Articles & FAQ

Articles & FAQ

NIR Filters

Astrodon NIR Tri-Color CCD Imaging Filters



Amateur imagers are producing incredibly deep and detailed tri-color images of deep sky objects that rival or even exceed the quality of film images taken with professional equipment only 20 years ago.  CCD cameras have quantum efficiencies (QE) approaching 90% that can bring out detail in faint objects.  CCD detectors as large as 35 mm film with 12 million pixels are readily available to amateur imagers. High resolution images are taken with plate scales of 0.4 – 1.0 arcseconds/pixel on modest backyard telescopes with apertures ranging from 25 – 50 cm.  Mounts can be actively guided to <0.2 arcseconds all night.  Detectors in these imaging cameras are made from silicon (Si).  They cover the ultraviolet (UV <400 nm), visible ( VIS   400 – 700 nm) and short-wavelength near-infrared (NIR) spectral regions up to about 1100 nanometers (nm; 1100 nm = 1.1 microns), as shown in Figure 1.  Professional astronomers use expensive CCD detectors made of different materials for NIR measurements that extend beyond the Si range.  For example, J- and K-band images are taken at 1250 and 2200 nm, respectively. The TMSS survey carried out 30 years ago was done in the NIR.  The “T” in TMSS means 2 (Two) microns (2000 nm).  The readily available photometric I filter covers the NIR region shown in Figure 1.  NIR commonly refers to the region from 700 to 2800 nm and infrared (IR) for longer wavelengths.


Figure 1. Spectral regions covered by various filters, including red, green, blue, NIR-blocked (L) and unblocked (C) luminances.  The QE sensitivity curve of the Kodak KAF3200ME detector is also shown.  The RGB, L and C curves represent Astrodon E-Series imaging filters.

Tri-color images are typically made by placing red (R), green (G) and blue (B) filters between the telescope and the monochrome CCD camera. They are combined in software to form the RGB color image.  The color images are generally taken at lower resolution by grouping (binning) pixels in the CCD camera, thereby permitting shorter exposures.  The lower resolution RGB image is then used to “colorize” the deeper, higher resolution (unbinned) luminance image.  This forms a high resolution LRGB image. 

There are two types of luminance filters; NIR-blocked (L) and unblocked (C = Clear), as shown in Figure 1.  L is designed to match the spectral coverage of the RGB filters. One school of thought is that this is the best combination for color images because of the spectral matching.  The other school of thought is that one wants all the photons possible.  C adds about 40% more signal from the NIR (using the Kodak KAF3200ME detector, for example) to produce a deeper luminance image. The question is how the additional NIR signal relates to the RGB image that will colorize it.  This is still debated.

Lastly, narrowband filters bring out detail in nebula and supernova remnants from the emissions of hydrogen (H-alpha or H-a) at 656 nm, oxygen (OIII) at 497 and 501 nm, sulfur (SII) at 672 nm, and others.  It is common to add H-a into the RGB of an emission nebula to enhance structural detail.  All of these colors and emission lines lie within the VIS spectrum. 

The NIR region remains largely unexplored by amateur imagers. Perhaps this is due to the impression that the Si CCD camera is not sufficiently sensitive in the NIR.  Perhaps it is not clear what can be imaged in the NIR.  Narrowband imaging has only recently become popular, allowing imagers to take images from backyard equipment (S&T, August, 2005, page 112) like the Hubble Space Telescope’s “Pillars of Creation” (Messier 17).  NIR imaging may follow this recent narrowband evolution and become another tool in the imager’s toolkit.  So, what can NIR imaging add?

  • Produce tri-color images entirely in the NIR.
  • Examine objects highly obscured by dust in the plane of the Milky Way, such as the IC342/Maffei group, that are difficult to image at VIS wavelengths.
  • Explore faint, extended red emission (ERE) nebula.
  • Uncover stars and other details in nebula that are obscured by bright VIS emissions from H-a, SII, OIII or other elements.
  • Mix NIR with RGB of globular clusters to enhance the appearance of cooler stars.
  • Minimize terrestrial light pollution from sodium and mercury street lamps and oxygen skyglow for imagers living in suburban or city locations.



A Santa Barbara Instruments Group (SBIG) ST-10XME camera employing the Kodak KAF3200ME detector was used on an RC Optical Systems (RCOS) 12.5” Ritchey-Cretien 12.5” reflector telescope at f/9 (2800 mm focal length) producing an unbinned plate scale of 0.5”/pixel.  A Software Bisque Paramount ME German-equatorial mount was used to guide the system with the internal guide detector in the ST-10XME.  Diffraction Limited’s MaximDL was used for camera control and guiding, and CCDWare’s CCDAutoPilot2 was used for automated data acquisition sequencing.  Astrodon E-Series RGB and NIR filters were used in an SBIG CFW10 10-position filter wheel.


