Lunar Imaging Basics - Part 1
In March, I managed to capture a fairly nice mosaic of the moon and a couple of people suggested that I share some of the details of mosaic imaging in a Night Times article. However, as I started looking at the process of producing mosaic images, I realized that much of the technique required some common ground regarding the process of capturing individual images of the moon. So I'm breaking the mosaic topic into two introductory articles and a subsequent mosaic article.
Taking images of the moon is one of the easier and more satisfying astrophotography processes if done right. It's probably the best place to start astrophotography. Yet many people seem to get easily discouraged along the way. Hopefully, I can introduce enough advice to help folks use their camera to get good lunar images. One source of discouragement is due to "comparison". Some novices compare one of their own raw images with the finished product of a veteran lunar imager. The raw image looks dreadful by comparison. However, the never-stated thing to remember is that the raw images of the veteran might not look only slightly better than the novice's raw images. Image processing - after image capture -- is one of the most important areas of lunar (and other) astroimaging and for the most part, learning about image processing tools and techniques is not optional. Almost no one captures a good lunar image without some processing.
Like most astroimaging, you face several obstacles: atmospheric seeing, transparency, camera noise, mount vibration, mount tracking, mount alignment, focusing, and dust on optics. However, for lunar imaging, mount vibration, alignment and tracking are less of an issue than it is for long exposures for galaxies, clusters and nebulae. (This is a key reason that lunar photography is such a good starting place.) Focusing is equally important, but a little outside the scope (haha!). So we'll skip those issues for now.
How do the masters of lunar imaging work their magic? Volume and image processing. "Volume" means that they capture multiple images. Any single image of the moon can be disrupted by bad seeing - atmospheric wobbliness. And any single image can have digital noise from the camera. By capturing and combining multiple images, the effects of poor (or even average) seeing can be minimized. Stacking multiple images also reduces the impact of digital noise because the statistical resolution of an image increases faster than the noise build up as the number of images in the stack increases. For lunar images, we generally want 5-20 images for a good stack. The quality of the final image depends on the quality of the raw images. So most imagers capture about 2-3 times as many images as they will stack and select only the best raw images for the stack. On a night with bad seeing, you might need 5-10x the number of images. So when preparing to stack lunar images to produce a single, high-resolution image, you'll need to capture a lot of images.
So this is one of the first discouragement points for novices: They capture a series of images of the lunar surface, but they find themselves dissatisfied with each of the individual frames. Sometimes, the raw image is totally unusable if it's out of focus or suffers from serious seeing issues. However, another seemingly poor raw image can contribute to a good final image though stacking and other image processing techniques. It's important to learn how to judge the quality of raw images despite their grubby appearance.
Image capture software keeps the images in different forms. JPG, FITS, AVI, TIFF and BMP are all popular image formats and each has its advantage. The key is to be sure your image capture and image processing software can find a common ground for image formats. Most people find that a single program for capture and processing isn't enough, so you should be prepared to work with two or more image management programs. Once you're ready to stack the images, you can tell the software which images to use and usually, you specify a feature that the software will use to act as an alignment point. In simple (one point) alignment, the software matches the correct horizontal and vertical offset of the images so the chosen alignment object is in the same spot. (In two point alignment, the software can adjust the angle of different images to align the photos.) Most stacking software will provide sub-pixel alignment capabilities and perform some image "fix ups" in the stacking process. Once the process is complete, you can usually perform some manual image tweaking and save the image.
Stacking images counteracts seeing and noise issues, but what about issues like dust or other imperfections on the optics or chip? These will show up as darker areas and sometimes they look like grey donuts. With small objects like planets, you can move the object to a "clean" part of the frame, but the moon often will consume most or all of the image space. Fortunately, many image processing programs have facilities to undo the effects of these imperfections. The trick involves capturing a "flat frame" and digitally compensating for image shortcomings. A flat frame is an image captured with the entire field of view with uniformly illumination. With the entire field of view illuminated, the captured image is a digital map to where light is being blocked by optical imperfections or chip obstructions. Software then can then adjust the processed image to increase the intensity in obstructed areas (proportional to the obstruction).
There are several techniques for capturing flat frames for example: sky flats, flat boxes, (See Ron Stanley's article in the August Night Times.) But one good trick for the moon is the "t-shirt flat". With this method, a thin, white cloth (e.g., a white t-shirt) covers the objective while the telescope is pointed at a bright object and the flat frames are captured. The nice thing about the moon is that for most of the month, it is bright enough to be the light source for the uniform lighting. (The thinner crescents just before and after New Moon are not bright enough.) The conventional wisdom is that a flat image should be about 1/3 of the maximum illumination in any one pixel. So if your camera pixels have a maximum value of 256, then the maximum pixel value you want in your flat image should be about 85. Adjust the shutter speed or gain to get the maximum value there and you will have a good flat field image. (Taking multiple flat field images and stacking the results will give better results.)
It's very important that the flat field images represent the same imaging path - including focus - that you use to capture your images. So if you change a filter or refocus or realign the camera relative to the optical path, then you need a new flat field set.
Once you've captured one set of flat images, you don't need to repeat the process unless you reposition the optics or modify the optical setup. But you do need to capture a set of flat images at least once during the night's session. Different software uses your set of flat images differently. Some software expects you to stack the flat images and produce a single, master flat to be applied to your "real" images. Other software will take the set of raw flat frames and combine them automatically.
At this point, we've addressed some of the biggest issues that make your raw images look poor compared to the experts' finished images: (a) digital noise from the camera, (b) seeing issues, and (c) optical flaws like dust. These variations between raw and polished images can make a world (pun intended) of differences. Next month we'll explore a short practical example of going from raw images to a finished moon photo.Published in the September 2005 issue of the NightTimes