After 3 months spent building my observatory, I finally captured the first light frames out of the dome with the setup halfway completed. One of the most challenging tasks was to ensure a proper slaving of the domes gap (shutter opening) which follows the mounts (scopes) direction… especially with higher altitude targets my parameters are still not perfect, but they work at least for some tests.
Setting the Dome Parameters in N.I.N.A.
For the coordination of the domes rotation together with the mount position (slaving), basically some geometric data has to be submitted to the N.I.N.A.`s Dome Options. Essentially, these are:
- the „TELESCOPE POSITION„: the exact position of the intersection of the right ascension and declination axes of the EQ8-R mount (mount center), and the height of this point in reference to the dome base (which is the domes plane of rotation), its location in north-south and west-east,
- the „GEM AXIS LENGTH“: According to the manual: ForAlt/Az-mounts, this should be 0. For an EQ mount, slew RA to +/- 90 degrees, and measure the lateral distance (in mm) from the axis to center of the telescope aperture. The purpose of this setting is to determine what should point to the center of the Dome aperture. If you have a guide scope, you should add half the length from the OTA to the top of the guide scope. For example, if the guide scope mount is 40mm and the guide scope aperture is 60mm, you should add 70mm to GEM Axis Length.
- If a SBS-setup (side by side) is used with 2 parallel scopes on the mount, also a „LATERAL AXIS LENGTH“ must be determined, specifying the offset (in mm) of the center of the OTA from the mount axis. A positive number for offsets to the right relative to the RA and DEC axis.
A short recap on the Question Mark Region
The target of choice was of Sh2-170 in Cassiopeia – a first test in narrowband should reveal details of this emission nebula.
Already in March, 2025 I´ve captured the upper part of the „question mark“ in narrowband. The core area of NGC 7822 revealed stunning details:
Sh2-170 is a nearly circular H-II region in Cassiopeia, physically associated with the very young open cluster Stock 18. The diameter is 20‘ or 55 light-years. Its ionization is dominated by the O-type system LS I +64 11. Sh2-170 is located at a distance of approximately 9,500 light-years, based on Gaia DR3 parallax measurements of its associated open cluster. In wide-field images, Sh2-170 and the nearby NGC 7822 complex often resemble the “dot” of a question mark, suggesting a physical association. In fact, NGC 7822 is much closer—at roughly 3.500 light-years—, so the alignment is purely line-of-sight.
In general H-II regions are volumes of interstellar gas in which hydrogen is predominantly ionized; the atoms have been stripped of their electrons, leaving a plasma of free electrons and protons (H+).
Imaging the Dot
At latitude 47◦N (Graz), Sh2-170 is circumpolar and culminates high. For imaging, Hα typically dominates, while O-III and S-II are present but fainter; balanced integration (or Hα-weighted acquisition) and rigorous subframe selection by FWHM/Eccentricity/Star count (with PixInsights SubframeSelector) should help to control noise in the channels.
The narrowband imaging sessions will start with the Hα-filter and will be followed by O-III and S-II (all Optolong 3nm). At least 3 hours of each filter (36×300″) is planned to be the minimum. RGB-stars may be captured as well.
Object Data
| Type | EN (Sh2-170), OSC (Stock 18) |
| Distance | 9.500 light years |
| RA | 00h 01m 29s |
| DEC | +64° 39′ 03″ |
Equipment
| Scope | 10″ Carbon Newton operated at f/3 (750mm) using the Starizona Nexus 0.75 Reducer |
| Mount | Skywatcher EQ8-R |
| Camera | ZWO ASI 2600 MM Pro |
| Filter | EFW (7 x 2″ Optolong, L, R, G, B, H-alpha, O-III, S-II) |
| Guiding | Skywatcher ED50, ZWO ASI 120 MM Mini |
| Sequencing | N.I.N.A. |
| Misc. | CAA, EAF |
Exposures (actual status)
Session 1 (H-alpha) – 11.10.2025
Although sky quality during the first half of the night was suboptimal (FWHM 2.5–6.5; median 4.59), I continued the session to verify dome–mount synchronization: for the first time, the dome (gap) successfully tracked the mount’s pointing.As a first-light validation, 34 of 43 H-alpha sub-frames were retained for integration with PixInsight’s WBPP, using the Subframe Selector metrics (FWHM, Eccentricity, and Stars). The actual status represents 34 exposures (with 300s each) in H-alpha (WBPP with 20 flats, 20 darks) whereby sensor temperature of the ZWO ASI 2600 MM Pro was at -20°C and gain/offset at 100/50.
