Thursday, July 06, 2006

Astronomical Data 2d. - Astrophotography

In this post, I’m going to take a brief deviation from the strict science behind astronomical data acquisition and discuss a related topic: Making pretty pictures.

Even if you’re not into astronomy, you’ve probably seen some stunning pictures like the pillars of creation. But what you probably haven’t stopped to consider is what was involved from taking those images to having an image worthy of display. So in this post, I’ll be giving a brief introduction to the art of astrophotography.

The first step is, predictably, choosing an object to photograph. The Messier catalogue has large number of objects that are favorites.

So once a target is selected we just have to find it. There’s various ways to go about doing this, from star hopping, to using coordinate systems, or just plain making the telescope figure it out in the cases of some newer ones.

Now that we have an object in the view of the telescope, it’s time to take a picture! Hooray! So we grab our digital cameras, hold it up to the eyepiece and snap a picture.

Chances are, if you get anything, it will either look like a big blur or streaked.

Nuts. So what went wrong? One of the first problems is focusing the image from the telescope onto the film or CCD. For amateurs, there are a number of devices specially designed to make the process simple. In professional observatories, the instrumentation packages are designed in such a way that the image should already be pretty close to focused.

The next problem is getting the camera to hold still. This is the reason that holding a camera up to the eyepiece of a telescope is extremely difficult. Chances are you don’t have anywhere near steady enough hands to do this. I know I don’t and I have no problem moving my mouse to pixel perfection. To fix this problem, the camera needs to be attached to the telescope and the shutter is triggered by remote.

Again, there are a number of methods available. Most SLR cameras on the market have adaptors built that are relatively inexpensive. For these cameras a cable release is also used. With astronomical CCD cameras, they’re built to go right on the telescope, and a computer program triggers the “shutter.”

So now we have the camera set up to expose. But how long do we leave the shutter open for?

This is entirely dependant on the object you’re trying to photograph, the equipment, and the intent. For very bright objects like the sun (with a proper filter) a 1/64 exposure is more than long enough. For the full moon, about twice that is necessary. For a thin crescent moon, good exposure times may be closer to 1/4 second. Planets will be in the range of a few seconds on film.

However, for the faint nebulae and the really fun deep sky objects, exposures of minutes to hours are required.

This introduces a new problem: tracking. Since the Earth is turning, it seems like the sky doesn’t hold still and spins overhead. If you have a telescope that doesn’t track that motion, the images will turn out streaked! If done intentionally, this can be a very cool effect:


However, if your goal is to get sharp images, the telescope must be able to follow these motions.

So assuming everything is done correctly, you should now have a nice astronomical image! However, if you’ve used a CCD, it’s in black and white since CCD’s don’t do color.

“But wait!” you’re thinking. “My digital camera does color and that’s a CCD, right?”

Yes, it is! But there’s a trick. Each pixel on the CCD is black and white, but sensitive only to a certain color: red, blue, or green. They’re arranged in a pattern like this:


Each one is very small, and the computer then averages each pixel with its neighbors to determine what color there is.

Not a bad trick. So why don’t those silly astronomers get a clue and do this too? In scientific astronomy the reason is because we don’t care about color images. In fact, we frequently only want to look at light from just one wavelength. But that will all be covered in my next post.

For amateur CCDs that are only going to be doing photography, many are color in this manner. However, it’s cheaper to have CCDs that are black and white so many amateur cameras are still black and white.

So how do astronomers get those color images out of a black and white image?

They go into Photoshop and get out the paint tools of course.

No. I kid. Astronomers use a trick similar to the one for color cameras. However, instead of taking one image, they take three or four. One will be taken through a red filter, another through green, and another through blue. A fourth one is frequently used with no filter to get luminance information across the whole spectrum.

Since things outside our solar system don’t generally have any perceptible motion over a few nights, we can get away with taking these images in different filters at different times. Each image is then recombined digitally to produce a color image (after the calibrations to remove noise that I talked about in my last post on this topic). There are several programs out there to do this. My personal favorite is Photoshop with the FITS Liberator plug in available from the ESA (astronomical images are stored in .fits format which is uncompressed and also has another file embedded that stores information about exposure time, location image was taken, and many other things, as opposed to .jpg format which inherently has compression artifacts).

The trick here is that the person processing the images can do a lot of customization from here. I’m partial to red nebulae, so when I do this sort of processing, I tend to put more emphasis on the red filter. When doing star clusters, I think they look prettier sort of bluish.

Thus, there’s a lot of customization involved in processing astronomical images. Two people can be given the same set of images can end up with very different results. Ultimately, this means that no astronomical image you see is “true color”. That would include those taken with “color cameras” since they are more sensitive to certain colors of light than others which lends an inherent favor to certain colors (generally red).

Is it possible to generate a truly “true color” image. Sure. If images are corrected for the biases proffered by the CCD, and not given any preference by the person doing the processing, it’s possible. However, such images are very boring to look at. The images generally come out very grey. So in the end, having a little bit of artistic license is a good thing.

This wraps up part two of my series of mini-essays on astronomical data. In part three, I’ll be looking at a number of ways we can actually extract information from these images. Assuredly, they’ll be more technical so before I get there, take a breather and check out some astrophotography websites like these.

No comments: