Ten Steps to Imaging a Globular Cluster

John A McCubbin

The Ten Steps

Globular clusters are a great first object to try with a CCD camera. First of all, globulars are bright and fairly easy to find. They are also fairly easy to image, yet require mastry of all the basic processes in CCD imaging. Processing the image of a globular cluster is fairly straightforward, but very demonstrative of the power of image processing. It is nearly perfect to illustrate the "gamma stretch". Imaging a globular cluster also tests your ability to focus your telescope. Even minor errors in focusing will degrade the image quality noticably. So if you haven't mastered focusing your telescope on the chip of your CCD camera, you might want to review the article Focusing Your CCD Camera.

Step one: Locate the cluster you want to image. You should chose one with a fairly open architecture if possible. M15, for instance, has a very compact core and is more difficult to resolve a full range of star brightnesses. M13 is a very good one to start with, but M2, M3, M55, and M30 also good initial targets. If none of those are available, chose any globular that is up.

You should then center the cluster on your CCD chip. It wouldn't hurt to check the actual size of the cluster with a reference source just to make sure it will fit. (See article on Will it Fit?) M13, for example, is 16.6 arcminutes. A 12" f/10 telescope coupled with an ST-7 (chip size 6.9 x 4.6 mm) has a field of view of only 7.78 x 5.15 arcminutes. This would cover only the central nuclear area of the globular. Even with a f/6.6 focal reducer, the area of coverage would only be 11.73 x 7.82 mm. This would be a much better choice for framing the object, but if you have a f/3.3 widefield telecompresser, this would fully cover the object (it would introduce some distortion in the far periphery, however). Even if your scope will only cover the nucleus, center this on the CCD camera chip with your camera software in the "find and focus" mode.

Step two: Focus the camera (this is covered in the article linked above). It is critical that your focus be as perfect as possible. Any errors in focus will immediately be apparent. This is one of the reasons why globulars are good "training objects".

Step three: Set up your tracking CCD, if you have one. This must be done after focusing the camera. If you use an ST-4 and separate guide scope, then focus it in the same manner as the imaging CCD camera then select a guide star. Calibrate the tracking rates using your software. If you don't have a guide scope and guide manually, you should select your star at this point and get your bearings and prepare to guide. If you are using track and accumulate, you should probably use a 15 - 20 second integration time for each exposure that will be stacked, and make the total time add up to at least 300 seconds.

Step four: Take the image, download it, then save it to disk. My personal naming convention for the filename is a string that consists of the name of the object, followed by the integration time, followed by the temp of the exposure, finally followed by the series number if it is to be one of a multiple of the same object. An example is M13_600n20a.ST8. This shows me the object is M13, the integration time is 600 seconds, and it was taken at -20 degrees (the "n" is for negative, "p" would be positive), and is "a" in the position in the series (in this case first). That way it is easy to match up dark and flat frames with the filename. Just for reference I name dark frames similarly. Ex: DM600n20a is a dark frame at medium resolution taken at -20 degrees and is "a" in the series.

Carefully examine the object for tracking accuracy and focus. If your stars don't look round or are fat and out of focus, you will have to check polar alignment, guiding, or focus as needed. This is where the "rubber meets the road" in CCD imaging. It's unfortunately the greatest test of your equipment's quality and your ability to set it up. If you get consistent streaking, you may have to recheck polar alignment (later - this doesn't have to be perfect on the first try) or balance of the scope. You actually don't want to be in perfect balance you need to be slightly heavy to the east and then either north or south (it doesn't matter as long as there is a sligh imbalance).

Next, examine a histogram of the image. You want to aim for a high pixel saturation. If the maximum brightness represented in the image is only 4000, then you should expose for a longer period than 300 seconds. I aim for full saturation in the brightest stars. A 600 second integration with my AP180EDT and the ST-8 binned 2x2 does this perfectly for me. Experiment and find out what it takes to produce good image depth without blooming. Take further exposures as necessary till you get the image you want.

Step five: Take a dark frame of the same length as your best image. It must also be taken at the same operating temperature.

