She Blinded Me with Eye Candy
The winner of the 2006 Science and Engineering Visualization Challenge (organized by Science and the National Science Foundation, NSF) shows "five well-known mathematical surfaces, rendered as glass objects in a highly realistic 'Still Life.'" Using reflection, colored lighting, and otherwise unstructured sufaces makes for an image that does not convey the actual shapes particularly well. But it sure is pretty.
The image is unimpressive to me for two reasons: rendering and shape representation. It is rendered using a raytracer (similar to the venerable POV-Ray), and these have been around since the 1980s. They have been improved, yes, but rendering glass-like surfaces is nothing new. Mathematical shapes have also been used to describe scenes more easily, and many raytracers allow you to define objects using fairly complex mechanisms. Renderings of shiny objects based on math have been used on magazine covers for 20 years. This just cannot count as innovative anymore in 2006.
There are also better ways of showing shapes than a rendering like that. Work has been published in recent years, specifically at the IEEE Visualization and Information Visualization conferences, that addresses the question of how to best use surface textures for shape perception. Notable are at least two pieces of work here, both based on solid science and experiments. Victoria Interrante worked with several students on texture orientation for shape perception and layered surfaces of nested objects. Alethea Bair is working with Donald House and Colin Ware to find textures for layered surfaces. Those are not nearly as beautiful, but they are perceptually effective at conveying complex surface shapes to the viewer (and they could undoubtedly be improved from an aesthetic point of view without destroying their effectiveness).
Now I may be wrong about the intentions of the contest. The contest web page quotes Felice Frankel, a science photographer who is highly respected (and for good reason), and who is working at Harvard and MIT. She is described as being "impressed by the image's ability to engage viewers and trigger their curiosity." Now there is nothing wrong with triggering viewers' curiosity, but that can't be all. This works for images that are put on magazine covers, those are meant to attract the viewers'/buyers' interest. And because of that, these images often are almost pure eye candy, although based on real science.
But looking at the evaluation criteria for the challenge, entries are supposed to have "scientific significance, freshness and originality, as well as have an aesthetically pleasing composition and drama." Under visual impact, the first sentence requires entries to "successfully [convey] the research to its intended audience enabling new scientific insight." So this is not (just) about marketing, but about communication – and visualization. Being able to read the image and gain insight from it is just as important as being visually compelling.
I understand the need for pretty images to make people interested in science. But only looking at the shallow prettiness rather than more scientific depth cannot be the answer. Science images must be richer than the cover photo for a fashion magazine. Science and NSF need to apply higher standards, and look beyond the superficial appearance.
Posted by Robert Kosara on October 8, 2006. Filed under criticism.