.glass

Readers should by now be aware that I work with 3-dimensional data and have a minor obsession with building a better FITS viewer. 3D rendering on a screen is all very well, but electronic data is sooo impersonal. Whatever happened to that bygone era when astronomers looked through telescopes and had no other option but to draw what they saw, or write poetry about it ?

Err, well, nothing, actually. Anyway my point, inasmuch as I ever have one, is that sometimes it's nice to have something good and solid that you can point to and, if necessary, bludgeon people with. Because laser pointers can only blind them in one eye at once. Or, if they're not trying to steal your research, you might just want to give them something to eat, and astronomical data may be many things, but no-one would ever describe it as nourishing.

For preference, I'd like all my data printed in .cake format. This would be something like a Battenburg cake (but with a better colour scheme). It'd be made up of hundreds and hundreds of little cubes of cake, each of just the right colour, so that the whole thing would be a physical model of an electronic data cube. As you slice through it, searching for galaxies, you'd also get a tasty snack. Of course, the resolution of the .cake format couldn't be much better than about 1cm per pixel. Which means that for a decent-sized data set, the cake would have to be about 7 metres long and 1.5 metres tall.

That's about right.

This of course is just idle fantasy. For now. But other approaches are yielding results. The .glass format offers the prospect of better data archiving (not being able to present your work because you ate it isn't normally a valid excuse), better resolution, and is much more shiny. Plus, when asked "what's a data cube ?" I can now just say "there's one on my shelf."

This cube is of HI data of the Virgo Cluster, which took 4 years of my life to analyse. Each blob is the gas in a distant galaxy. It measures approximately 14.5 x 14.5 x 6.5 cm. It looks somewhat better in reality, especially when properly illuminated. It doesn't photograph very well though, and owing to a lack of black surfaces I had to use a coat for a background. Here's what the data looks like electronically :

To create the physical cube, I converted the original FITS cube into a text file of x,y,z,v format using an IDL script I'd already written. Then I sent the data off to Bathsheba Sculpture. I stumbled on them having searched for custom glass etching companies. I couldn't find any that would just allow me to submit a model or text file, but I saw the astronomy section of the site and sent them an email - reasoning that since they'd obviously had prior dealing with astronomers, so were less likely to be thrown into a state of bewilderment.

Indeed they weren't. It took a couple of months to go from a state of raw data into a physical cube sitting on my desk, but this isn't very long considering the length of time I spent looking at this thing in its original format. Now all I need to do is find an illuminated base so that it's easier to see in daylight.