Sh2-105 | C27
Sh2-105 | C27, Crescent Nebula - Click here for full resolution
The Crescent Nebula (also known as NGC 6888, Caldwell 27 and Sharpless 105) is an emission nebula in the constellation Cygnus, about 5000 light-years away from Earth. It was discovered by William Herschel in 1792. It is formed by the fast stellar wind from the Wolf-Rayet star WR 136 (HD 192163) colliding with and energizing the slower moving wind ejected by the star when it became a red giant around 250,000 to 400,000 years ago. The result of the collision is a shell and two shock waves, one moving outward and one moving inward. The inward moving shock wave heats the stellar wind to X-ray-emitting temperatures.
source: Wikipedia
NGC/IC:
Other Names:
Object:
Constellation:
R.A.:
Dec:
Transit date:
Transit Alt:
NGC6888
Crescent Nebula, Sh2-105, Caldwell 27
Emission Nebula
Cygnus
20h 12m 00s
-38° 21.0′
26 August
89º N
Conditions
During the month of August, Sh2-105 is high up in the northern skies with altitudes well above 80* and visibility throughout the astronomical night. For the current image, Sh2-105 was photographed over 11 nights during August/September 2024 from the remote observatory at IC Astronomy in Oria, Spain.
Equipment
The default rig at the observatory was used. The core of this rig is a Planewave CDK-14 telescope on a 10Micron GM2000 mount, coupled to a Moravian C3-61000 Pro full-frame camera. The RoboTarget module in Voyager Advanced automated the process to find optimal time-slots during astronomical night.
Telescope
Mount
Camera
Filters
Guiding
Accessoires
Software
Planewave CDK14, Optec Gemini Rotating focuser
10Micron GM2000HPS, custom pier
Moravian C3-61000 Pro, cooled to -10 ºC
Chroma 2” H-alpha and OIII (both 3nm) as well as Red, Green and Blue unmounted, Moravian filterwheel L, 7-position
Unguided
Compulab Tensor I-22, Windows 11, Dragonfly, Pegasus Ultimate Powerbox v2
Voyager Advanced, Viking, Mountwizzard4, Astroplanner, PixInsight 1.8.9-3
Imaging
The Crescent Nebula is one of those staple objects that finds its way into many astrophotographers’ portfolio. Back in 2018 when my astrophotography journey had just kicked off, it was one of the earlier targets that were shot. Now, in mid August 2024, when a full moon made it almost impossible to photograph some nice galaxies, it seemed like a great opportunity to re-visit this object and see how much different it was imaging under better conditions, with better equipment and six years of experience under the belt. The Ha and OIII narrowband Images were shot using 10 min exposures each. For the stars, quick 3 minute RGB images were shot. Unfortunately the imaging conditions were not always great, and somehow especially some OIII images had poor quality due to issues with seeing, clouds, etc. During a first run of WBPP, almost 40 OIII frames were rejected. So on 09 September, when conditions were great, I shot an extra 30 images, just to compensate for these losses. In the end, about 12h of each Ha and OIII were obtained. This makes the total exposure just over 25h, just about where I’d aimed for.
Resolution (original)
Focal length
Pixel size
Resolution
Field of View (original)
Image center
9036 × 5958 px (53.8MP)
2585 mm @ f/7.3
3.8 µm
0.30 arcsec/px
45' x 29'
RA: 20h 12m 08.077s
Dec: +38° 21’ 10.80”
Processing
All images were calibrated using Darks (50), Flats (50) and Flat-Darks(50), registered and integrated using the WeightedBatchPreProcessing (WBPP) script in PixInsight. This time I applied a manual Pedestal value of 200 ADU for all frames. Adam Block explained this in one of his videos, that if you image from a very dark site, the pixel-to-pixel variation in your darks can be large enough to create zero-values in your calibrated frames and he showed a nice trick to demonstrate them. And indeed, my frames were full of zero-values…. With some testing it turned out that a Pedestal value of 200 ADU should be more than enough to never have a zero-value again.
As mentioned above, a lot of OIII frames were rejected as they did not reach the minimum set weighting threshold of 10%. But with the extra frames captured, about 12h of exposure per narrowband channel was obtained.
As a first step I attempted a gradient correction in all channels. The problem however with this image is that it is in a very nebulous region, and it is hard to see what is gradient and what is real signal. I had the impression that both GraXpert and GradientCorrection removed a lot of detail in the background, so I decided not to use it. Only for the RGB images, where background was not very important, GradientCorrection was applied.
