"High-Speed, Dithered, Dual-Shot Astrophotography"

By Minos K

| Introduction |

Messier’s celestial object #16 and ‘The Eagle Nebula’, a place of star-forming clouds and ever expanding molecular dust. In its core, three dark and gaseous pillars, of creation and stars. Emerging like the ‘Hand of God’, these pillars of creation point towards the open cluster M16, a swirling array of huge balls of fire. Both are situated in the next inner spiral arm towards the center of our Galaxy, 7,000 light-years away from Earth.

| Methods |

Location: Sia Space Observatory, Cyprus | Latitude: 34° 57' 2.32" North | Longitude: 33° 24' 23.71" East | Altitude 300 meters.

-Each night, I aimed to capture these ancient photons with 1000 exposures (what I'm going to call a 'KiloStack'), 4 seconds each exposure, with the camera set to a high gain (84%). I used the fantastic Sharpcap 3.0 program to do my initial polar alignment with and then into a straight capture and calibrated each light frame with matched dark frames (acquired immediately after kilostack) in DeepSkyStacker to reduce amplifier glow noise.

-Polar alignment confirmation and guiding was performed within PHD2 and dithering was set on a timer for every 2 minutes (approximately every 30 frames) with PhD2 Dither App (as Sharpcap does not currently have built-in dithering functionality and the ASCOM camera driver has no black point offset to use effectively for proper histogram conformation with short exposures in other software that does i.e. APT). Dithering scale was set at 12-pixels minimum (two-star-widths, spiral formation). I would then reject 33% of exposures in the Kilostack that would mostly contain star trails due to dithering commands.

-Framing and focus was done by manual camera rotation and adjustment in a 1.25" compression ring fixed in a dual speed R&P focuser. Target centrality confirmed using Astrotortilla.

-Rather than approaching this image from a traditional perspective, where one shoots LRGB channels, or narrowband channels using a single monochrome camera sensor, I took a different approach, thinking outside the box - but with a twist. Instead, I employed two identical cameras - the excellent Altair Hypercam IMX178 - one in mono and the other a one-shot colour (OSC). By doing this, I could save heaps of time by acquiring all three colour channels in one-shot instead of independently, and then by using the super-sensitive mono version to acquire the Luminance channel, I could ultimately boost the colour signal and sharpness in the final image.

-Filters were a very important choice, and this is were the twist comes in. After much experimentation and deliberation, I decided to use the Baader Moon & Skyglow for the Altair 178C, as this is a wonderful filter for good white balance (tames the green Hulk in OSC's) and good skyglow reduction (although I may need to purchase a Baader Semi-Apo next time to reduce the purple fringing inherent in my doublet) and for the Altair 178M and Luminance channel, I decided to go with a more aggressive filter, the Altair CLS-CCD, which is a very interesting and effective dual-band filter that manages to isolate all of the desirable narrowband emmissions of Hydrogen-Alpha & Beta, Sulphur-II and Oxygen-III in one-shot instead of independently, whilst blocking absolutely everything else. I decided that this filter would offer me these important celestial wavelengths of light in one-shot, again saving time and effort in their acquisition.

-Stacking was performed in the super user-friendly and powerful-when-setup-correctly DeepSkyStacker program:

-> Stacking mode: Intersection Standard

-> Alignment method: Automatic

-> RGB Channels Background Calibration: No

-> Per Channel Background Calibration: No

-> Method: Median Kappa-Sigma (Kappa = 2.00, Iterations = 5) for both Lights and Darks

-> No Offset

-> Dark: 1 frames (single master average obtained from 23 dark exposures acquired after a Kilostack in Sharpcap) with no Hot Pixels Detection and Removal (dithering and sigma-stacking takes care of this)

-> No Flats

-> No Cosmetic Correction

-> FITS DDP Settings - Generic RGGB with AHD interpolation

-> All kilostacks were registered and then filtered at 67% of best frames (to eliminate dithers) with frame #1 used as the reference frame (thanks to Astrotortilla)

-Processing was performed exclusively in Pixinsight and involved the following workflow:

-> StarAlignment of Kilostacks

-> DynamicCrop

-> Image integration of Kilostacks

-> RGB extraction from OSC integration

-> AutomaticBackgroundExtraction on LRGB channels

-> LinearFit of all channels to strongest signal - Red

-> ChannelCombination for RGB

-> LRGBcombination for Luminance

-> BackgroundNeutralisation

-> ColourCalibration

-> Deconvolution

-> MultiscaleLinearTransform on Luminance and ChrominanceRestore

-> HistogramTransformation x 3

-> LocalHistogramEqualisation

-> MorphologicalTransformation

-> CurvesTransformation - Colour Saturation and Luminance

-> SCNR - Taming the "Green Hulk"

-> Masked LocalHistogramTransormation

-> DynamicCrop for framing

-> Annotation

| Discussion |

Acquiring colour from a OSC with a Bayer matrix/CFA might not be as good as using individual interferometric colour filters with only a mono sensor, but at the resolution I'm sampling (best @1.9"/pixel), you'd be hard-pressed to tell the difference and it only took me a fraction of the time. If you're thinking of employing this technique on high-resolution, high-focal lengths, you might wanna stick to traditional LRGB methods where quality expensive filters can shine through.

84% gain on the camera was selected as the estimated lowest read gain (calculations beyond the scope of this report), but I was not happy with the banding and noise in the CMOS sensor. Although most of this noise got clipped when sigma stacked, experimentation with lower gain settings around 30% produced a much, much visibly cleaner signal with a lower noisebed. Further investigation is required, but I propose to use lower gain settings in the future with this sensor, not only with regards to lower noise, but also to boost dynamic range.

As for the argument of high-speed vs long-exposure deep-sky astrophotography, consider the following logic; In a single 300 second exposure, one can capture X amount of astrophotons. In comparison, if one were to shoot 300 x 1 second exposures, then by subtracting a given amount of sensor downtime between exposures, say for the sake of argument - 10 seconds in total from 300, then one can acquire for example 290 seconds worth of X astrophotons in 300 frames.

Now, why would one want to do high-speed imaging instead of the classical long-exposure?

-Well, for starters, the starlight signal will not saturate anywere near as much as it would clip in a long-exposure.

-Second, noise reduction with sigma-clipping (standard-deviation outlier rejection) will be much more effective with hundreds or thousands of frames versus a few tens of frames.

-Thirdly, capturing so-called "faint-photons" has nothing to do with long-exposure photography, as those photons are travelling to Earth irregardless of whether you acquire them in a single 300 second exposure or 300 - 1 second exposures (one can argue though that if the frequency of these faint photons is indeed very, very low, they will get sigma-clipped when stacked in the multi-exposure approach). Indeed, I believe I've captured "enough" faint nebulosity above using the high-speed approach.

-Fourth, super-accurate guiding for long-exposure is no longer a issue.

-Fifth, one can dither less sparsely rather than every frame or every other frame, as one is acquiring hundreds to thousands of exposures in total.

| Conclusion |

To summarise, this is a multi-frame (>6000 selected from >9000 ~67%), multi-night (9-nights - 6 for luminance, 3 for RGB), high-speed (4 seconds exposures), two-camera (Altair Hypercam 178M & 178C) and two-filter (Baader Moon&Skyglow, Altair CLS-CCD) approach.

I very much like this method I've come up with, and apologies to anyone who might have done this before, but I believe I'm the first to try this specific amalgamation of approaches with the above logic in mind. Therefore in that respect, I would like to coin this method, my method, as "High-Speed, Dithered, Dual-Shot Astrophotography". Dual-Shot was conceived on the premise that one-shot is for all the RGB signal to be used for Colour, and the other one-shot is for all the desirable narrowband emmissions to be used as Luminance.

Finally, I would like to dedicate this photo to my daughter Gia, who turns 1 year-old today. Little girl, you are made of starstuff, that has floated around the cosmos for eons, only to recombine into a living human here on Earth, that has given me more joy and love than anything else has ever done in this life. Live long and prosper baby, love, your Dad x