Light Frames - In deep-sky astrophotography, the primary signal that we are interested in comes from the photons of light that come from the astronomical object and that are recorded in our long-exposure light frames. As discussed in detail in Chapter 3, in addition to the signal recorded from the deep-sky object, other signals, such as heat-generated thermal signal, are also recorded. To reduce or eliminate these undesirable signals, support frames are used to calibrate the light frames. Support Frames
The latest generation of DSLR cameras are very low noise. Depending on the brightness of the object you want to shoot, the sky brightness at your observing site, and the ambient temperature, you may be able to get away with only shooting light frames and with not shooting any support frames at all (see M17). However, as you become interested in improving your long-exposure deep-sky astrophotos, you will want to start shooting support frames, the most important of which are dark frames. Dark frames are time exposures of the same length as the light frames, but with no light reaching the sensor. They essentially take a picture of the signal generated by the heat in the camera, as well as the bias signal. Dark frames are more important on warm nights with higher ambient temperatures. In addition to undesirable signals, the signal of interest can be modified by the optical and mechanical aspects of the imaging system. For example, the optical system can suffer from defects such as vignetting. There will also probably be tiny specs of dust (and some not so tiny) on the glass that protects the sensor in the camera. These dust spots may cast a shadow on the sensor during the exposure and will be recorded in the image. These can also be removed by shooting a "flat-field" frame and using it later when the images are processed. Bias frames record the signal from the low-level charge that is applied to a CCD or CMOS sensor in the form of a fixed offset voltage value. The bias signal can be removed by subtracting a bias frame - a zero second exposure that records just the bias and readout noise. These support frames are used to calibrate each individual light frame. If you have the latest model low-noise digital camera and are shooting under cold ambient temperatures, you may be able to get away with shooting only light frames. It just depends on how critical you want to be with the final image. In most cases, shooting support frames, and in particular dark frames, will tremendously help your final images. For all four types of frames, you should shoot raw file format in the camera. Levels of Effort and Complexity There are different levels of effort and complexity that can be pursued in astrophotography with a DSLR camera and that lead to different levels of quality in the final image. From the most simple to the most complex, and from the lowest quality to the highest quality, they are:
Light frames The light frame is the time exposure that records the light from the actual object in the sky. It also records the light from the sky background from sources such as interplanetary dust (Zodiacal Light and Gegenschein) and any light in the sky "foreground" from sources such as light pollution primarily, and also the faint airglow at dark-sky sites. With the linear sensors in CMOS and CCD cameras, it is best to shoot a series of shorter exposures (each called a "sub-exposure") and average them together to approximate a single longer exposure. The longer the total exposure, the more signal is gathered, and the higher the signal-to-noise ratio in the final image. Each individual exposure should be long enough to reach a minimum signal-to-noise level so that the signal is not lost in the read-out noise of the camera. The exposure necessary for this is easy to determine by examining the histogram. The bulk of pixels should not be touching the left hand side of the histogram. There is a limit to how short an exposure can be used in a sub-exposure. For example, 10 five-minute exposures (50 minutes total exposure time) will come close to equaling a single fifty-minute exposure. But, 3,000 one-second exposures, which also equals 50 minutes total exposure time, will not work because the miniscule signal gathered in one second will be overwhelmed by the readout noise present in each frame. Too long of an exposure can be less than optimal also. Overexposing an image can cause loss of detail in the highlights and possibly cause "blooming" on bright stars. Blooming is vertical lines by stars caused by pixel overflow. Blooming is usually not a problem with DSLR cameras which have excellent anti-blooming technology built into them, but can be a problem with non-anti-blooming dedicated astronomical CCD Cameras. If you are not guiding, you may also need to limit the length of the exposure to how long the mount can track without error. This can be determined by trial and error. If you are shooting with an altazimuth mount, your exposures will be limited by field-rotation. The light frame should be exposed long enough to get the most signal-to-noise, but not long enough so that poor telescope tracking smears the image, or so long that bright portions of the object are overexposed (clipped) so that they contain no detail. Exposure length will also be limited by the ISO used and the thermal noise present at whatever ambient temperature the exposures are made at. According to Wei-Hao