NGC6946 - Fireworks Galaxy
NGC6946 - Fireworks Galaxy. Click here for full resolution.
The Fireworks Galaxy with classifications NGC6946 and Caldwell 12, is a face-on intermediate spiral galaxy, right on the border between constellations Cepheus and Cygnus. The distance to Earth is approximately 25 million light-years. Originally it was considered to be part of the Local Group of galaxies, but that appeared to be wrong.
The name Fireworks Galaxy originates from as recent as 2005 and relates to the unusual high number of supernovae that have been observed in this galaxy. No less than 10 supernovae have been observed in the last century, making it significantly more active than the second-most active galaxies M61 and M31, with six each. Also new star formation takes place at a tremendous high rate. It is not clear why this galaxy is so active. A possible explanation could be that the galaxy has recently picked up a pile of low-mass neutral hydrogen clouds from the surroundings.
Planning
Object
Visual Magnitude: 8.9
Apparent size: 11.4 x 10.8 arcmin
R.A.: 20h 35m 18.01s
Dec.: 60º 13’ 26.7”
Conditions
Astr. night: 23:18
Astr. dawn: 3:44
Moon: 0%
Moon set: 21:02
Humidity: 56-62%
Pressure: 1023-1024 hPa
The Fireworks galaxy is visible throughout most of the winter and reaches fairly high altitudes. Its magnitude of 8.9 does not make it a particularly bright galaxy, but more than bright enough to be photographed with a good amount of detail. With a size of approximately 10 arcmin, it is a medium-sized galaxy in the sky, a good object for the TOA-130 + ASI1600.
Images were captured on four consecutive nights in April 2020. From April 20-24 the observatory could be left outside thanks to the prolonged period of clear skies and generally good conditions with low humidity. Wind was quite strong on the first day, but settled down a bit on subsequent days.
Capturing
The image is captured with 180s exposures at gain 139 (unity gain) using LRGB broadband filters. In addition, a 5nm H-alpha was used to bring out the areas in the galaxy of high star formation activity. The H-alpha frames were also 180s exposures, but with an elevated gain of 300. H-alpha frames typically have a longer exposure and/or a higher gain because they have a restricted amount of light coming in.
Framing was fairly straightforward by just positioning the galaxy right in the middle of the frame. The 1º field of view of the TOA-130 with ASI1600 camera is significantly larger than the 10 arcmin width of the galaxy, leaving the galaxy still fairly small in size.
Since there were enough clear skies for three consecutive nights of imaging, there was an opportunity to collect some real long exposures on this target. In total 15.7h of data was collected. A general ‘rule’ has developed over time to collect about as much luminance data as RGB data, meaning the exposure for each colour channel is about one-third of that of the luminance. In this case it meant that 7h of luminance data was collected and a little over 2.5h of data for each of the colour channels. The H-alpha data was meant to just add a little key lights into the image, for which 1h of total data was considered sufficient.
Technical details
Telescope
Mount
Camera
Sensor Temp.
Exposures
Takahashi TOA-130 + FL67 flattener
10Micron GM1000HPS
ZWO ASI1600MM Pro
-25ºC
Luminance
Red
Green
Blue
H-alpha (5nm)
Total Exposure
140 x 180s @ Gain 139/21
51 x 180s @ Gain 139/21
51 x 180s @ Gain 139/21
51 x 180s @ Gain 139/21
20 x 180s @ Gain 300/50
15.7h
Processing
All frames were calibrated with Bias (100), Dark (50) and Flat (25) frames, registered and stacked using the BatchPreprocessing script. Unfortunately, in the bottom-right corner, there was some kind of a reflection visible in the frame. The origin remains unknown. By speculation one could think of some sort of reflection against the flip-flat from one of the lights of neighbouring houses. If the whole frame would have been essential for the image, there would probably have been ways to correct for the reflection. In this case, given the relatively small size of the galaxy, a simple crop dealt with the problem quickly.
A slight gradient in the background from right (bright) to left (darker) was removed. Usually the DBE tool is used, especially if the gradients are a bit more complex. But here it was such a subtle and simple gradient that ABE was more than sufficient.
The next step was noise reduction. Especially on linear images good results have been obtained from the MureDenoise script. What makes this script different than other noise reduction algorithms, is that it does not try to detect noise in the image, but it tries to model inherent noise of the camera, and correct for that. This can give much more pleasing looking images than algorithms that essentially smooth out small-scale detail in the image. In the past it was quite tricky to feed the MureDenoise script with the right parameters. But in the current version of PixInsight, there is a ‘helper’ script called MureDenoiseDetectorSettings. That script calculates the relevant characteristics of the camera from a set of two flat frames and two bias or dark frames. These parameters can then be easily entered into the MureDenoise script. The variance scale is then the remaining parameter to tweak. One way to do that is to make multiple previews and apply MureDenoise to each of these previews with a different variance scale for each. In that way the variance scale with the most pleasant results can be selected. After MureDenoise, there is often some high-frequency noise left. This was removed using MultiscaleLinearTransform, focusing only on layer 1 (3,0.33,3) and layer 2 (1,0.33,3). In the image below, close-ups show the increasing level of noise reduction after subsequently applying MureDenoise and MLT.
A very different approach to noise reduction is software called Topaz Denoise. This software uses artificial intelligence to figure out what is detail in an image and what is noise, and then smooth out that noise. Unfortunately the results obtained with Topaz Denoise were not very impressive. Star shapes turned into strange diamond-sort of objects, especially in areas of structure, such as the spiral arms of the galaxy. Over time, perhaps when Topaz Denoise learns what an astrophoto is, this technique will become better, but noise reduction methods like MureDenoise, MLT or MMT so far have given better results.
Topaz denoise gives smooth backgrounds, but also strange morphological changes to the stars
After the noise reduction, the RGB channels were combined and photometric colour calibration (PCC) was applied. Especially for galaxies, PCC gives often very natural colours. The luminance data was subjected to a careful deconvolution, to concentrate the brightness of the stars and pull out a little more detail in the galaxy. All channels were then stretched using HistogramTransformation. During stretching special care was given to not clip the stars too much. If stars get too bright they tend to loose their colour in the final image. In order to bring the balance between galaxy and stars a little bit more towards the galaxy, a further star shrinking was applied using MorphologicalTransformation (Amount 1.0, size 3). Then Luminance and RGB were combined using the LRGBCombination process. A bit more structure was then extracted using LocalHistogramEqualization (Radius 64, amount 0.18).
The H-alpha data was given a similar treatment to the luminance data and was added to the LRGB image at the final stages, using the NBRGBCombination Script and H-alpha mapped to the Red channel. This is a subtle process. If done too much, or if the brightness levels between H-alpha and LRGB are too far apart, the whole image can become quite red. Reducing the scale of the H-alpha channel can be necessary to balance the final colours well.
As a final step the overall saturation was increased by 0.16, to give the image a little bit more punch.
This image has been published on Astrobin.
