Astrodevice logo
Dual filter imaging

Dual filter imaging

Usually the cosmos is photographed using a single optical filter at a time. It can be, for example, a nebula filter, a narrowband filter, a light polution filter, a filter that transmits only one of the primary colors, but the obvious seems to be that it is always one filter for a single frame. Although this is a common practice, nothing prevents us from trying to break out of this pattern. In this article, I will show that it can give surprising results.

The Idea

Let me start by saying that when I was developing a filter drawer for ZWO large-sensor cameras dedicated to the RASA 8 telescope, I introduced into the design the possibility of using two filters at once. This solution allows optical backfocus adjustment by screwing in an additional UV/IR filter. This is potentially useful when using broadband filters such as R, G or B in the drawer. You can read more about that there here.

However, the ability to screw in an additional filter opens up many more possibilities. After all, if a complex narrowband filter is screwed into the Astrodevice drawer, with the help of additional filters we should be able to cut out or pass through only specific parts of the spectrum. But is this possible in practice?

Color vs monochome

I was prompted to test this idea by a trend I observed among my astrophotographer friends. Many of them, using a Celestron RASA telescope, shoot with a ZWO ASI 2600 MC Pro color camera and a complex multiband IDAS NBZ filter. This filter was designed specifically for high-speed telescopes just like the RASA. The glass transmits only the O-III and H-alpha spectral lines, allowing spectacular color images to be taken without having to juggle two separate narrow-band filters. When using a color camera, this is a gigantic time-saver. However, the price for this convenience is the quality of photography.

To understand this problem, consider what makes a color camera sensor different from a monochrome camera. Well, while the sensor itself is identical, the top of the sensor in a color camera is coated with a special layer, the so-called Bayer filter. Each pixel is covered with a blue, green or red coating. This treatment allows only light of a certain color to pass through. The pattern of colors is regular and strictly defined. For example, in the aforementioned ZWO ASI 2600 camera, a Sony IMX 571 sensor with an RGGB array was mounted. Red, green and blue filters are alternately applied, with the pattern enforcing a twofold predominance of the green color in the entire pixel pool. When light enters the camera, only 1/4 of the pixels register red, 1/4 blue and 1/2 green. Using a special mathematical process called demosaicing, by examining neighboring pixels, the algorithm guesses what the intensity of the missing components should be in the pixels that didn’t physically register them. This all works very smoothly, but from a measurement point of view it is just a clever mathematical trick. For the intensity of the missing component of light, determined at a given location, is just a calculated statistic based on an assumed a’priori probability characteristic. It is not the truth but the result of mathematical modeling. It is merely an assumption.

Bayer pattern on sensor

Sensor coated with Bayer filter.
Source: Wikipedia

Let us now turn to astrophotography. Suppose we want to record the emissive light of a nebula in the H-alpha line. This line reproduces a wavelength of about 656 nm, which translates into the deep red color characteristic of many color cosmic images. However, if you capture an image with a color camera, only 1/4 of the pixels will register the light of interest. The rest will be blind. This means that for a 26 Mpix camera, we actually have a 6.5 Mpix device. From 19.5 million pixels, the light reflects and is irretrievably lost. For space photography, where every photon is precious, this seems to be a real waste, both of time and resolution. Moreover, there are very few interesting sources of green color in space. So if someone is interested in the emission spectra of nebulae, half the camera physically captures almost nothing of interest. That is why traditionally astrophotography images are taken with a monochrome camera, putting on a suitable filter. Then each pixel captures a selected spectrum and we always use the full resolution of the device.

Now let’s assume that someone who normally uses the color camera set-up, wants to take advantage of the monochrome camera’s benefit and take separate images for the O-III band and a separate one for the H-alpha band.

However, he has only the mentioned IDAS NBZ filter and does not want to invest in additional narrow-band filters. What then?

To test this, I decided to perform a simple experiment.

Dual filter imaging

My idea was based on a simple assumption, which I mentioned at the beginning of the article: if I cut certain bands from the multiband filter with a second filter, I will be able to record only the spectral lines of interest as if I had a dedicated narrowband filter.

Let’s look at the spectral charactristics of the aforementioned IDAS NBZ filter. As already mentioned, it only transmits H-Alpha and O-III lines. The former is deep red, while the latter is turquoise. It’s worth mentioning here for the sake of accuracy that doubly ionized oxygen emits light not in one line, but in two close frequencies, which corresponds to two closely spaced spectral lines – 500.7 nm and 495.9 nm.

We can also learn important information from the spectral profiles containing the characteristics of Astronomik’s Deep-Sky Red and Blue filters. Looking closely, one can see that the wavelength corresponding to O-III lies in the blue spectrum and is transmitted at a level comparable to the rest of the blue spectrum. From this graph we can also draw the fact that the filter transmits only about 93% of the light near the spectral line of oxygen, so when we decide on an experimental setup, we will lose about 7% of the data by scattering it through an additional color filter. Nonetheless, a few percent is not the few tens of percent we lose on the Bayer matrix.

Obtaining such selected spectral profiles turns out to be quite simple. Astrodevice M87 series drawers designed for ZWO cameras are equipped with an additional 2″ front thread. So I screwed an IDAS NBZ filter on the front and decided to use R and B color filters in the drawers. This made it easy for me to control whether the hydrogen or oxygen spectral line was passed through during a photo shoot by changing filters.

The test lasted three hours. I took 90 H-alpha and 90 O-III photos, each one lasting 180 seconds. 

In principle, the taking of photographs itself was no different from a standard process, that is the one that would involve a single filter. The IDAS NBZ filter, screwed in from the front, was mounted throughout the exposure and its presence was completely transparent to me, not interfering with or hindering anything.

After the three-hour session, it was time to preview the results.

And here’s the final, post-processed result of the experiment:

Pacman nebula

Importantly, I found that imaging with two filters does not introduce any additional color distortions that would significantly affect the final result. One indicator that is a measure of possible deviations in the spectral distribution is the ability of the resulting stellar image to match the theoretical black body model. In the case of the test carried out, this match succeeded quite well. Excluding the intensity ratios of the oxygen and hydrogen components, the distribution is actually perfectly linear.

Calibrated stars

Summary

By placing a second filter in the optical path, we can selectively cut the appropriate range of transmitted light. Thanks to Astrodevice’s filter drawers, such a procedure is possible even when imaging with the help of a telescope with a very small backfocus, such as the Celestron RASA 8. Despite the small amount of space for placing any additional accessories, I was able to use the IDAS NBZ multiband filter in a monochromatic camera and make it work like two separate narrow-band filters.

Scroll to Top