There are two ways to think of getting the “most” out of an image:  capturing the faintest possible details, and capturing the finest possible detail.  The ideal situation would involve getting an image with both extremely fine detail (high resolution) as well pulling in the dimmest wisps of nebulosity or the most elusive galactic spiral arms.  Often, however, these goals are almost mutually exclusive.  For deep-sky imaging of objects such as nebulae and galaxies, faintness is usually what counts, and sacrificing some resolution is not a problem.  For planetary or other imaging involving objects with small angular size (double stars, planetary nebulae, globular star clusters, etc.) it is high resolution we seek.

Sky Conditions

Novice astronomers often confuse clarity and steadiness of the atmosphere.  The clearness of the sky, the transparency, is a measure of how much light is making it to Earth from space.  A night of good transparency will be a crystal-clear night, with black skies and countless stars.  A steady night, one of good seeing conditions, means the atmosphere is calm and fine detail is easily visible.  Good seeing is essential for planetary observation but not so critical for deep sky.  Good transparency is best for deep-sky viewing and imaging, while it is not necessary for planetary observations.  In fact, good seeing and good transparency are almost mutually exclusive.  The best seeing usually comes on hazy nights when few stars are visible, and often the seeing is best before the sky is even completely dark just after sunset.  Transparent nights, on the other hand, are often associated with turbulent conditions.  The ideal nights have both good seeing and good transparency, but are far rarer than we would like.

Off the Deep End – The Faintest Detail

If you hang around a telescope shop, or with a group of CCD imagers or photographers at an astronomy club, before long you will hear exclamations of awe upon viewing someone’s latest image of a familiar object.  Many people wonder what could be so exciting about an image of something that has been photographed thousands of times.  The fact is that the equipment and techniques available to amateur CCD imagers have allowed them to begin capturing familiar objects in so much detail as to render them new again!  Professional images from mountaintop observatories, taken with multi-million-dollar instruments just a few years ago pale in comparison to what many amateurs now routinely capture from their backyards, often in light-polluted suburbs!

What are the keys to capturing the faintest possible detail in a CCD image?  The first consideration is the weather.  While this is an important factor in all astronomical pursuits, it is probably most critical for imaging.  To bring in extremely faint detail it is necessary to have very clear skies which will transmit as much might as possible.  On a hazy night, or when windy conditions during the day have kicked dust into the air, it is best to hold off on trying to reach for the faintest subjects.  While seeing conditions are not as critical as transparency, very poor seeing — often associated with the passing of a front and the arrival of clear skies — will hinder any efforts at super-deep imaging.

Once you have a the perfect weather conditions, the other factors are in your hands.  There are two factors left:  technique and equipment.

Techniques for Getting Faint Detail

Using the telescope and CCD camera you already have, technique can significantly improve the amount of detail you are capturing.  The single most important factor is exposure time.  The more light you capture, the fainter the objects you can image.  There are limitations to this, but in general if you want fainter detail, spend more time imaging the subject.  This does not necessarily mean taking a single longer exposure, but taking multiple exposures at an ideal exposure length.

The key is signal to noise ratio (S/N).  The signal comes from the object you are imaging, the noise from the inherent electronic noise of the camera as well as skyglow from artificial and natural lighting.  The idea is to maximize the signal to noise ratio:  more signal, less noise.  Below are some ways to reduce or avoid extraneous noise sources.

  • Electronic noise is minimized by cooling the CCD camera and by taking dark frames.

  • Other “noise” in the optical path such as dust appearing as faint out-of-focus donuts in an image can be removed using flat fields.

  • Skyglow can be minimized by getting to as dark a site as possible away from all city lights.  Extra elevation is always helpful, so take advantage of it if you live in a mountainous part of the world.

  • Natural skyglow is caused by auroras, which are rarely a problem outside of the far northern and southern parts of the world, but keep it in mind if you live in the high latitudes.

  • The sky glows from reflected sunlight for quite some time after sunset, especially in the summer and for longer and longer times the farther north you live.  To get the faintest detail, try to image only after astronomical twilight has ended in the evening and before it begins in the morning.

Note:  Evening astronomical twilight ends when the sun is more than 18 degrees below the horizon and morning twilight begins once the sun gets within 18 degrees below the eastern horizon before sunrise.  The exact time depends on the time of year and your latitude, but astronomical twilight normally ends about 90 minutes after sunset.

By taking a longer exposure, the signal to noise ratio is increased, but only up until the point where the noise, especially from skyglow, starts to overwhelm the faint detail in the object being imaged.  Beyond this it is possible to continue increasing S/N by stacking multiple exposures.  Since the signal remains the same from image to image, adding two images together will double the value of the signal.  Noise, on the other hand, is random, so adding two images together only increases the noise by √2 (about 1.4) times, or 40%.  Adding three images would triple the signal, but the noise doesn’t quite double, so the overall S/N is increased 75% over a single image, and so on.  There is a point of diminishing returns, but in general, more images are better.

n Images Stacked

S/N Increase Over n-1 Images

S/N Increase Over Single Image

1

2

40%

40%

3

33%

75%

4

27%

100%

5

24%

124%

6

21%

145%

For a more detailed discussion of signal-to-noise ratio and capturing the most detail possible, visit the CCD Theory page on Ideal Exposure Timesas well as the Ideal Exposure Calculator.

Equipment for Getting Faint Detail

In visual observing aperture is everything.  A larger telescope will show you more detail than a smaller one.  To some extent this is still true for imaging, but less so.  For a given exposure time, a larger diameter telescope will pick up fainter stars, but not necessarily fainter extended detail such as nebulosity.