Step six: Take a flat frame. I take an unfocused image of a slide projector screen that resides in my observatory. When I did portable imaging, I took it with me then took an image of it in the field. Since flat frames are quick, you should take 4 or 8 then average them (see articles, How to Take a Good Dark and Flat Frame, and Image Stacking, How and Why). This will reduce noise in the image and allow you to bring out the faintest stars in the globular with gamma stretching.

If you are unable to take a flat frame, then the image will still be very good. Globulars clusters are so bright that the signal to noise ratio is very high to begin with. Not taking a flat frame will still produce a high quality image. You will only be sacrificing the dimmest stars. You should try to get a flat frame, if for no other reason, for the practice.

Step seven: Calibrate your image. Open your image processing software and open the raw image. Calibrate the image by first subtracting the dark frame. Second, divide by the sensitivity of the flat frame (MaxIm and CCDSoft do this for you by just selecting the file). Your particular processing software may do this differently. Now you have a calibrated unprocessed image.

Step eight: Stretch the image using a linear stretch. This prepares the image range for gamma stretching. To make the background black you should select a value that just barely removes the background brightness. I personally like black background with no levels of gray at all. Don't go overboard or you will actually process out the stars you can't see yet. Don't forget your 16 bit camera can see 64,000 levels of gray and your monitor and eye can only see around 64 levels of gray. There's information in the image you can't see yet. Once you set the background level, set the highest brightness level until you can see detail in the core of the globular. This is a personal choice. Because of the extreme levels of brightness the more detail in the core that you preserve sacrifices dim stars in the outer areas of the cluster. Experimentation is the key here. Try it both ways (just not on your original image).

Step nine: Gamma stretch the image. This is a key feature of higher end image processing software. In MaxIm, you are allowed fine control over the exact mathematical options of this stretch. A gamma stretch is a nonlinear stretch that emphasizes the dimmer or brighter portions of the image. In MaxIm, a gamma of <1 emphasizes the low levels of brightness without changing the higher ones. A gamma of >1 does the reverse. If you have Photoshop, and aren't using scientific software, or your software doesn't have a gamma stretch routine, you are still in buisness. I will illustrate how to gamma stretch in MaxIm and Photoshop.

In MaxIm, you select Process > Stretch. You are then presented with a dialogue box like the one at the left. You should select gamma as the stretch type with a gamma value less than one. I started with a gamma of 0.5 (actually a fairly agressive number, but I knew that I had good image depth and this is where it pays off ! ). You should base the stretch on the "screen stretch", which are the values you selected in step eight. Therefore you only apply the gamma stretch to the values you select, not what the software selects. Put another way, you select both the dimmest and brightest values upon which to base the gamma stretch. Leave the output in 16 bit format.

If you have the screen stretch window visible, you can play with the "pregamma stretch" and the preview button in the dialogue box to test your gamma settings and outcomes.

When you are satisfied, click okay and see how you did. After all there's always the undo button to reverse your decision.

If you want to finish the image in Photoshop (I use v5.5, previous versions do this almost the same way), you must first save the linearly stretched image from your imaging software to disk in a 16 bit format as a *.tif file. You can save it in an 8 bit format, but you will lose a little information. Once the image is saved as a *.tif, open it in Photoshop and chose Image > Adjust > Levels. You will be presented with a histogram graph with three sliders below it. The left one represents the "floor" or dark level. Adjust it to darken the background. The right slider button controls highlights. Slide it to empahsize the levels of upper brightness. The center slider button controls gamma of the image (ahhh, were all software packages this easy). Slide it to the left to achieve a gamma value less than one and emphasize the lower brightness levels. Sliding it to the right does just the opposite. Once you have achieved your perfect result. Save the image under another name. You will want to convert it to 8 bit format for display on the web or posting on a newsgroup.

Step ten: Apply a light unsharp mask. If your stars are a little less than pinpoint you can apply a light unsharp mask. This can be done in almost any software package. I sometimes do this in MaxIm, sometimes in Photoshop (it has a very controllable unsharp mask routine with immediate feedback, but you can only access this routine in 8 bit mode).

Examples of the Process: Object M2, Globular in Aquarius

		The calibrated unprocessed image
		Linear stretched to select intial brightness range
		After 0.5 gamma stretch
		Light unsharp mask

Now go try it for yourself...

Back to Resources Page

I consider your heavens, the work of your fingers, the moon and the stars, which you have set in place ... Psalms 8:3