On these pages I describe the specific techniques of data visualisation I've developed. This is mostly about how I load 3D astronomical FITS files into Blender, but I also cover some of the more technical processes of how best to process radio astronomy data to better reveal different features.
Radio astronomy data visualisation is still very much a niche field in the pejorative sense. Understandably so, because after all, what we're ultimately paid for is not pretty pictures but scientific advancements. But I firmly believe that the two are intertwined, that the more deeply we think about how to process the data to make it visually appealing, the more we also think about how best to extract different scientific value for the data. Case in point : only after about a decade did I realise that my very first radio observations could have new information extracted from them by the extraordinarily simple method of plotting them as contours.
And why not extract both artistic, aesthetically appealing imagery from our data purely for its own sake ? Sometimes the data is a bunch of crap (no, really, it is), but sometimes it's beautiful. Explaining the vagaries of galaxy evolution to the general public is difficult, and to avoid using something which is immediately appreciable is a mistake. Data visualisation, then, is of enormous threefold importance : in getting us to understand the scientific implications of the data, in creating artistically valuable content, and in engaging the public. But the fourth, most important reason of all is philosophical : that it gets us to think on all levels about what the data really means. Data visualisation, in short, is something everyone should do, because it's good for the soul.
Data cleaning : Methods of cleaning the data to improve the S/N. I look at why these effects are more subtle than just measuring the rms, with lots of visual examples of how much better the final data product can become.
Basic FITS in Blender : My very early experiments with how to load astronomical FITS files into Blender, what works and what doesn't. Interesting for abstract art, but for far more practical solutions, see below.
Non-Cartesian Data : A couple of techniques for dealing with data cubes that aren't in Cartesian coordinates. The first is clever way of UV-wrapping the data in Blender and transforming it back to its proper geometry. The second is a bit of a cop-out, but more useful, describing a way of converting the data to Cartesian without any ugly gaps.
N-Body Simulations : Some simple techniques for loading n-body simulation data inside Blender, avoiding the need for Blender's own particle systems, and best practises for making the data look pretty.