One of the first efforts to do tri-color NIR imaging was reported by Paolo Candy in (Astronomy, May, 2002).  He used a series of long-pass NIR filters shown in Figure 2 that block shorter wavelengths and pass NIR light.  These are made with readily available colored glasses (e.g. Hoya, Schott).  He obtained different bandpasses by subtracting these filters.  For example, a bandpass covering 700 – 800 nm resulted by subtracting IR2 from IR1. 

Figure 2. NIR cut-on filters used by P. Candy for tri-color imaging.

A different approach was taken for the present work with Astrodon NIR imaging filters.  Subtracting images was avoided to minimize noise.  Instead, more efficient dielectric-coated filters were designed and produced to have the desired bandpasses, as shown in Figure 3.

Figure 3. NIR tri-color imaging filters (Astrodon).


These are analogous to LRGB filters.  The luminance filter is a NIR long-pass filter starting at 700 nm. NIR1, NIR2 and NIR3 correspond to B, G and R in order of increasing wavelength, respectively, and are correspondingly color-coded Figure 3.  NIR3 is again a long-pass filter, whose final shape is most strongly defined by the decreasing QE of the specific CCD used. For example, the bandpasses shown in Figure 4 are produced if the filter transmissions shown in Figure 3 are multiplied by the QE curve of the KAF3200ME detector shown in Figure 1.

 Figure 4.  Astrodon NIR filter bandpasses combined with KAF3200ME QE sensitivity.

Similar profiles would be expected with the larger Kodak KAF6303E detector.  Back-thinned detectors (e.g. E2V) will result in even more sensitive NIR2 and NIR3 bandpasses due to their flatter and more sensitive QE in the NIR.   

White-Point Balance and Sensitivity

The Astrodon RGB E-Series filters shown in Figure 1 were designed for a generic class of Kodak blue-enhanced, E-Series CCDs to produce a G2V (sun-like) white-point balance with equal exposures. Thus, the fluxes among the bandpasses have been roughly equalized when accounting for the QE curve of the CCD and the solar photon flux. The G2V white-point has been adopted by most of the imaging community (S&T, December, 1998, page 142).  This means that equal exposures made through these R, G and B filters could be combined in software with weights approximately 1:1:1, such that a G2V star will appear white.   Since one exposure time can be used for each color to achieve comparable signal-to-noise, only one dark exposure time is needed.  This simplifies calibration and reduces file storage.  With larger detectors such as the KAF6303E and KAI11000XM producing 10-20 MByte files, file storage becomes even more important.


A similar approach was taken with the design of the NIR filters based upon a G2V white-point, as best as possible, considering the strongly decreasing QE in the NIR.  Tests on a G2V star (SAO83953) with the ST10XME and 12.5” RC resulted in weights for NIR1, NIR2 and NIR2 of 0.85:1.00:1.21 for equal exposures.  The NIR2 (Green analog) filter has 40% of the G filter’s integrated signal, as also measured on SAO83953.  As expected, the NIR filters will require more image exposure.






The NIR tri-color filters are unique in that they provide an opportunity for the amateur imager to explore objects that cannot be easily imaged with conventional RGB filters.  The IC342/Maffei (McCall and Buta, AJ, 1999) group objects are excellent examples. In fact, with all the discoveries of galaxies, it was only as recent as 1968 that Paolo Maffei discovered these highly obscured galaxies that now bear his name located within 1° of the plane of the Milky Way.  If the giant elliptical galaxy, Maffei 1, was not obscured by 5 mags from so much dust, it would be one of the brightest objects in the night sky, along with M31 and M33! These objects are just below the Double Cluster (NGC869, NGC884) in Cassiopeia.  Figure 5 is a NIR tri-color image of Maffei 1.  This is a relatively short exposure, but Maffei measures at least 10’ in its longest direction from this image.  It is one of the few amateur images of this object.



 Figure 5. NIR Tri-Color image of the nearby giant elliptical galaxy, Maffei 1. 02h 36.3m 59deg 39m SBIG ST10XME at -15C on RCOS 12.5” RC at f/9 taken near Bend, OR at 4500’ on August 3, 2005. The total exposure time was 100 min.  The image is ~18’ wide.

Maffei 2 is slightly dimmer than Maffei 1 by ~ 1 mag, and is a barred spiral galaxy with a core having an unusual nuclear burst of star formation.  Only about 1% of its blue light reaches us due to scattering from dust in the galactic plane.  Figure 6 is a NIR tri-color image of Maffei 2 where the spiral structure is readily apparent.

