M31 - Andromeda Galaxy
M31 - Andromeda galaxy. Click here for full resolution image.
The Andromeda Galaxy, classified by Messier as M31, is the closest galaxy to our Milky Way, at a distance of approximately 2.5 million lightyears and located in the constellation Andromeda. It is easily the largest galaxy of our night sky, and has the very traditional shape of a barred spiral galaxy.
The Andromeda Galaxy and the Milky Way are the two largest members of what is called the Local Group, a group of approximately 80 galaxies, all within a radius of about 8 million lightyears. With approximately 1 trillion stars, the Andromeda Galaxy has about twice as many stars as the Milky Way, but in mass both galaxies are probably rather similar. The Andromeda Galaxy and Milky Way are moving towards each other and will eventually collide. But this will likely take another 4.5 billion years or so.
Located next to the Andromeda Galaxy is a small, but very dense elliptical galaxy, the satellite galaxy M32 with a supermassive blackhole in its center.
Planning
Object
Visual Magnitude: 3.28
Apparent size: 178 x 70 arcmin
R.A.: 0h 43m 51.36s
Dec: +41º 22’ 49.3”
Conditions
Astr. night: 21:34
Astr. dawn: 05:17
Moon: 16%
Moon set: 21:29
Humidity: 86-92%
Pressure: 1015-1020 hPa
The Andromeda Galaxy is perhaps one of the most photographed objects in the night sky. Its brightness with a visual magnitude of 3.28 together with its size of well over 2 degrees, make it visible to the naked eye. And with a small telescope, beautiful pictures can be made. The best time to photograph the Andromeda Galaxy is in autumn and early winter, where it reaches altitudes well over 80º.
The first set of images was captured on March 01, 2020. However, M31 was quite low on the horizon and the conditions were not optimal. A much larger set of images was captured on September 20 and 21, 2020, when M31 was very well visible, with no moon interference.
Because the images were captured with a mirrorless SLR camera not supported by the software, there was not a lot of automation involved. Two consecutive nights of imaging from around 22:00 until midnight was the result, after which the system was reset to the regular astrophotography setup to image another target automatically through the rest of the night.
Capturing
This image is actually a test project. For other photographic purposes, a Fujifilm GFX100 camera had been acquired. This is a medium format camera with an image sensor of 100 megapixel. How suitable would such a camera be as an astro-camera? With such large sensors it is important to have an image flattener with a large diameter, in this case 55mm. The FL67 flattener has an imaging circle of 90mm, so that should be more than enough to give a flat image on the GFX100.
The stand-out feature of this camera is of course its 100 megapixel resolution. Interestingly, because of the large 44x33mm sensor, the pixel-size is 3.8 µm, the same as for example an ASI1600. So one way to look at these photos is to consider each frame a mosaic of 6 ASI1600 images. Would it be possible to capture a large object like M31 in high detail without having to go through the trouble of making a mosaic?
In a separate blog post there will be more technical information on the camera, how to connect it to the telescope, the special adapter, etc. It will also detail the workflow, which is quite different. There is not yet native support for the GFX100 in the LibGPhoto2 library, used by the GPhoto driver in INDI. The older models in the line-up (GFX50s and GFX50r) are supported, so hopefully this is a matter of time. It means that focusing is not that easy and automated imaging is not (yet) possible.
With the object being quite large and stretched out, the positioning and rotation in the frame will have a big impact on the aesthetics of the picture. It was decided to frame the galaxy diagonally from bottom left to top right. A medium format sensor in combination with a 1000mm focal lengths provides only just enough field of view to capture the whole galaxy comfortably.
Some test shots were made with 1, 2 and 4 min exposures. 4 Minute exposures showed no highlight clipping, but did show decent detail in the shadow-areas. So this exposure was selected. During the session in March, 2m exposures were taken, showing slightly less detail.
Technical details
Telescope
Mount
Camera
Sensor Temp.