Session 2 (O-III) – 24.10.2025
It was actually the first time, that the whole workflow in N.I.N.A. worked completely remote! Within N.I.N.A.`s Advanced Sequencer it was possible to begin with the connection of all devices, which was followed by a parallel instruction set for unparking the mount, cooling the camera, opening the flat panel, moving the EAF to a predefined position near focus and forcing the dome to find home and allowing it to sync with the mount. After a filetr change (to Luminance) the AF (autofocus) completes the first part of the starting sequence.
In order to calibrate the guiding in a favorable position, the scope is instructed to slew to an Azimuth of 180° and an Altitude of +43° to point towards the ecliptic in southern direction (this time the scope was perfectly accompanied by the dome). After calibration at a location with maximum movement, it is now time to start the target sequence.
The scope slews and centers on the target (SH2-170 again) and changes to the narrowband O-III filter. The foilter change triggers an AF again und the sequence starts with the first 300s subframe. After capturing the first 4 subframes some clouds moved through (frames 5,6,7 had to be selected as trash) the following frames were captured perfectly, with a proper guiding performance and quite good conditions.
At 10:54 pm the northern transit was reached and N.I.N.A. performed an automatic meridian flip, bringing the scope from the western to the eastern side of the pier. Centering again, AF worked well, just PHD2 made problems (again): „the pulse guide command could not reach the mount – guiding therefore will be ineffective …“ The first attempt to solve this problem by disconnecting all devices and starting them again didnt work. After the connection clouds moved in and – completely out of character for me – … i paused the session.
In the end, 35 subframes in O-III (representing nearly 3h of data) were collected within the first half of the night. This should be enough for a first processing of the target using an HOO-approach.
Processing Notes
After taking 20 O-III-flats the next morning the frames were analysed using PixInsights SubFrame Selector. A FWHM-median of 4.24 characterized the conditions: seeing was moderate, with a weak start followed by an intermittently usable plateau. During the best phase FWHM was ≈ 4.0.
A slight tilt in the image train and the dome not yet being thermally equilibrated at the beginning of the session may have contributed to the early degradation and some eccentricities.
HOO-Bicolor Workflow
After a processing pause lasting several months due to the „Observatory Obsession“ it actually feels like starting from below zero when it comes to Narrowband processing in PixInsight. Reason enough to review all the steps in detail. The following documentary are my processing notes while tryin to again become familiar withe the right steps in a reasonable order …
Star Alignment, Dynamic Crop & AutoDBE
Since the optical train (including my ZWO CAA Rotator) was reassembled after the first session, the rotation of my O-III captures did also slightly change. The StarAlignment process in PixInsight (PI) was used to align the two channels again and Dynamic Crop was applied to receive a common overlap for all following steps.
AutoDBE is an automated wrapper around PixInsight’s DynamicBackgroundExtraction provided by a comfortable SetiAstro script. It places background samples to model gradients and vignetting, and subtracts or divides that model from the data. The result is consistent gradient removal.
Balancing the two Channels with LinearFit
LinearFit equalizes the intensity scales of the channels—compensating for differences in filter bandpass, quantum efficiency, and atmospheric transparency—thereby producing more stable color blends and reducing color fringes around stars. For narrowband combinations, choosing the weakest channel can keep the Hα channel from overwhelming the others (see BRACKEN (2022), p. 177).
After alignment, cropping and DBE, I therefore equalized the channels intensities using a median match: median(Hα)=70.175 and median(O-III)=82.428. Surprisingly the O-III median showed to be higher and – as we will see in the MAD chapter – also the cleaner channel. As we do not want the weaker and noisier channel (in my case the Hα) to be pushed further, therefore the O-III channel was downscaled instead of pushing Hα to the level of O-III. In LinearFit I choose Hα as the reference for O-III (factor s = median(Hα)/median(O-III) ≈ 0.85135). The Statistics-process in PI revealed the used mean figures and also told me about the background noise of the image.
