never works, because the eye struggles to focus on the dominant elements.

By emphasizing the first, second, and third most important aspects of the visual information, the eye is directed and come to rest on the most important information. When the eye scans the visuals to discern more information, a hierarchy allows it a resting place, and provides a visual context.

Specific comments


In the RENCI example, the logo, tagline  and the titles fight with the storm visualization. In addition, putting text over the flow visualization, or any visual makes legibility of the text difficult. The designer seems to have sensed this, and so added a grey field. Nonetheless, the grey field is not strong enough to dampen the confusion among visuals.It only works when the visual has very little contrast - 10% or less. As a rule of thumb, never put text over an image.

This is clearly identifiable as a genre of storm animation sequences, and works well in the greyscale test. The sequences themselves are more legible than is the image on the website page because the animations don’t have to compete with overlapping, transparent information. One thing that may improve the animation is interactivity IF the context (i.e.,the users) warrants it. For example, the directional arrows dominate. Can they be greyed down or interactively disappear so only the storm is visible?

How do the designers know this was a “successful” design? Who are the users? Does the visual solution work simultaneously with an indication of geography? More of these issues need to be discussed.

2.        Chart Tamer

Not knowing the full context of the problems this designer faces (which seems to be immense), it is difficult to find fault with the design. The colors – not all primary or all secondary – work especially well. The light grey background works to enhance legibility. The one issue that could be discussed is the apparent overemphasis on boxes (boxology) in the left Chart Picker. They were used I think for two reasons. First, the boxes make the distance from the work length to the buttons seem to be equidistant. Second, the boxes seem to make everything of one type == selectable buttons. But this is where the question lies: EVERYTHING seems to be a clickable button, despite the designer’s careful attention to hierarchy and subtle use of grey for emphasis. I’d like to see the Chart Picker with everything flush left, and without the boxes for #1 and #2.

3.        Horizon Graph (Panopticon)

I was prepared to dislike what seemed and overly dense graph. However, it seems to work. It stands the grayscale test surprisingly well. In terms of hierarchy, I thought the slicing down of information wouldn’t work; for the ranges depicted though, it does work. Could interactivity makes this better? I’m wondering if, for example, a rollover and zoomable lens might offer users a “closer look” at defined time stages.  One issue with the color though is the red-to-orange selection. It is somewhat of a good choice because it is the opposite of the “cool” blue.  However, it appears harder to discern between layers, especially in the denser (more extreme) areas.

4.        Students Assesment Data

Several things could be addressed to enhance the legibility with this example. First, the grid lines are too dominant and interfere with the visual hierarchy. Second, this one does not stand the test of being transformed into grayscale. The color is the most important visual to change, as it is now difficult to discern where the colors are (which is why the designer used the zoom in window), and what they distinguish. This is because the white background fights with the very saturated green color. It is difficult to see past the green, both because of its saturation, because it fights with the white background, and especially because the shape reduces the information area to very small bits. It works better than if it were a pie, but at the same time, the ”spokes” reduce what can be legibly read at such a small size. Even a slightly larger size would enhance legibility. I would experiment with shapes and especially color.

5.        AstraZeneca

Abandoning the “before” visual was really a good idea – it didn’t work for many reasons. The one thing that the “after” succeeds at is the separation among various phases. That works quite well. What still doesn’t work as well as it could, however, the legibility of the years and the legibility of Active, Delayed, Discontinued and Superended (block titles). The latter would be easier to read if it read in the “up” rather than “down” direction, because then all of the vertical text would be going in the same direction. What is the least legible, and possibly more important, is that the “after” solution makes following individual drugs over time very, very difficult. I would imagine in order to do so, either an interactive solution would need to be deployed – such as a rollover – or a text-based search. It would be better if neither were required. Nonetheless, the “after” solution is a dramatic improvement.

6.        Redesigning Parallel Sets

The redesign of this is apparently minimal. Upon investigation, however, it makes reading the information quite a bit easier. This is because the designer has reduced components into more legible chunks of information. Second, though the change in color may initially seem minimal, it is far easier to read, with the partial exception of the yellow. Third, the designer uses more of the visual field for the set, rather than saving some of the visual area for textual information. Thus, I’d agree with the designer that small, informed changes can mean a big difference in how well a visual communicates.   Could interactivity improve this solution further?

Diane Gromala is an artist, designer, visual technology innovator, and Professor in the School of Interactive Art and Technology at SFU. Her work has been at the forefront of emerging forms of technology, from the earliest form of multimedia (HyperCard, at Apple Computer) to one of the very first instances of Virtual Reality art at the Banff Centre in 1991. Gromala’s art has been performed and exhibited in North America, Europe, the Middle East, Asia and New Zealand. It has also been featured on the Discovery Channel, CNN, the BBC, the New York Times, the Los Angeles Times, to name a few. Along with collaborator Lily Shirvanee, Gromala was a semi-finalist for Discover magazine's Award for Technological Innovation in 2001. Gromala's design work has received numerous awards from organizations ranging from the AIGA to the American Institute of Architects.