At this point I can't avoid talking about my research a little bit. For the last 5 years I've been studying the Virgo Cluster, the nearest rich cluster of galaxies to us at about 17 Mpc distance (or 55 million light years if you insist). Since it's close, it's been intensely scruitinised in the optical wavelengths. The principle catalogue - known simply as the VIrgo Cluster Catalogue or VCC - is available in handy electronic form on the GOLDMine database.
By writing a simple Python script, I was able to chart the positions of each galaxy in Blender. Thanks to the wonderful Sloan Digital Sky Survey, RGB images were available of nearly object, so I was able to map the cluster. This involved a lot a blood, sweat and tears. Unfortunately Blender's Python documentation is great if you already know how it works, and only marginally useful if you don't. So, unable to figure out how to automatically assign textures, I UV mapped each RGB image to each galaxy by hand. All 781 of them.
You can see the resulting mab below. The size of each galaxy is scaled by its major optical axis listed in GOLDMine, though I exaggerated this (by a factor 5) to display them more clearly. Click the image to show it in its full, glorious resolution. Or click here to see an unadulterated image with no size exaggeration.
Of course, Blender is a 3D tool and not designed for 2D illustrations. However, we have some knowledge of the 3D shape of the cluster thanks to redshift measurements. These tell us the recessional velocity at which each galaxy is receeding, which turns out to be a rough proxy for distance. Actually, in galaxy clusters it's only a very rough proxy indeed, but it's still enough to tell that certain groups of galaxies within the cluster are distributed differently to others.
This is why the map is restricted to a mere 781 objects, when the VCC contains about 2,000 entries. Many of them have velocities much too high to be cluster members, and many others are so faint that measuring a velocity is just too difficult. But for the remaining 781 objects it's possible to show them in 3D space. This video was shown at the summer 2011 AAS meeting in Boston.
Many real galaxies are discs which are inclined at some angle to us. It's possible to correct for this, but I didn't attempt to do so. This means that spiral galaxies tend to disappear as the camera flies around them as they're just simple planes. Elliptical galaxies are believed to be essentially giant balls of stars, so I used the traditional billboarding technique with these.
A handy feature of UV mapping in Blender is that it allows the textures to show up in the realtime view, which is a lot of fun in combination with flight mode. However, realtime transparency is not handled correctly, so the above views are renders.
A final key feature that should be mentioned is that the SDSS images have in some cases been heavily processed (again, by me, with a toothbrush) to remove foreground stars and the like. So, although this view uses real data, don't think of it as in any way a true picture of the cluster. It's better to think of it as "broadly correct."