Background Noise (MAD)
In general the MAD (median average deviation) values (Hα = 2.163, O-III = 1.894) are robust proxies for background noise; if needed, multiplying by 1.4826 yields Gaussian-equivalent σ, but their ratio already is a good choice for weighting the channels. Because O-III’s MAD was lower, it was considered to be the cleaner channel in my linear images. These findings are important also for the next processing step, which is responsible for the fine details of the final image.
Creating a synthetic Luminance
A synthetic luminance is a grayscale master built by linearly combining the Hα and O-III emission-line channels to concentrate the highest-SNR structural signal independent of color. The Luminance channel is then applied to the color composite (via LRGB-Combination) so that fine detail and contrast are driven by the clean luminance while chrominance stays flexible and less noisy. The channel weights for creating the Luminance are chosen based on the MAD figures derived before:
Using MAD as a noise proxy, I take σHα ≈ MADHα = 2.163 and σO-III ≈ s MADO-III = 0.85135×1.894 ≈ 1.612.
with the inverse variance 1/σH2 beeing normalized (weightHα +weightO-III = 1) it follows 0.357 for Hα and 0.643 for O-III
In PixelMath therefore the expression L = 0.357*Hα + 0.643*O-III will yield the desired synthetic Luminance with SNR-based weighting. The weighted Luminance was processed with BlurXTerminator and NoisXTerminator to apply Deconvolution and Noise removal. After sharpening a bit (with UnSharp) also the darker structures were enhanced (DarkStructureEnhancement). In all the frames the stars were separated from the nebulae (using RC Astro`s StarXTerminator) for specific processing and a GHS (generalizedHyperbolicStretch) was applied to all nebulae frames.
Within PixelMath I allocated R = H, G = 0.1 x H + 0.4 x O and B = O to reach a RGB-Image out of the stretched Hα and O-III channels. This arrangement had the focus on the O-III emission, which should be revealed in blue. An ABE (Automated Background Extraction) was used to establish a neutral background to the RGB-Image as well as the Luminance Image.
Combining RGB and matched synthetic Luminance
The challenge is now to integrate the Luminance data into the RGB image without producing a „Luminance overdrive“. Brightness and Contrast of the Luminance must therefore be aligned with the RGB figures.
Working with CIE L*a*b* makes it possible to separate luminance (L*) from chromatic components (a*, b*). This decoupling allows me to enhance sharpness and structural detail in L* while preserving color fidelity in a* and b*. So the first step ist to extract the Luminance from the RGB image. With PI`s ChannelExtraction process I extract therefore L_rgb representing only the Luminance of my combined RGB image.
With LinearFit the intensity of my synthetic Luminance is now aligned with the L_rgb figures. Within the 3 frames below one can see the RGB-Luminance, the unmatched synthetic Luminance and the matched Luminance after the LinearFit process:
If we would now just simply add the RGB and the matched synthetic Luminance channel (1:1) the colors again would become pale, therefore mixing the Luminance has to be controlled. I generally add Luminances using the LRGB-Combination Process with a smooth weighting of the Luminance channel – this time 0.6 has delivered quite good results in adding contrast and details and not washing out the chroma.
I can name no explicable reason why I – in most cases – tend to use the Hubble-Palette, when it comes to assign colors to narrowband data… Also while processing this target, the Hubble-Palette is for me the most appealing view. I therefore used PI`s NBNormalization in order to assign HOO (preserving the brightness of Hα and slightly boosting the O-III emission. Finally a CurvesTransformation made the image pop a bit more.
Actual state of processing with blurred and displaced halo stars …
The actual version reveals mostly the central area with the O-III emission ionized by the hard radiation of the star cluster Stock18 – especially BD+63 2093p. The outer Hα-regions are according to my opinion a bit „ragged and weak“ when I compare my image with others – most of them captured the 20´ diameter target with longer exposures, better skies and additionally also with S-II filters, revealing more structures at the outer boundaries. Also „something“ happened with the stars (which were combined out of the narrowband data using the Forraxx-process). The spikes nearly dissappeared and also some halos shine through as if the positions were shifted somehow to south-east…
…searching for the causes of the star problem
…