Takahashi TOA-130 + FL67 flattener
10Micron GM1000HPS
Fujifilm GFX100
Ambient (appr. 12ºC)
Exposures
OSC
OSC
Total Exposure
39 x 120s @ ISO800 (March 01, 2020)
41 x 240s @ ISO800 (September 20-21, 2020)
4.0h
Processing
All frames were calibrated with Bias (50), Dark (25) and Flat (25) frames. Unfortunately, there was a little hair on the sensor, that gave a shadow on the images that was difficult to deal with. Flat frames were taken the following morning, to not use up too much observation time. But that meant that the camera had been taken off the telescope and put back on again, and the little hair had moved a tiny bit. So instead of calibrating it out, flats were only making things worse.
Single frames showing dust particles in the September images. Note the increased detail of the 4 min exposures (September) relative to 2 min exposures (March).
In the above single frame examples (luminance extracts from the OSC images), the two dust shadows in the September frames are clearly visible. But another issue also shows, and that is the different rotation angle between the sessions in March and in September. Normally one would crop to fit, but the goal here was to have a 100 megapixel image, and the subject did not leave a lot of space to crop. It was decided to try to solve both issues in one go, using the StarNet process. This used to be a custom process, StarNet+, but in a recent upgrade of PixInsight, the process is part of the standard installation. What the process does is separate stars from the background. It creates two images, one with stars, one with background. Typical use of this process is to create starless nebula images. Here it was used to correct the artefacts by using the CloneStamp tool on the background image. This turned out to be very effective in those areas that did not have much detail in the background. Unfortunately the dust spot from September 21 was right in the middle of spiral structures of the galaxy and could not be corrected without affecting the galaxy structure. For this reason, in the final image, all the images from September 21 were discarded. Disappointing, but that is part of astrophotography.
On the background image, using the CloneStamp tool, the edges were filled in and the dust/hair shadow was erased. Overall this created a very smooth image that was then combined back again with the stars. This method really saved this image, even though some frames had to be dropped. Once a better method comes along, a second attempt might be made to process all frames (5.3h) into one image. As a lesson learned, going forward flat frames will be taken at the start of the imaging session, during dusk hours, if in any way possible.
Background images are created by the StarNet process corrected using the CloneStamp tool to fill in the edges of the frame and delete the dust shadow.
The rest of the processing was relatively standard. After calibration and registration, all images that could make it into the stack were integrated in one go. The background extraction, neutralisation and color calibration were applied followed by stretching to the non-linear state. Typically images at AstroWorldCreations come from monochrome cameras, so default workflows are usually to deal with color channels and luminance channels separately. And since the StarNet process had to be applied, this seemed like a proper approach here as well, even though obviously the images were originally all OSC images. There are probably much better/faster processing workflows on OSC images.
Noise reduction was applied to each individual color channel using the MultiscaleMedianTransform method as described here. There was still a bit of a halo present around the galaxy, which was further reduced by a Dynamic Background Extraction on each channel. This was followed by some further mild noise reduction on the high frequency noise with TGVDenoise and a slight contrast increase by increasing the blackpoint a little. Then the R, G and B channels were combined into a colour image. Colours were tweaked to taste. Some M31 images show very saturated blue and red colours. It was decided to keep the colours in this image a bit more neutral. To further reduce the chromatic noise, a slight blurring was applied using the Convolution tool. With a StDev of 4.8 and shape of 2, it smeared out the chromatic noise just enough.
The Luminance image was then added to the colour image using the LRGB Combination process. Structure in the galaxy was enhanced by subsequent application of the Local Histogram Equalization process (Kernel radius 344, Contrast Limit 2.0, Amount 0.3) and the DarkStructureEnhance script.
The final image was an image of 11634 x 8734 pixels, or 101.6 megapixels. It is difficult to appreciate the resolution while watching it on a computer screen. The image at the top of this post can be zoomed in to 100% magnification. Above are some samples of closeups of detail.
This image is published on Astrobin