Wang, an astronomer and astrophotographer at the National Radio Astronomy Observatory, it is only at extremely dark sky sites (darker than mag 21 per square arc second) with slow f/ratios (slower than f/8) that the maximum exposure is limited by the thermal signal generated in a modern low-noise DSLR camera. At less-than-perfectly-dark-sky sites, the maximum exposure is usually limited by light-pollution sky fog or amp glow in those cameras that do not turn off the readout amplifier. Exposure length for light frames is determined by a number of considerations, but on a good mount with guided exposures at a dark-sky site, a good starting point at f/8 seems to be around 5 - 10 minutes at ISO 800 or 1600, at temperatures from 50 to 70F (10C to 21C) with the latest generation of DSLR cameras. At colder temperatures, lower ISOs can be used with longer exposures. Shorter exposures can be used with faster focal ratios. Let the histogram guide you. We'll discuss this in detail in the next section. A light frame is actually made up of different "signals". A Light Frame = (Sky and Object signal * Flat signal) + Dark signal + Bias signal. For optimum results, light frames usually need to be "calibrated" by subtracting a dark frame and dividing by the flat-field frame that has had the bias subtracted from it. Most astronomical image processing programs will do all of this automatically if you tell them which frames are which.
Dark Frames A dark frame is an exposure that is made with no light reaching the sensor. It records the thermal current signal of the camera for a given ISO, temperature, and exposure time and varies by camera model. All openings to the telescope, or camera lens, should be capped, and the camera eyepiece covered. Dark frames should be shot at the same ISO, temperature and exposure as the light frames. As many dark frames as possible should be taken and then averaged together later in software to produce a master dark frame. This will increase the signal-to-noise ratio between the thermal current signal and the thermal current noise. Because dark noise is reduced by the square root of the total number of exposures added or averaged together, more dark frames produce better results. When shooting on a clear night at a dark-sky site, it is better to spend most of the time making light frames. Dark frames can be made during down times between acquiring and shooting different objects, when clouds roll through, or at the end of the night if the temperature has stayed relatively constant. Try to shoot a minimum of 9 dark frames if you shoot them on a imaging night. 16 dark frames are better. More than that can produce even better results, but large numbers should be shot at another time, on a non-imaging night, at the same temperature as the light frames. You will need more dark frames when the ambient temperature is high. High temperatures produce more thermal signal, and therefore more noise. A master dark frame is created in software by averaging the individual dark frames, and then the master dark is subtracted from each individual light frame during the calibration phase of image processing. A dark frame technically includes a bias signal as well as a dark current thermal signal. Bias is present in all frames. Since it is present in the dark frame, when we subtract the dark frame from the light frame, we also remove the bias signal. It is more time efficient to shoot a large set of dark frames at another time, say on a cloudy night. A master dark frame is then created that is subtracted from all of the light frames shot at the same temperature. A library of dark frames for different temperatures can be created and used with different imaging sessions. Make sure all openings in the imaging system are covered, such as the front aperture of the scope and the viewfinder eyepiece in camera. You don't even have to have the camera on the scope or with a lens on it to shoot dark frames. If you are worried about light sneaking in somewhere in the scope, say around the primary mirror in a Newtonian, just take the camera off of the scope completely, and put a body cap on the camera, and cover the camera's viewfinder eyepiece and shoot the dark frames. If you are really serious about producing the best results, shoot a whole lot of dark frames to produce a really good, low-noise master dark frame. Don't try to shoot your darks on the same night as your lights. This is inefficient because it wastes precious clear dark-sky time when you could be gathering actual photons from the object of interest, instead of shooting the back of a lens cap under clear dark skies. If dark frames are shot later, after light frames are gathered, conditions should be as close to possible for both. For example, if the light frames are shot with a 120 volt adapter powering the camera, then the dark frames should be shot the same way. It may be tempting to think that you can shoot the dark frames with the camera in the camera bag on the drive home. This is not a good idea as the camera will heat up considerably in the bag compared to when the camera was on the scope out in the open air shooting light frames. Internal Camera Temperature Stability Because DSLRs do not have any regulated cooling, the internal temperature of the camera will change during an exposure session. It will heat up during use. This is true for both light and dark frames. This is a problem for accurate dark