Monday, June 25, 2007

Pet Peeve #115

People who use "three-dimensional" graphs. Paper is two dimensional. There is no way to represent a general black-and-white three-dimensional object using a single black-and-white two-dimensional figure. Even if you use Matlab. Even if you use wireframes or shading. Even if you really, really want to.



You are encouraged to use color to create an artificial third dimension. Back when I was writing real-time spike analysis software for electrophysiology recordings (read: stick wires into rat brains; try to extract the signals of individual neurons from the muck), I tried using the red and green color channels to represent the two dimensions of our four-channel tetrodes which wouldn't fit on the (by definition) two-dimensional scatterplots. Note that the human eye contains three (3) distinct color receptors, so this strategy unambiguously encodes the desired information, although it can be a bit difficult to decipher visually. I thought it worked pretty well. My adviser thought it looked "unprofessional"; i.e. unlike the stupid commercial system we were sacrificing a year of our lives to replace. Meh.

If you're some sort of crazy chemist (e.g. Gylbert, 1973 or Hackert and Jackobson, 1971) and want to use a stereogram, that's pretty swanky, but you'll need to depend upon the ability of your audience to freely rotate their eyeballs in their sockets. Maybe the format is standardized and they give out stereoscopes at chemistry conferences they way they used to give out those horrible red/blue cellophane glasses for 3-D movies. In any case, a single two-dimensional view of a three-dimensional object is necessarily ambiguous. Computers let you do all sorts of things you really ought to avoid. Powerpoint presentations, for instance. The 80's are dead. Get over it.

1 comment:

Anonymous said...

As someone who has spent hours practicing the free rotation of my eyeballs in order to glimpse three of the four dimensions of a hypercube, I can attest that these stereograms are awesome. The first one looks like a flat mesh in 2D, but in 3D it's all over the place. It also makes a great party trick to stare at Julesz's random dot stereograms in books and be able to say, "I see the square in the middle!" Unfortunately for me, the hypercube is based on crossing the eyes, whereas these pictures are based on separating the eyes, so front and back have been reversed. If you could just switch the left and right pictures, I'd be much happier.