Figure 6. NIR Tri-Color image of the nearby barred spiral galaxy, Maffei 2.  02h 41.9m 59deg 36m. SBIG ST10XME at -15C on RCOS 12.5” RC at f/9 taken near Bend, OR at 4500’ on August 4, 2005. The total exposure time was 100 min.  The image is ~18’ wide.

The difference in brightness between VIS and NIR luminances for Maffei 2 is shown in equal single exposures in Figure 7.  As indicated above, the NIR flux for this system is approximately 40% of the VIS flux.  Therefore, the NIR image on the right would have to be multiplied by 2.5 to more properly compare brightness of Maffei 2. This emphasizes how much
VIS light is scattered by dust for these objects lying so close to the plane of the Milky Way. 


Figure 7. Visible NIR-Blocked Luminance (left) and VIS-Blocked NIR Luminance (right) of Maffei 2 showing how much brighter this dust-obscured object is in the NIR.  Single 10 min exposure of each.

The NIR and VIS luminances can be compared and even subtracted to show differences in galaxies.   These are compared in Figure 8 for M33, the Pinwheel Galaxy in Triangulum .  Notice the brighter NIR core (right) showing less contrast with the dark lanes.  In contrast, notice the brighter HII regions in the VIS image (left).  As an aside, these differences lie at the heart of the debate regarding the inclusion of NIR signal into the high resolution luminance and colorizing that with RGB, since the RGB directly relate to the image on the left.

Figure 8. Visible NIR-Blocked Luminance (left) and VIS-Blocked NIR Luminance (right) of M33 showing significant differences I the core and HII regions.


The NIR image was subtracted from the VIS luminance in MaximDL.  The results are presented in Figure 9.  This shows the relatively NIR-rich core (darker areas) and brighter VIS HII star-forming regions.  VIS and NIR images can be taken without refocusing, since all of the filters are all parfocal.

Figure 9. NIR-subtracted image of M33 from Figure 8.  Darker areas are richer in NIR.


The ability to penetrate emission nebula can be seen in M17, The Swan Nebula, in Sagittarius shown in the mouseover in Figure 10.  Notice how many more stars are present in the NIR image.  There is considerable emission from H-a, OIII and SII in M17.  However there is still considerable signal outside the VIS region, which could suggest that we are observing light scattering or even absorption of light and re-emission at longer wavelengths (lower energies).


Move Mouse Over Image to See VIS Luminance

 Figure 10. Visible NIR-Blocked Luminance (left) and VIS-Blocked NIR Luminance (right) of M17 showing that NIR can penetrate further into emission nebula and detects stars hidden at visible wavelengths.


This emission nebula may be contrasted by a comparison of the Dumbbell Planetary Nebula, Messier 27.  Figure 10 provides an unusual look at M27. It is a combination of

H-a and OIII for the nebula and RGB for the stars.




Figure 10.  Deep Ha, OIII, and RGB image of the Dumbbell Nebula in Vulpecula.

Figure 11 presents a NIR tricolor using the filters shown in Figure 3, where NIR1, NIR2 and NIR3 are mapped as blue, green and red and IRLP is used as the luminance.  There is only a small amount of residual nebulosity shown in the NIR and again, many more stars are apparent.


Figure 11.  NIR Tricolor Close-Up of the Dumbbell Nebula, M27, taken with the same orientation as Figure 10.


Lastly, the same NIR approach can be taken with globular clusters.  Figure 12 is a comparison of M3 taken in LRGB and NIR tricolor.



   Figure 12.  M3 Globular Cluster taken with LRGB (left) and NIR tricolor (right)


The colors are more distinct in the VIS image (left) due to including hot blue stars from the Blue filter.  This is less color distinction among the stars in the NIR image (right) because the blackbody curve is flatter in the NIR.  Nevertheless, there is still considerable detail present in the NIR image. Because the filters are all parfocal in most telescope systems, interesting possibilities emerge.  One can develop any combination of filters – B, G, R, NIR1, NIR2 and NIR3.  So, a tricolor image can be developed with BGNIR2 to capture hot blue stars and cooler red stars moreso than the traditional red filter, for example.



As with narrowband emission line filters, NIR imaging provide yet another tool for to explore features that are obscured by dust or hidden by bright emissions.  They provide means to capture images where light pollution from street lamps may otherwise be a problem.  They provide the means to produce interesting combinations of both VIS and NIR light.  They allow imagers to examine familiar objects in a very different “light”.