subtraction because the amount of thermal signal will not match if the lights and the darks are taken at different temperatures. Subtracting dark frames taken at higher temperatures than the light frames can result in "dark holes" being left behind in the image. These holes are the size of individual pixels or small groups of pixels. If the dark frame was colder than the light frame, light spots will be left behind. These light and dark artifacts will smear across the frame and make an ugly mess when the stars in the lights are aligned to correct for slight movement between frames due to tracking inaccuracies and flexure between the guide scope and imaging optics. The internal temperature will increase as the camera is taking exposures until it stabilizes after about an hour or so of constant use. If different objects are shot during the night, the camera will cool off slightly during down time as a new object is acquired, framed, and focused. The camera will heat up more depending on whether Live View is used, and how long it is turned on, as well as how much the LCD is used for reviewing images. Conditions while shooting dark frames should duplicate as closely as possible the conditions when the light frames were shot. Ideally this means starting with the camera at the same ambient temperature at the beginning of the exposure run and exposing for the same total exposure time so the camera is at the same temperature at the end of the run as well. In practice, this can be difficult to accomplish. What you usually end up with during an exposure run is a series of frames with increasing different temperatures until the camera's internal temperature stabilizes at its maximum relative to the ambient temperature. For example, shooting darks with my Canon 550D (T2i) at an external ambient temperature of 12C, for 10 minute exposures at ISO 1600, the camera's internal temperature, as recorded by the EXIF data written to the image file, shows the first frame at 14C, the next at 20C, then 23C, 24C, 25C, and finally stabilizing at 26C after about an hour. This rapid heat up in the beginning and large amount of heating is typical of the latest cameras that shoot video, probably because of the extra electronic circuitry inside the camera. On the positive side, thermal signal in these latest cameras is well controlled, so even though they heat up more than cameras without video, the total amount of thermal signal appears to be less. This may be because the new sensors actually produce less thermal signal, but is probably because of the sophisticated digital signal processing that is applied to the data from the sensor. This same heating up will occur for both light and dark frames. For the dark frames, we can program the camera to shoot darks all night long, and then just throw away the first hour's worth that have different temperatures. The rest of the frames should all be at the same temperature. Light frames are more problematic. We don't want to throw away an hour's worth of signal. We want to use these frames, but many will be at different temperatures. There are several solutions to this problem.
Dark Noise Remember, even though thermal signal looks like noise, it is not. It is signal. Dark (thermal) current signal in a light frame can be removed by shooting a dark frame at the same exposure and temperature as the light frame, and then subtracting the dark frame from the light frame. Along with any signal, however, is associated noise. Even after subtracting the thermal signal with a dark frame, the thermal noise in the light frame will remain. This cannot be removed. There is also thermal noise in the dark frame. When the dark frame is subtracted from the light frame to remove the thermal signal, the dark noise in the dark frame is unavoidably added to the light frame. This may seem confusing because we are subtracting the dark from the light, so how can it be adding noise? The reason is because the noise is random. Subtracting random noise just makes the noise already present in the light frame get worse. The amount of image degradation from the additional thermal noise that is added in the dark-subtracted light frame can be reduced by averaging a large number of dark frames to create a good low-noise master dark frame. With this method, the thermal signal in the master dark is reinforced and the random dark noise is averaged and reduced. Just as the signal-to-noise ratio is improved in light frames by the square root of the total number of frames, the noise is reduced in the master dark frame by the square root of the total number of sub dark frames. For example, if a certain amount of noise is added to the light when a single dark frame is subtracted, then only half of that noise would be added if a master dark frame was made up of 4 sub dark frames. The square root of 4 is 2, so a 2x improvement in noise would result in only 1/2 of the noise added compared to a single frame. Likewise, compared to a single dark frame, only 1/3 of the noise would be added if 9 frames were averaged for a master dark, 1/4 for 16 frames, 1/5 for 25 frames, 1/6 for 36 frames, 1/7 for 49 frames, 1/8 for 64 frames and so on. At about 64 frames you start to reach the point of diminishing returns, but if you shot 100 subs for your master dark, you would add only 1/10 of the noise from a single dark. It is important to use a lot of subs at higher temperatures. Amp Glow Amp Glow is