For the narrowband image, first a linear fit was performed on the OIII channel to make the overall signal strength in line with the Ha image. This is an old practise that I stopped using, and replaced with SPCC and NarrowBandNormalization to balance the various colour channels in a narrowband image. But a recent training video of Adam Block (Narrowband Fastback) demonstrated its value again. When the HOO image was assembled using ChannelCombination, the resulting image already looked rather nice. Next up was deconvolution using BXT, followed by SXT. But unfortunately SXT was not able to extract only stars, and also took out a serious piece of the OIII signal in the image. Again a recent technique from Adam Block showed how to prevent that from happening. After making a pure black image and a pure white image, you can use the CloneStamp tool to brush away the artefact in the extracted star image by using the black image as the source. But rather than execute the CloneStamp tool as you normally would, you now drag its triangle over the white image. This paints black wherever the CloneStamp tool had been applied. The white image can now be used as a mask on the HOO image before SXT is applied. As with most black/white masks, a little convolution to blur the edges is usually a good habit to make the result blend in well. As can be seen in the images below, this trick was very effective in maintaining some significant detail in the OIII signal that would otherwise have gone lost. One other lesson from this experience is that it is always good to create the stars-only image and inspect it. Even if it wont be used, it might indicate some artefacts from SXT that are worth dealing with.
SXT was not 100% successful in extracting only the stars. It took bit of OIII signal out (left two images). Using te CloneStamp tool and a pure black and pure white copy of the image, a mask can be created at exactly the spot of the artefact. SXT can then be applied with the mask in place. In this case, the little trick save some valuable detail in the OIII part ofte nebula (right image).
Although the noise was not dramatic, I decided to run a bit of noise reduction while the image was still in its linear stage. Just to prevent it to become somewhat dominant during stretching. In Adam Block’s Narrowband FastTrack course, there are instances where he applies NarrowbandNormalization in the linear stage, with some pretty good results. So far I have always been under the impression that NarrowbandNormalization should be applied in the non-linear stage, but inspired by Adam’s videos, I tried it here. Unfortunately the resulting image was quite underwhelming, so I decided to revisit later. For stretching a two-step approach was used. First with StatisticalStretch from SetiAstro, a rough stretch was applied. Then, using GHS, some areas of low contrast (horizontal lines in the log-based histogram) were stretched carefully to add a bit of contrast. At this point the image already looked quite nice, but when NarrowbandNormalization was applied, it changed a lot of the colours, and it just was not getting much better. So instead I decided to fine-tune the last bits using ‘manual’ controls. The background had quite a serious teal colour to it, originating from the OIII channel. Some of it could be reduced by just adding some contrast, but that alone was not enough. I decided to tackle the background separately. A inverted mask was created from a contrast-enriched luminance channel. With the mask applied, the black-point for both the blue and green channels in HistogramTransformation were lifted just enough to make the shadow-colours a lot more neutral. To finish off, some contrast and colour enhancements were applied using CurvesTransformation.
The Red, Green and Blue channels were combined using ChannelCombination, followed by SPCC colour calibration and BXT deconvolution. The stars were removed and stretched using StarStretch from SetiAstro (stretch amount 5.5, color boost 0.9). The stars were put into the HOO image using ScreenStars. There are many ways to put stars back into a starless image, and I have used several of them. My current understanding is that ScreenStars from Mike Cranfield and Bill Blanshan is the superior one, as it keeps the colours of the stars intact when blended into a very colourful image. As final steps a second and modest noise reduction was applied, followed by a small crop to focus a bit more on the object itself.
The same target was shot very early on in my astrophotography journey, back in 2018. One of the objectives was to see how better conditions, better equipment, and 6 years of experience would affect such an iconic image in the portfolio. And the results do not disappoint. The 2024 image shows a lot more detail, has much better detail in the OIII envelope around the Ha nebula and shows much more detail in the Ha clouds in the background. While the image from 2018 shows a lot of contrast, it compares as rather harsh compared to the much softer tones of the 2024 image.
One of the goals was to compare an image taken in 2024 with one from 2018. A lot has changed since then. The conditions from the remote observatory in Spain are obviously much better than from my backyard in The Netherlands. And the 350mm mirror of the CDK collects a lot more light and detail than the 130mm objective of the Takahashi TOA-130. One of the things I notice while processing these newer images is how much more malleable they are and how much easier it is to put them together without combating all kind of shortcomings during the imaging itself. And it must be said, recent processing tools have made life a whole lot easier. Tools such as the AI-based RC-astro suite, the SetiAstro scripts, the Bill Blanshan/Mike Cranfield scripts and numerous other PixInsight improvements have greatly enhanced the power of PixInsight. And finally, after watching hundreds of training- and YouTube videos on processing while having processed now almost 100 images myself, I do feel more confident on getting the desired results out of a dataset.
All that said, it is very rewarding to see the progress made over time. Perhaps I will revisit some other targets as well at some point, but there are still lots of new objects to shoot from the Messier, Arp, Sharpless, Caldwell etc. libraries, so that will remain the priority.
Processing workflow (click to enlarge)
This image has been published on Astrobin and received Top Pick Nomination status.
