The Virgo Cluster
Galaxies aren't found scattered randomly in space. Most of them live in groups that can contain anything from a few to a few thousand members. The Virgo Cluster is the nearest large galaxy cluster (generally reckoned to contain ~1,000 members or more), so is an important and very well-studied region. I decided to make a map of the major galaxies using data from the uniquely detailed Virgo Cluster Catalogue via the GOLDMine database.
The images of each galaxy come from the Sloan Digital Sky Survey. The colours are not what you would see with your eye, but they accurately show differences between galaxies (things which look redder really are redder, but the difference might not appear so pronounced if you looked through a telescope). A lot of processing had to be done to clean things up.
First, I only selected galaxies which have redshift determinations (telling us how fast they're moving away from us - which is ROUGHLY equivalent to distance, so we can tell if they're really in the cluster or actually much further away). That restricted the sample to 781 galaxies. This map is a few years old so their are probably more measurements available by now.
Secondly, I manually downloaded the images of the 781 galaxies, and removed foreground stars and other unplesant artifacts. This was done in a very unsophisticated, photoshop-esque way, the goal being to create a pretty picture.
Thirdly, I wrote a little Python script to import the images into Blender and place them at their correct position and scale them to the correct size. The sizes of each galaxy were then exaggerated by a factor 5 to make them clearly visible (you can see them at their correct size here.
The main reason I only used galaxies with redshift measurements was because it's a rough approximation for distance - not a very good one in cluster, unfortunately, but the best we've got. That means the map is actually in 3D, which is of course what Blender's good for. This video was shown at the summer 2011 AAS meeting in Boston. Although velocity is not really the same as distance, it's good enough to show major sub-structures within the cluster.
Spiral galaxies are circular, so it would be possible to correct for how they're inclined towards or away from us. This is more complicated, however, so I didn't try to do this. You may notice that the spirals sometimes disappear in the video, since their images are just shown as thin planes which don't rotate as the camera moves. "Elliptical" galaxies, which are basically just big balls of stars, are shown as multiple different planes pointing in different directions (so they never disappear).
So the map is a reasonable depiction of the major galaxies in the cluster. The relative sizes, brigthness and colours of the galaxies are all correct. The faintest galaxies are missing (because they don't have redshift measurements), but that's OK because they probably wouldn't be visible anyway. The video is accurate on a large scale, but all of the details are wrong - redshift isn't good enough to tell us if individual galaxies (which may be close together on the sky) are really close together in true 3D space.
What I would like to do is extend this to other wavelengths - very hot gas that can only be seen in X-rays fills much of the cluster, and long streams of atomic hydrogen have also been detected. One day, maybe.
You can download the file used to make the images here, but please also see this page for more details.