The stellar limiting magnitude of a given exposure is determined by aperture;  the amount of extended detail captured, however, is a function offocal ratio.  This means that, with a given exposure time, an 8-inch f/3.3 telescope will capture just as much nebulosity as the 200-inch Hale telescope on Mt. Palomar, which is also an f/3.3 system.  The image scale will be much greater in the Palomar image, since the focal length of the giant scope is 25 times longer than the small one, but the faintest details will be the same (assuming you have the 8-inch telescope sitting on top of Mt. Palomar, anyway).  It also means that a 4-inch telescope imaging at f/5 will pick up fewer stars than a 8″ telescope at f/10 during the same exposure time, even though the bigger scope is “slower”.  The ideal solution is the largest and fastest telescope possible!

As for the CCD camera itself, sensitivity is the key.  A more sensitive camera will pick up more detail in a given amount of time.  Aside from actualquantum efficiency — the measure of a CCD’s sensitivity — pixel size is important: larger pixels detect more light in a given exposure than smaller ones.  However, if pretty pictures are still a goal, the best match for the telescope may not be a high-sensitivity camera, and many very deep images have been taken with relatively “slow” telescopes and CCDs with small pixels (which are not usually considered to have the highest sensitivity).

Tip:  Try binning the pixels to increase the sensitivity of the CCD.

The Sharper Image – High Resolution Imaging

As important as the sky conditions are for imaging faint objects, they are the single most important factor for high resolution imaging, especially of the planets.  All the best equipment, techniques, and processing will not make any difference if the seeing is poor when the images are taken.  This is what makes planetary imaging so much more difficult than it seems like it would be.

Once you have a good night, the trick is to take advantage of it and maximize the resolution obtained.  For most purposes matching the pixel size to the focal length of the telescope is not really critical, but for high-resolution imaging it becomes important.  For deep-sky imaging, a relatively fast focal ratio is usually desired to minimize exposure times;  when high-resolution is the goal, however, a longer focal length provides more image scale, and for most instruments this will require a slower focal ratio.  For planetary imaging, a very long focal length is required to provide sufficient image scale to resolve small details on the tiny disks of the planets.

Stellar detail is also dependent on resolution.  Stellar signal to noise ratio is a function of sampling, which depends on the pixel resolution.  If after very fine details in celestial objects such as globular clusters or detailed galaxies and planetary nebulae, having the right pixel size versus focal length is important.  Undersampled images will have higher signal to noise ratios than properly sampled images.

Note:  For an in-depth discussion on pixel size vs. focal length (sampling), visit the CCD Theory page on Pixel Size.

Deep-Sky Pixel Size

The most common figure given for pixel resolution is 2″/pixel, meaning each pixel covers 2 arcseconds of sky.  The idea behind this is that the best data is obtained when a star image is 2 pixels wide.  If the seeing conditions from a typical backyard site average 4″ over the course of a long exposure, then during a long exposure when seeing blurs the star image, a star will appear 4″ across.  An image scale of 2″/pixel means each star is covered by four pixels, providing good “sampling”.

 It can be argued that this is still less than ideal, and that something closer to 3 pixels per star-diameter is better, in which case 1.66″/pixel might be better.  Of course, this is still assuming 4″ seeing.  If you get a night of very good seeing, even higher resolution might be suitable.

Why not just use smaller and smaller pixels to achieve higher sampling?  Because there is a trade-off in sensitivity.  Suppose a star covers one 18-micron pixel and produces a pixel count of 40,000 during a given exposure.  Using 9-micron pixels divides the star into four pixels, now each recording only 10,000 counts during the same exposure.  Smaller pixels, all other things being equal, are less sensitive than larger pixels.

The ideal solution is to stick with the tried-and-true rule of thumb that states “Use a pixel resolution of one-half the typical seeing conditions”.  Image processing software will allow you to measure the average star size to determine the long-term seeing during an exposure.  This information can be used to optimize the imaging setup for the best resolution.

Above:  This image of M51 was measured to have 2.5 arcsecond star images.  The pixel resolution was 0.5 arcseconds/pixel, yielding a sampling rate of 5x.  This would be an oversampled image.  However, it isn’t exactly a bad picture either, so once again, if you are after pretty pictures over absolute resolution, over- and undersampling are often fine.

Planetary Pixel Size

In this case, oversampling is a good thing.  Some sensitivity is lost, as explained above, but then again the planets are about 1000 to 2000 times brighter than an average 10th magnitude galaxy.  Also, oversampling provides more information which can be extracted later using advanced image processing techniques.

A recommended resolution for planetary imaging is 0.25″/pixel.  This is much easier to achieve than trying to obtain the proper pixel resolution for deep-sky imaging.  For deep-sky you are pretty much limited to the focal length of the telescope being used.  With planetary imaging, it is a simple matter to change the focal length by using a Barlow lens, or by using eyepiece projection.

Using an ST-10XME CCD camera with 6.8-micron pixels, a focal length of 5600mm is required for an image scale of 0.25″/pixel.  This can be obtained by using an 11″ SCT at f/20 (easily obtained by using a 2x Barlow lens) or by using a 6″ refractor at f/37 (requiring the use of something in the neighborhood of a 15mm eyepiece and an eyepiece-projection arrangement).  The advantage of aperture for planetary imaging can be seen:  the exposure time will be much less for a larger scope since a faster focal ratio can be used for the same scale.

Note:  See the Planetary Imaging section for more details on pixel size for planetary imaging.

To calculate the focal length needed for 0.25″/pixel resolution, use the following equation:

    Focal Length = Pixel Size * 825

To calculate the eyepiece required to obtain a certain focal length with eyepiece projection, use the following calculator:

Eyepiece Projection Calculator