a red glow in the corners and edges of a long exposure image caused by electrons in the metal-oxide semiconductor field-effect transistors (MOSFET) that are in the readout amplifier of the camera's sensor. Amp glow is not caused by heat, it is caused by electroluminescence. If not too severe, it can be removed by subtraction of a dark frame. If exposures are too long, the amp glow will saturate a portion of the frame and no meaningful data can be recorded there. The amount of amp glow varies with the particular camera model, so you will need to make some tests at various exposure times to see how bad it is. Most new camera models turn off the readout amplifier during long exposures greatly reducing the amp glow. Scaling Dark Frames Dark current scales linearly with exposure time. This means it is possible to create a master dark frame taken at the same temperature as the light frames and scale the master dark frame for use with different exposure times for the light frames. To create a scalable master dark frame, the bias must be subtracted first, resulting in a "thermal frame" which contains only the dark current. If this method is used, the bias should then also be subtracted separately from the light frames since it won't be removed when the dark frames are subtracted. When creating a scalable master dark frame, an exposure should be used that is longer than the exposures for the individual light frames. Creating a Master Dark Frame Library Using a large number of dark frames is very beneficial in terms of controlling dark noise. Some experts recommend five times the number of darks as lights! This is obviously extremely difficult to accomplish if darks are taken at the same time as lights. A better method is to create a dark frame library of master dark frames taken at the same exposure, ISO and temperature as you will usually be shooting your light frames. It helps to standardize on a couple of parameters to cut down on the total number of dark frames you will need for each set of conditions. Try to create a set of master dark frames at 10 degree intervals over the temperature range at which you shoot. For example, I normally image from either my driveway in the suburbs with a fair amount of light pollution and a naked-eye limiting magnitude of about 4, or from a reasonably dark-sky site with a naked-eye limiting magnitude of about 6 where the Milky Way can be seen. At 70 degrees F (21C), I can shoot about 60 second exposures at ISO 800 from my driveway for the light frames. From my dark-sky site, I can shoot 300 second exposures at ISO 800 at 34F (1C) for the light frames. All I need to do is to shoot a series of dark frames at these same temperatures and ISOs and exposure times. The more individual dark frames I can shoot to create the master, the better. Since they are dark frames, and all openings to the camera are covered, this library can be created indoors or outdoors, during the day or night, just as long at the temperature stays fairly constant. If the frames are made at cold temperatures, the camera should be placed in a closed zip-loc bag before it is brought inside to warm conditions to prevent moisture condensation inside the camera. It should be left inside the bag until the camera warms up to room temperature. I created my dark frame library in my garage. I use a TC-80N3 to automate the imaging session, and let the camera shoot dark frames all night long at a particular temperature and ISO and exposure time. The camera will warm up for the first hour or so of shooting and then stabilize. If you have an older camera that does not record the internal camera temperature in the EXIF header in the image file, you can tell when the camera has stabilized by examining the file size of the individual frames. I simply throw away the early ones that were taken when the camera was still warming up and had not reached thermal equilibrium. I shoot a lot of bias frames at the same temperature and ISO as the darks so I can create a good master bias frame with which to scale the master dark. The library of master dark frames I have created go from about 30 degrees F to 70 degrees F, spaced at about every 10 degrees F. Most are at ISO 1600, which is the optimum ISO for my camera (Canon 20Da) and are 10 minute exposures. Since I also create a master bias frame, these 10-minute exposures can be scaled down and used for shorter exposures at the same ISO and temperature. A new dark frame library should be created every so often over time as the camera ages if the dark current properties change. In-Camera Dark Frame Subtraction Many DSLR cameras offer an option for long-exposure noise reduction. This is accomplished with a dark frame that is shot automatically by the camera immediately after the light frame. This essentially doubles the time necessary to shoot one frame. Clear dark-sky time is very valuable and can unquestionably be better spent gathering photons since dark frames can be shot at another time. There is no other way to increase the signal in the all-important signal-to-noise ratio of our images other than by gathering more signal through more exposure and more light frames. This in-camera, single frame method of noise reduction by dark frame subtraction is also applied to a raw frame, making it not really "raw" anymore. If calibration with a master dark frame is planned, then in-camera, long-exposure noise reduction should be turned off.
