4. The Filters

4.1 The Spectrum and Monochrome CCD Sensors

Light is made up of particles called photons. Photons can have a wide range of frequencies and therefore wavelengths. Depending on the wavelength (or frequency) of a photon, they lie in a different part of the electromagnetic spectrum. When we are imaging in astrophotography, we are mainly interested in capturing the visible part of the electromagnetic spectrum - that is, in registering the photons that lie within this region of the spectrum. At times, we are however interested in regions of the electromagnetic spectrum that lie within the infrared range. CCD sensors are built to be sensitive to a wide range of wavelengths within the spectrum. As discussed in the previous section however, colour cameras have a Bayer filter in front of the CCD sensor that pre-filters the incoming light into Red, Green and Blue, to fall on specific pixels and thus produce a colour image in one exposure. Moreover, cameras such as DSLRs contain internal IR filters to block most of the infrared light that comes in. This makes colour cameras, in particular DSLRs, less usable for astrophotography in terms of user control over what is being imaged. 

Monochrome CCD sensors do however present the ultimate form of control over imaging. In particular, within astronomical CCD cameras, which have zero filters in front of the CCD sensor and when monochrome, have no Bayer filter on the sensor itself. This allows the CCD sensor to image the entire range of wavelengths it is built to be sensitive to, without hindrance. For example, the Kodak KAF-8300 is a popular CCD sensor used in astronomical CCD cameras. The following is the spectral response for this sensor:

The spectral response above shows how well the CCD sensor in question responds to light of varying wavelength, spanning the ultraviolet, visible and infrared regions of the electromagnetic spectrum (the response continues slightly below 400 nm wavelengths). The height of the line (absolute quantum efficiency) defines how many photons of light cause a signal. For example, in the middle of the visible region, in Green, the absolute quantum efficiency is 0.6. This means that if 100 photons come in that are of this particular wavelength, 60 of them will be registered and the other 40 will not. As discussed in the previous section as well, this means that someone using a monochrome camera and imaging Red, Green and Blue with exposures of equal length, will find that the resulting image has a green tint to it. This is because more signal was registered in Green than in Red and Blue within the specified exposure length. This can of course be altered later in post-processing, but it is worth-while pointing out in relation to how CCD sensors detect light. 

What the monochrome CCD sensor naturally fails to provide, is a colour image in a single exposure. All exposures captured are, as expected, monochrome. This is where filters come into play, to later produce colour images with a monochrome CCD sensor. 

4.2 Filters Used in Astrophotography

Filters used in astrophotography tend to be of high quality coated glass that have well-calibrated transmission curves. The transmission curves essentially define which wavelengths of light are allowed through the filter and which are rejected. For users of monochrome CCD cameras, it is important that filters used for Red, Green and Blue have very specific transmission curves that complement each other well. This is important so that the Red filter does not allow a lot of Green light through, for example. Having the cut-offs at very nearly the same wavelength is therefore important. The following are the transmission curves for Baader's (a popular manufacturer of astronomical filters) Luminance, Red, Green and Blue filters:

It is very clear from the above transmission curves that the Red, Green and Blue filters have sharp cut-offs, to encompass only the specific areas of the visible spectrum corresponding to them. There is a little overlap, naturally, but this is minor considering the transmission percentage. Slightly less visible in the above transmission curves is the Luminance (see the grey L to the right of Red). Luminance is used extensively by professional astrophotographers because it images the entire visible spectrum (Red, Green and Blue - the filter looks like clear glass to the human eye) and provides an image that is essentially a map of brightness. This brightness map can be post-processed meticulously to bring out the detail in the RGB colour image (an image formed by digitally combining the Red, Green and Blue exposures together). The exact details of this are beyond the scope of this section and are the domain of post-processing tutorials (see the Tutorials page). Since the regions of the spectrum allowed through by these filters is quite broad, these are called Broadband Filters

Another advantage of monochrome CCD sensors is the ability to use more scientifically-oriented filters - filters that only transmit a particular, very narrow portion of the spectrum. Atoms of elements such as Hydrogen that lie within deep space can be excited by radiation and other neighbouring particles. When these atoms relax, they emit light of specific wavelengths and these are the wavelengths that can be specifically selected out by these filters - called Narrowband Filters. When we speak of atoms relaxing, we in fact mean the electrons bound to these atoms relax from a higher energy level to a lower one. By conservation of energy, this energy has to go somewhere and it is in fact emitted as a photon of light of energy equal to the energy level difference. This is an important concept because different narrowband filters select out different such electron energy level transitions. Hydrogen-Alpha is, as an example, an extremely popular narrowband filter. This is because Hydrogen is the most common element in the visible Universe. Other popular narrowband filters include Hydrogen-Beta, Oxygen-III and Sulphur-II

