Simulation visualisation I did over the weekend for my friend Rory Smith. He has some huge super-fancy simulation showing what happens to a dwarf galaxy as it falls into a cluster. It's a very complicated data set so this one's shown in stages.
First we see the stars as simple points. You can see they're in a sort of grid distribution, but this is just because their positions weren't output to high enough precision. They're rendered very crudely here because Blender has problems combining halo materials with transparent planes. I guess I could probably do something with z-buffers and composition, but time's a-wastin'.
Second we overlay the dense gas in the disc. There's a strong warp in the centre, not sure what's going on there. The peak density is more than a million times the lowest density, and the lowest-density material is found inside the cluster rather than originally in the disc. So it's useful to start by showing only the gas that originated in the disc, otherwise it gets very confusing. Then we see the velocity vectors in the gas, so you can see how it's rotating and being disturbed as it enters the hot gas in the cluster.
Then we overlay the thin gas from the cluster itself. You can see the bow-shock where the gas piles up against the galaxy, and a lot of very complicated, turbulent structures in the stripped wake. Finally we overlay the velocity vectors of this thin gas so you can see just how complicated these structures are (not sure how good the compression is on Google video).
This isn't totally maxing out FRELLED's capabilities, we could also colour the gas by temperature instead of density, do a larger cube or a time series etc. But it's quite a nice way to demonstrate many of the major features.
Is the dwarf galaxy falling into a bigger spiral galaxy or into a cluster of galaxies?
ReplyDeleteIt's falling into a cluster of galaxies. Not sure how much detail is included in the other galaxies.
ReplyDeleteHow's your experience been with Blender? I'm familiar with some of the dedicated scientific visualization packages like Paraview and VISIT, but I've never tried using Blender.
ReplyDeleteJonah Miller I'm biased because I started using it as a non-astronomical hobby. Of the free CGI tools available at the time, Blender was hands-down the best : the only one with a vaguely-sensible interface and enough features to do anything useful. I really like the variety of ways you can access different features : through GUI menus (easy to remember), through the keyboard (extremely fast when you learn them), and originally through mouse gestures (which were a nice balance between intuition and speed, but I understand why they removed them as they're somewhat unconventional). Used in combination, they are extremely powerful. But, by the time I started doing astronomy in Blender, I no longer had to think about the interface any more than I have to think about where the letters are on the keyboard. So it's tough for me to compare other tools
ReplyDeleteHeck, I have a paper on this, see in particular section 2.8 :
http://arxiv.org/pdf/1510.03589v1.pdf
Or the blog version :
http://astrorhysy.blogspot.cz/2014/08/hydrogen-dinosaurs-and-user-support.html
TLDR : I am a massive Blenderhead. :D
Rhys Taylor Your abstract says Blender FIFTY TIMES faster than other tools?!!
ReplyDeleteIf that performance advantage carries over to other forms of visualization, you may have just won a convert.
On the other hand, I see you're comparing to specialized tools, which may have less 3D rendering capability. So I'm not sure. It'd be interesting to compare to other tools.
Jonah Miller The 50x is in relation to cataloguing sources in a data cube, not visualisation per se. The reason the performance increase is so dramatic (it really is a factor of 50, no lie !) is because other tools utterly lack the ability to easily define and manipulate a 3D region, which for Blender is very, very basic. Once you've got a region defined, it then becomes trivial to do analysis on that region. If cataloguing sources in 3D is your thing, FRELLED is the tool for you. :)
ReplyDeleteIn terms of displaying the data, it's way slower than other tools by some orders of magnitude in terms of loading the file. Once loaded, however, it's similarly faster than other tools to display the data - at least as far as specialist astronomy software goes. How it compares to more generic viewers I've no idea. I got put off by the 30 page "quick" start guide of VisIt. None of them seemed to have the tools I needed, as far as I could tell.
See also :
http://frelled.wikia.com/wiki/FRELLED_Wikia
http://www.astroblend.com/
http://www.cv.nrao.edu/~bkent/blender/