July 2006 Archives

Yeesh, it's been almost two weeks since I've posted here. Haven't done that since I don't know when.

Basically it's been my two week window between finishing teaching and leaving for a bit of vacation time. This does not add up to a great deal of time in which to get stuff done. Mostly I've been futzing with computers in preparation for getting interesting things done later. With occasional breaks for being horrified by the world burning outside.

Sadly, I have little of interest to say about any of this. (Not even about Kinky Friedman and the Weirdest Governor's Race Ever, although my whole family's been rooting for him for months -- I should try to come back with some bumper stickers or the like.) That is a sure sign that I need a change of scenery. I leave Saturday morning, and may or may not post here during the two weeks that follow. My attempts to photoblog my life in real time never pan out, mostly because I'm too picky about processing the photos I post, but I should have nice photos from the Tetons when I get back.

Popped out to watch B.I.K.E. at the Bicycle Film Festival this evening, which was ridiculous and self-absorbed, but featured much tallbike-jousting, a working jetbike, and brief appearances by a handful of Chicago bike freaks I've met, so it gets a pass in my book. The preceeding short was more vicerally entertaining, though, consisting mainly of a death-defying bicycle messenger race across Manhattan from a helmetcam's perspective.

Now, a little more astronomy. Despite dramatically improving the signal-to-noise ratio of my Jupiter pictures, the composite photo I posted yesterday is still dramatically more blurry than one might expect from a high quality 10-inch instrument. This brings us to a concept called "seeing". As the animation below demonstrates, from one instant to the next an image will distort in numerous different ways. This is caused by turbulence in the atmosphere, and is similar to the phenomenon that causes stars to twinkle. However, most of this turbulence is much closer to us than that high-altitude effect; the ripples visible here are primarily the result of hot air rising off the building's roof and mixing with the air streaming through the observatory dome opening. In spring or autumn the seeing from the 10-inch improves considerably, as the days are cooler and the sun sets earlier, and thus the temperature gradients involved are less extreme. In any research-grade observatory, steps would be taken to ensure that the temperature inside the telescope enclosure is equalized with the outside before observations begin.

There's no getting rid of the higher-level turbulence, though, which is why the Hubble telescope produces such wonderfully crisp images using a comparatively puny aperture by current standards. That's changing, though. Since these are quick snapshots it's easy to see that each one is a reasonably crisp image and basically identical with some random pattern of squashing and stretching applied; it's only when they are combined together that you'd get a blurrier image (and for that matter, if I had a more light-sensitive camera I could get even sharper images by taking even shorter exposures). But you could imagine, say, choosing fixed landmarks on the Moon and designing an algorithm that would distort each image until the landmarks lined up every time, essentially un-doing the distortion. Most astronomy, however, is concerned with details much smaller than the distance between landmarks on the Moon. It turns out you can do something similar, though, by distorting an image until a star appears as close as possibe to a round point. This is called adaptive optics, because it isn't done with images after the fact, but in real time using a computer to distort a flexible mirror in just the right way to un-blur the image.

Now if only they could get it to work with visible light.

Sequence of six 1/15 second exposures of the northern Lunar highlands, combined as an animation to highlight varying atmospheric distortion. The images are roughly centered on the broad Clavius Crater; the famous young Tycho Crater is the smaller, deep crater with prominent central massif to the lower right. 2006:07:05 21:37:35

Single 1/15 second exposure of Jupiter through the 10-inch refracting telescope, scaled to 1/4 original size. 2006:07:05 21:49:28

Picking up where we left off yesterday, recall that I had a bunch of shots of Jupiter like the one shown here or a few days ago. The major problem with taking such photos is that, due to the magnification and quantity of light involved, you typically are faced with choosing between a long, blurry exposure or a short, grainy one. If I had a decent mount, better seeing, and an astronomical CCD this wouldn't be such a problem, but with these consumer camera-based solutions, that's the tradeoff.

For something like a planetary image, there's a way around that. Most cameras of this sort have a rapid-fire mode, which I used to take a series of 15 1/15 second exposures like the one at left. Now each one of those is dark and pretty grainy, but crucially, each one is a picture of the same planet, only with some different random-ish pixel noise and offset by random amounts due to jitter. Offsets are easy to take out by hand, although you could also write an image correlation algorithm to do the work if you want to do this a lot. Then it's just a matter of combining all those images in such a way that the noise averages out, leaving a clearer picture of the planet.

For now let's continue working in the Gimp. I can load up each image as a fresh layer in a single file. To line them up, what I find works best is to set one layer to divide the one below it; slewing the top layer around until the result is as close to 1 (white) as possible does the trick very sensitively. This would only work for a large, bright object like a planet, however.

The Gimp doesn't have a concept of averaging many layers, but you can fake it with transparency blending. The average of many layers is an image where each layer contributes equally. So from the bottom upwards, I set the opacity to the series, 1, 1/2, 1/3, 1/4, and so on, so at each level the resulting image is an equal combnation of those lower in the stack. The final result has enough color information that it's actually worthwhile to play with the transfer function to enhance Jupiter's banding, and the final result is below. Whee!

Stack of 15 1/15-second exposures of Jupiter through the 10-inch refractor, color enhanced to emphasize the cloud bands. There is a moon just coming out of eclipse at lower right, but it is totally invisible with this transfer function. I could alternately have added the layers to build up enough photons to clearly show Europa, at the expense of washing out the planet. There is a hint of a dark spot on Jupiter's lower left limb, which is right where Europa's shadow ought to be.

Took the whole weekend off, more or less. Worked a bit on the garden, lit lots of fireworks -- apparently I'm already slated to be pyrotechnics man for next year's July 4th party -- got in a couple of longish bike rides. Graded lots of papers, which was a little bit like work. Fed my roommate's fish.

Tonight I took more of my students up to the roof to look through the telescopes. I'm finding that, while the optics in our little 8-inch scopes are decidedly superior to those in the 10-inch refractor, the massive refractor is a much more stable platform for astrophotography. This week I'm experimenting with image stacking. So you get more of that. Combining two images is easy, even with something like the Gimp, simply using layers and opacities. At low magnification and high light levels you don't gain a whole lot, though, so consider this a proof-of-concept.

Starting with an easy case, the average of two 1/24 sec exposures of the Moon. Taken tonight through the 3-ish-inch finder scope on the refractor. 2006:07:05 21:35:06

Yesterday I took a field trip with one of my roommates to Wisconsin. Now my living room table is piled high with "party favors" for Clay's July 4th party. Below, another photo from my Thursday night observing with my students.

In other news: dude, the Blue Man Group is touring again! The set of musical acts I'd cross state lines to see is exceedingly small, but if they don't announce a Twin Cities date soon I'll start planning a fall field trip to Milwaukee. With a billing like The How To Be A Megastar Tour 2.0 I'm gathering they plan to continue the theme of deconstructing the arena rock concert form.

Lightly enhanced 1/19 second exposure taken through the 10-inch refractor. Jupiter is at center, obviously. If you squint you can just make out the Great Red Spot just high of center. At lower left is Ganymede (I initially mis-identified this as Io, but Io had just gone into eclipse at lower right and I forgot which way the image flips due to the corner prism), and in the top left corner is field star Tycho 5575-473-1. 2006:06:29 23:04:48