Jupiter Lollipop
Using data from the Hubble Space Telescope, scientists have produced groovy new maps of Jupiter and projected them onto a rotating globe.
Credit:
NASA, ESA, A. Simon (GSFC), M. Wong (UC Berkeley), G. Orton (JPL-Caltech), and G. Bacon (STScI)
Watch:
http://www.spacetelescope.org/videos/heic1522a/
#nasa #esa #jupiter #space #hubble
At first I was thinking why not do this with images from Cassini and get a more detailed map? But of course you can't stitch together a series of close up images with changing atmospheric features and have things visually fit together well. The Hubble images would also need to be combined side-reel but at their resolution and a 10 hour day cycle, the cloud eddy shapes would still fit together reasonably.
ReplyDeleteSean Walker Couldn't model-based approaches achieve a better job by perhaps not even skip rotations but interpolating all missing frames realistically?... or to sound leading edge, deep-train an AI artist to fill in the blanks?
ReplyDeleteBoris Borcic I missed this comment - sorry. Model is of course a stretchy word, but usually means a method of designing an independent symbolic/mathematical/ algorithmic construct that parallels the thing being modeled. I'm sure you could do this with Jupiter's surface but I think a simpler technique would simply be use something like inter-frame morphing to interpolate. I guess you could call it a model too - but it's not really a parallel of the subject system in the sense that models usually are.
ReplyDeletehttps://en.wikipedia.org/wiki/Motion_interpolation
As a side note, wavelet processing is often used to derive a nice interpolated images from multiple samples. This wouldn't help though with knitting together discontinuities that arise when you come back to where you started with with a non-static subject.
https://en.wikipedia.org/wiki/Wavelet
Sean Walker However Moore's law translates to continuous progress of natural phenomena simulations, sensitivity to initial conditions does limit, whatever, the 'asymptotic' expectation that Laplace famously answered to Bonaparte (when asked about a supreme being) -- "I just need an initial condition". It took me a long time to extirpate myself from the notion of physics models going to fully jail reality into their predictions, and effect a shift from models-as-shells to models-as-tuning-forks (to summarize in metaphors).
ReplyDeleteOTOH, what we are discussing here is not to jail reality in predictions, but to smoothly fill in blanks in captured data -- to make an interpolation whose result won't have to pass any other test than viewer satisfaction with the realism.
Boris Borcic Laplace's most famous occasion of the principle of Occam's razor suites well the purpose of interpolating between frames. The energy and creative use of any number of free variables serves no purpose if in the end a simpler non-model produces the required result from initial conditions. ;)
ReplyDeleteBut now I wonder if the more interesting parallel isn't the one between this discussion and another we've had. More, I wonder Boris Borcic my friend, if the parallel wasn't an intended, if subtle, weaving. =D
Sean Walker At first blush I don't see what discussion you're referring to. What happened here is that I had a contention which on second look wasn't justified, that the choppiness of the animation was due to it assembling samples that were over or near a full rotation apart, what would require some modelling to "anti-alias".
ReplyDeleteI was referring playfully Boris Borcic to your allusion to the Laplace retort about a supreme being with Bonaparte and our conversation a couple weeks back on a similar subject. I saw an irony in the implied application of Occam's razor principle. My comment wasn't meant as a jibe and I trust you wont take it that way.
ReplyDeleteSean Walker I think you refer to the linked conversation, but I am not sure you've read my last reply (including its element of... perhaps rectification).
ReplyDeletehttps://plus.google.com/u/0/+BorisBorcic/posts/6hEAZpiTiPB
Muito bom!
ReplyDelete