From CS294-10 Visualization Sp11
Lecture on Apr 27, 2011
- Animation: Can it facilitate? Tversky, Morrison and Betrancourt. (pdf)
- The Value of Visualization. Van Wijk. (pdf)
- Visualization Research Challenges. NIH/NSF Research Report. Johnson et. al. (pdf).
- Principles of Traditional Animation Applied to Computer Animation. Lasseter (acm)
- Representing motion in a static image: Constraints and parallels in art, science, and popular culture. Cutting. (pdf)
- On creating animated presentations. Zongker and Salesin. (pdf)
Saung Li - Apr 28, 2011 07:04:39 pm
It amazes me that there are so many animated visualizations that are complicated because there are so many moving things happening at once. I used to think that I get confused because the concept is hard to understand, but in fact it is the visualization that is making it difficult. The alternative shown in class is to use static images showing the direction of things happening, which sounds boring compared to animation, but does potentially help tackle the problem. There can be cases were static images are insufficient in conveying something, though. In situations where there are not many things happening simultaneously, animations should be able to aid the user a lot in understanding a concept. Perhaps we can have a way for users to interactively select which part of the visualization to animate so that they can focus on one part at a time. Of course, they would need some guidance for knowing which parts they should choose before others.
Matthew Can - Apr 28, 2011 07:21:04 pm
After reading the paper by Tversky, Morrison and Betrancourt, I was surprised that animation is largely a failure at conveying information over static graphics. But this may just have to do with the way the research community has approached the problem of evaluating animation. As mentioned in the paper, the utility of an animation is evaluated against a comparable static graphic. The problem is that it's difficult to find a static graphic that is equivalent to an animation in every way except for the animation itself (i.e. same exact same information content). I don't think it's fair to conclude that animation has been a failure. Rather, I think we should take a step back and think about whether we've done the proper evaluation of animation.
Brandon Liu - Apr 28, 2011 08:48:40 pm
Paper on animation of statistical data graphics: http://vis.berkeley.edu/papers/animated_transitions/
I think one confounding distinction that could be cleared up is the gains from animation vs. gains from interaction in a visualization. A major flaw of the animated visualizations I have seen is that I need to experience the animation multiple times to make sense of anything, and I need to choose what to attend to each time. http://www.guardian.co.uk/football/world-cup-match-replay is a not-so-good visualization of world cup matches using the same technique as Gapminder. While it is instructive to be able to scrub back and forth in a match timeline, it is hard to interpret the areas that contstantly move and change.
Sally Ahn - Apr 28, 2011 11:50:37 pm
The video of animated transitions Brandon posted demonstrates cases where animation can be greatly beneficial. It's surprising that Tversky, Morrison, and Betrancourt's findings indicate that this is rarely the case. I think one important aspect of Jeffrey Heer's animated transitions is the separation of changes into stages. Namely, most stages restrict all changes to occur along one axis. Tracking the dots of a scatterplot that all moved horizontally was much easier than tracking dots move freely on a 2-dimensional plane (like the white dots example in lecture). Staged animations help to overcome the difficulty of perceiving "fleeting...minute changes simultaneously" that Tversky points out.
Michael Cohen - Apr 30, 2011 03:44:34 pm
I think squash and stretch could be very useful in addressing the problems with instructional animations that Tversky et al identify. In class, we saw a few animations of mechanical systems that were accurate and clearly drawn, but still hard to understand. One reason that I often find those diagrams difficult is that they don't show causality very well, because all the pieces are so tightly coupled and rigid. In the cylinder animation, it's hard to intuit when the ignited fuel is pushing on the piston (power stroke) and when the piston is doing the work of drawing in the next round of fuel/air (intake stroke). If the piston held still for a moment and we saw a bit of squash on it before it was pushed down by the ignition, it would be immediately intuitive where the energy was coming from in that stroke. Likewise, if we saw the piston stretch out a bit during intake, and the air/fuel particles lag behind the top edge of the piston, we could easily see that in that stroke the inertia of the piston is what's driving the motion, not the pressure of the incoming mix.
A "cartoony" representation would be somewhat less accurate physically, but would communicate the basics of the cycle to novices more clearly. And note that this is an effect that could only be achieved with animation, since squash and stretch wouldn't make much sense in a static drawing.
Hm. This would have been an interesting final project.
Michael Hsueh - May 01, 2011 12:46:14 am
I admit I was one of those that liked Hans Rosling's visualizations. After some thought, I realized that indeed it was because of the quality of the presenter as well as the fascinating subject matter that covered for a mostly deficient use of animation. I will say that one positive aspect of the Gapminder playback is that it conveys a feeling of passage of time as the animation plays. I know the same can be said of static time series data, but in a different way. I think the real-time, dynamic playback of the data makes it engaging, even while it admittedly obscures a lot of information. Perhaps the effect is analogous to what was discussed in a previous lecture about how much creativity, accuracy, etc. people derive from hand sketches versus CADs. The Wijk paper had a very interesting discussion about the role of visualization in presenting / "selling" data, as opposed to exploring it. The peculiar example of commercials is given as an example. In this class, I would say we've mostly been focused on the technology and science views of visualization, as described by Wijk. If we consider the Gapminder animation through a presenting (or even artistic) lens, does our evaluation change?
Julian Limon - May 01, 2011 07:41:37 pm
Both Michaels made great points here.
I agree with Michael Hsueh about Gapminder. It may not be the best visualization to display exact data--if we want accuracy and comparison there may be other visualizations that achieve this goal more effectively. However, there is something "natural" about representing time with time (just as we say last week when we represented text with text). I believe that concepts like speed and acceleration are easier to grasp when looking at an animation. After all, speed and acceleration metrics are both models that we have created to understand the world and made mathematical operations, but we definitely experience them in time. I believe that a viewer can easily create a mental model of the speed of growth in certain regions when looking at Gapminder. It conveys a story (although not a data-heavy one) and conveys it well.
I also found Michael Cohen's points about squash and stretch to be very interesting. I wonder whether they could also be used in visualizations like Gapminder to illustrate fast-moving countries.