Flat-field Frames A flat-field frame is a picture of the optical system's peculiarities, such as vignetting, uneven illumination, and dust shadows. It also records sensitivity variations between individual pixels in the sensor, which are usually small, but present nevertheless. A flat-field frame is made by photographing an evenly illuminated subject, such as the cloudless sky at twilight, or a light box, or by placing a double white T-shirt over the aperture of the telescope and pointing it at an evenly illuminated extended light source such as a blank wall. The flat-field frame will be a relatively short exposure because you can use a light source that is much brighter than the faint deep-sky objects you will normally be shooting. Because the exposure is short there won't be much thermal noise, but to be precise, a dark frame, taken at the same exposure and temperature as the flat-field, should be subtracted from the flat-field frame. The bias will be subtracted with the dark frame, so subtracting a separate bias frame is not necessary. Flat-field frames should be acquired with the telescope in exactly the same configuration as far as focus position and camera orientation that will be used for the light frames. If the optical system suffers from an asymmetrical illumination problem, such as from the shadow of a pick-off mirror in an off-axis guider, then rotating the camera to frame an object differently will change the orientation of this uneven illumination in relation to where it was when the flat-field frames were taken. Flat-field frames should be shot at a low ISO setting. Several can be collected easily because these exposures are short since they are of a relatively bright subject. They are then averaged together to create a master flat-field frame. Shoot 9 or 16 flat-field frames to create a master. Flat-field frames do not have to be taken at the same ISO and exposure or temperature as the light frames, unless you are using a program like DeepSky Stacker that expects them to be at the same ISO. Make your exposure so that the histogram's peak is in the middle. A good and easy way to do this is to just put the camera on auto-exposure on Av (Aperture Value where the aperture is fixed and the shutter speed changes to correct the exposure). The histogram of the flat-field frame will reveal how much uneven illumination is present in the system. The narrower the histogram, the better. If a sky flat is used, the predominant color will be blue. This frame can simply be turned into a grayscale frame for division from the light frame. An astronomical image processing program's calibration routine will perform this calculation automatically. With sky flats, it is also possible to pick up subtle sky gradients during twilight. If you are shooting with a camera lens, you can rotate the lens and camera body together 90 degrees between exposures and then median combine them to remove these gradients, but this can be difficult to accomplish with some lenses, and impossible to do with most scopes. Use of a t-shirt or light box for flats will eliminate this problem. Naturally, t-shirts and light boxes introduce their own problems, the main one being getting the illumination even. If you are using camera lenses, each lens must have its own flat-field frame to correct for geometrical vignetting, which is present in almost all lenses, especially when used wide-open or near its maximum aperture. For telescopes, you must shoot a flat-field frame every time you change the orientation of the camera for framing purposes. To shoot a sky flat, aim the scope towards the section of sky opposite the Sun after it has set and it is still twilight. Move the scope slightly between exposures in case there are stars in the field, or just take the flat-field frames with the drive off. You can later median combine all of the flat-field frames to remove stars that may be in individual frames. This is another reason to shoot more than one flat-field frame and combine them to create a master flat-field frame. It is important to reduce the noise in your flats or this noise will end up in your final images. Use a low ISO and average several images to create a master flat-field frame. While flat-field frames can remove vignetting, they cannot compensate for the reduced exposure in the sections of the image where vignetting is present. For example, if a vignetted portion of the frame is receiving only 50 percent illumination, stars and nebulosity will have a lower signal-to-noise ratio than in the non-vignetted portion of the image. If vignetting is a serious problem, it is better to locate its source and correct it in the optical system than try to fix it with flat-field frames. Good flat-field frames use an illumination source that is bright enough so that exposures are kept relatively short, so that thermal signal and thermal noise are not a problem. This usually means exposures shorter than 1 second. If you must use longer exposures, it may be necessary to take dark frames for your flat-field frames, create a master flat-field dark frame, and subtract it from each individual flat-field frame before making a master flat-field frame. Some telescope's illumination can vary because optical and mechanical components move from flexure when the scope is pointed at different parts of the sky. If your scope does this, you should construct a light box and use it to shoot new flats whenever you shoot an object that is in a different part of the sky.
