My poster for EAS 2024. It should work publicly now, though the official poster gallery doesn't seem to be up yet. Anyway, this continues my efforts to promote Blender-based data visualisation in astronomy. Next week I'll recycle and update a talk covering actual science.
#Astronomy
#Blender
#DataVisualisation
#3D
#VR
https://k-poster.kuoni-congress.info/eas-2024/poster/803490a3-04a4-46ca-9094-22a296467039
Couple of papers on doing something a bit more unusual with AI and suchlike in astronomy. The first attempts to construct a knowledge graph to show how different concepts are used in different fields and how this changes over time. It certainly looks very nice but I'm not really sure what the point of it is. The second is more a controversial look at whether machine learning is useful and what makes for a good theory : it's got some useful points, but by and large, I would hope that most of its declarations are obvious to anyone actually trying to do anything. Then again, it's always good to derail a hype train, I guess.
#Science
#Astronomy
#Philosophy
#DataVisualisation
https://llittlephysicists.blogspot.com/2024/06/the-data-wont-learn-from-itself.html
Finally submitted a paper which has been basically ready to go since February :
A FITS viewer providing multiple techniques for astronomical data visualisation in Blender
I present a new version of FRELLED, the FITS Realtime Explorer of Low Latency in Every Dimension. This is a 3D data visualisation package for the popular ‘Blender’ art software, designed to allow inspection of astronomical volumetric data sets (primarily, but not exclusively, radio wavelength data cubes) in real time using a variety of visualisation techniques. The suite of Python scripts that comprise FRELLED have been almost completely recoded and many new ones added, bringing FRELLED’s operating environment from Blender version 2.49 to 2.79. Principle new features include : an enormously simplified installation procedure, a more modular graphical appearance that takes advantage of Blender 2.79’s improved interface, much faster loading of FITS data, support for larger data sets, options to show the data as height maps in 2D mode or isosurfaces in 3D mode, utilisation of standard astropy and other Python modules to support a greater range of FITS files (with a particular emphasis on higher-frequency radio data such as from ALMA), and the capability of exporting the data to Blender 2.9+ which supports stereoscopic 3D displays in virtual reality headsets. In addition, in-built help files are accessible from each menu panel, as well as direct links to a complete wiki and set of video tutorials. Finally, the code itself is much more modular, allowing easier maintainability and, over the longer term, a far easier prospect of migrating to more recent versions of Blender.
The main hold-ups have been waiting for feedback from colleagues and, to a less extent, finding a journal. I’ve given up prompting and decided to the caution to the wind regarding the former; it’s a single-author paper, and such comments as I did get were along the lines of “seems fine to me”.
The latter is… uglier. The original version of the code was published in Astronomy & Computing, and in the halcyon days of 2015 this was free. Of course, every journal under the sun has switched to “open” (open ! HAH !) access, which mains that would now cost $2900. We can’t afford that - it’s simply a silly price. Which is a shame because A&C is an ideal journal for this kind of thing. So now I’ve tried A&A, which is more geared towards science papers rather than code, but has the tremendous virtue of being free because we subscribe to it. So, we’ll see - not sure this is the right journal, so this might be an awkward refereeing process.
(The code, of course, is here : http://www.rhysy.net/Code/Software/FRELLED/)
This demo video gets its own post because the effort gone into this project has been frankly outlandish.
I finally resume updating my website. This new section collects a lot of different data visualisation projects in one convenient location. Come explore the gas in the Milky Way, the Virgo cluster, and the entire nearby Universe... and then take a walk along the Arecibo telescope in virtual reality and see the sky at radio wavelengths across the world.
Probably needs a good check for typos but that can wait.
Training session I gave last Friday on viewing 3D FITS files in #Blender. I must have really caught most of the bugs by now since nothing went wrong during the live demos...
Preparing a data visualisation training program and having some fun with the demonstration files. This is an ALMA PHANGS map of the CO clouds of the galaxy NGC 2903. Basically clumps of the densest molecular gas where star formation happens, shown as a volume render and with contours at 5 sigma (4 sigma gives a more interesting result, but that creates a 12 million vertex file which is a right bugger to work with).
More website updates. A smaller one this time, looking at how we turn radio #astronomy data into visually meaningful images.
I stole some of my increasingly depleted time to make this little animation of the hydrogen contours of M33. Well, isosurfaces to be more accurate. The third dimensions is velocity, not distance. I've done these before, but now I extended it to show the surfaces at different flux levels, which is quite fun.
(More here : https://llittlephysicists.blogspot.com/2019/11/wibbly-wobbly-spacey-wacey.html)