The above transmission curves demonstrate how narrowband these filters truly are (compared with the previous transmission curves for the broadband filters). In theory, the narrower the bandwidth of narrowband filters, the higher the quality of the filter and the better the resulting image contrast is as it selects out what you want more specifically. Baader, for example, have 35 nm and 7 nm Hydrogen-Alpha filters. One could argue that the 35 nm filter is better because it allows more light through, building image brightness more quickly. This is true, but actually a narrowband filter is built for selecting out a particular wavelength of interest and as a result, the 7 nm proves to be much better, giving a sharper image with more contrast. Higher end filters exist in the market, particularly by Astrodon, who make 5 nm and even 3 nm narrowband filters. These are however, significantly more expensive. Narrowband filters can be used to produce false colour images akin to some produced by the Hubble Space Telescope (e.g. the famous Pillars of Hercules image). The excellent thing about using these narrowband filters in imaging is that one can combine the narrowband images in a multitude of ways to produce colour images of different colour palettes. Moreover, narrowband images such as Hydrogen-Alpha can be combined with a specific visible spectrum colour (in the case of Hydrogen-Alpha, Red) to produce incredible contrast and vividness in that colour channel. 

An important property of narrowband filters is that they overcome most light pollution by their very nature. This happens because light pollution tends to come from sodium or mercury vapour lamps, which mostly emit light of specific wavelengths. Since these wavelengths are actually rejected by certain narrowband filters (e.g. Hydrogen-Alpha), light pollution from these sources is not captured by the CCD sensor. It is also common knowledge that deep space astrophotography is not viable during nights with the full Moon (or close to it), high up in the night sky. Filters such as Sulphur-II and Hydrogen-Alpha however, particularly very narrow bandwidth ones, can be used during these nights. This allows astrophotographers to use these nights to capture some of their data and save the really dark nights for visible spectrum data with broadband filters. Other narrowband filters, such as Oxygen-III, cannot be used during moonlit nights since in the case of Oxygen-III, lie deep within the visible spectrum. 

In these modern times, it can be hard to find truly dark night skies. We are mostly littered with light pollution and this plagues imaging whether or not the Moon is visible. Thankfully, as aforementioned, most light pollution sources work within specific wavelengths and these can essentially be selected and blocked (reflected back out of the system before it reaches the CCD sensor). Filters exist for this purpose and are called light pollution suppression filters. These filters should be threaded at the very front of the optical train that goes into the focuser, reflecting back out the undesired light. Depending on the quality of the filter, these tend to avoid most light pollution ever being imaged and thus strongly enhance contrast, especially in broadband images (visible spectrum). A very high end filter for this purpose is the Hutech IDAS LPS:

As can be seen from the transmission curve above, the filter very specifically rejects light of certain wavelengths and these correspond to the various sources of light pollution common today. Other wavelengths have near-100% transmission and allow for the user to image the visible spectrum including wavelengths common to narrowband filters. It is noted however, that because light pollution suppression filters reject quite a bit of light, it essentially makes the optical system slightly slower at building image brightness (even within accepted wavelengths of light). Moreover, depending on the quality of the filter, it can produce a slight change of colour balance. This slight change however, can be easily corrected in post-processing and is less of a problem for those imaging with monochrome CCD sensors. It is worth noting at this stage that the Hutech IDAS LPS does actually provide the industry's best colour balance with no changes caused. 

4.3 Choosing the Right Filters

Choosing the right filters for the job is not a difficult task. There are numerous options in the market but for good reasons, there are well-established ones that produce incredible results and do not break the bank. In general, filters are divided into two categories - visual and imaging. For astrophotography, we are obviously interested in imaging filters and those are the suggestions listed here. Having chosen imaging filters, we can divide choices in sub-categories:

With the above information, one should be able to make a more informed decision of the filters to buy. The following are recommendations to date (30 / 12 / 2013) of filters for astrophotography. These vary in price and capabilities and are only some as there are of course more in the market. The Baader sets of broadband and narrowband filters are of particular value in terms of performance and price.