Bias (Offset) Frames A bias (offset) frame is zero second exposure that records the bias signal present in every frame. The bias signal comes from electrons from current that is applied to the sensor to get it charged and ready to collect electrons generated by the detection of photons from the sky. Bias adds an offset or pedestal to the voltage presented to the Analog/Digital converter. The bias signal is constant in every frame taken at the same temperature. It is essentially a picture of the constant voltage offset, the amplifier readout noise, and camera electronic noise. The bias signal is different for each pixel but very nearly the same from frame to frame, varying only by random electronic, readout, and digitization factors. It may show up as fixed pattern noise in some cameras. The bias signal is specific to an individual camera and changes with different ISO settings. Ideally a bias frame for calibration is a zero second exposure that records only the bias signal and readout noise. But practically with a Digital SLR camera you will have to use the shortest exposure (highest shutter speed, such at 1/4000th second) of which the camera is capable. Average many bias frames to create a good master bias frame. The exposures are so short it is not much effort to shoot a lot of frames. If the camera is on the scope, make sure the aperture is capped so no light can get in, and make sure the viewfinder eyepiece of the camera is closed or completely covered. If the camera is off the scope, use the body cap and viewfinder eyepiece cover. Be aware that with Newtonians and other open tube scopes, light can sneak in through the bottom of the tube and other places. With these types of telescope, it is better to remove the camera from the scope. To preserve the camera's orientation if you are shooting flat-field frames, leave the 2-inch adapter locked into the focuser, and just remove the camera from the t-mount. For normal image calibration, subtraction of a bias frame is unnecessary. Because the bias is part of both the light and dark frames, when a dark frame is subtracted from a light frame in the calibration process, the bias is automatically subtracted also. Bias frames are useful in calibrating flat frames. However, if you calibrate your flat frames by subtracting dark frames, then the bias is subtracted there also, just as it is when the light frames are calibrated by subtracting dark frames. In both of these cases you don't need bias frames. If you don't subtract a dark from the flat, then you do need to subtract a bias from the flat. As we have already discussed, bias frames are useful in creating scalable master dark frames.
Order In Which To Shoot Support Frames If you are going to shoot support frames, you will have to decide on the order to shoot them in. Obviously if you intend to shoot twilight flat-field frames, you will need to shoot them immediately after you have set up before it gets completely dark. Don't forget to focus the telescope on infinity and don't change the orientation of the camera afterwards. Twilight flats can also be shot in the morning before sunrise if you stay out all night. Other types of flats, such as those made with a light box can really be made at any time. If your telescope suffers from a lot of flexure, you might want to shoot them immediately before or after shooting the light frames without moving the telescope to another object. Bias frames can also be shot at any time as long as the temperature is near the same as when the light frames are shot. Dark frames are best shot on a cloudy night. If you intend to gather support frames, especially if you are incredibly critical and shoot dark and bias frames for the flat-field frames, you can eat up a lot of memory space on your compact flash card or computer hard drive in a short amount of time because the raw files are fairly large. If storage space is a problem, archive the files off the compact flash card in the camera to the hard drive, or burn a CD-ROM if your laptop or desktop has a CD burner. Otherwise, pay attention to how many files you have written to the Compact Flash card and be careful not to schedule a large number of frame acquisitions with an automated process that will write more data to a compact flash card than it can hold. You will end up losing data this way. |
||||||
Back | Up | Next |