Imaging the Summer Triangle

The constellations Cygnus, Lyra and Aquilla with their bright stars, Vega, Deneb and Altair forming the ‘Summer Triangle’, lie overhead in August as seen from the UK.  The Milky Way lies along this region with a long dark dust cloud region called the ‘Cygnus Rift’  running through it and, over to the left of Altair, lies the tiny constellation of Delphinus.  Though taken under light polluted skies, this image does bear comparison with many images of the region seen on the web and I hope that you might find the processing used in its production interesting and useful.  The most significant was the use of Sequator in aligning and stacking the frames as it was able to remove the light pollution from the image without removing the structure of the Milky Way – something that I was not able to achieve using Deep Sky Stacker.

On a transparent night with no Moon, I attempted to image this region from a location within a mile from the centre of reasonably large town.  Now that LED have replaced Sodium streetlamps the light pollution is less (though cannot be filtered out as before) and I could just make the Milky Way arching overhead.  Happily, towards the zenith, the light pollution should be least.

I decided to use what is, perhaps, my best constellation imaging system which comprises a ‘legacy’ Zeiss Contax G 45mm, f/2, planar lens coupled to a Sony A7S full frame camera.  The lens is said to be one of the sharpest prime lenses ever made and shows virtually no coma when stopped down to f/4.  The camera has only a 12 megapixel sensor but has the wonderful ability to show stars in ‘live view’ on its tilting rear screen.  This makes it very easy to select the area to be imaged.  To avoid what has been called ‘The Sony Star Eater’ problem, the result of a noise reduction algorithm applied to the raw files, the exposures need to be kept to 30 seconds and below.  The camera was mounted on a ‘Nanotracker’ tracking mount as described in the digest article ‘Astrophotography Tracking Mounts’ and the exposures were controlled by an external intervalometer.

The area of sky imaged with this lens/camera combination cannot cover the whole of the Summer Triangle region as seen in the Bridgewater Skies Field of View Calculator (see article in digest) and it would have been easier to employ a shorter focal length lens but, instead, I chose to manually combined two ‘panes’ in Adobe Photoshop to cover the required field.

The individual frames had an exposure of 15 seconds using an ISO of 2,000.  The A7S has a ‘dual gain readout system’ and there is a reduction in the noise level of the image moving from ISO 1600 to ISO 2000, hence my choice of ISO.  However, the dynamic range will be reduced compared to lower ISOs.

The raw frames were aligned and stacked in Sequator (see article in digest) with three of the processing variants selected: High Dynamic Range, Reduce Dynamic Noises and Reduce Light Pollution.   The removal of light pollution where there is nebulosity in the image – as in the Milky Way – is really difficult and I was really impressed as to how well Sequator  had achieved this as the Milky Way showed up well in the resultant images.  In order that the two panes that resulted from the stacked frames could be composited, it was vital that the two sets of frames were processed identically with the resultant image shown below having applied a little stretching.

To make the dust clouds show up better, one can apply some local contrast enhancement using the ‘Unsharp Mask’ Filter with a large radius and relatively small amount.

However, when processing images that contain both stars and nebula, I usually ‘split’ the image into two; the background nebula and stars.  This is easily done.  The ‘Dust and Scratches’ filter was applied to the composite image with a radius of ~12 pixels and, magically, the stars disappear.  This image is saved as ‘Nebula’. 

The original image is brought back and copied and pasted over the nebula image.  The two resulting layers are then flattened using the ‘Difference’ blending mode leaving just the stars, this image being saved as ‘Stars’.  By allowing the Nebula image to be processed separately, I was then able to make one edit that would not have been possible working on the original image.  (One useful result is that the effects of hot pixels will only be present in the Stars image.)

Hot pixel removal

The standard way to remove hot pixels is to employ the ‘Long Exposure Noise Reduction’ feature used, by default, in many cameras when long exposures are made.  In this mode, each light frame is followed by an identical dark frame and the two are differenced to remove any hot pixels.  Fine, but there are two problems; firstly the dark frame subtraction actually adds some noise into the light frame and, secondly, the time collecting photons from the sky is halved.  I thus normally deselect this mode.  Hot pixels may thus be present in each light frame. 

Though looking at a .Tiff file made from one of the raw files (.ARW) produced by the Sony A7S, I could not actually see any hot pixels when viewed at full size on the screen, they did become apparent when viewed at 100%.  So, coloured streaks, most obviously blue in colour, appeared in the aligned and stacked Stars image.  I specifically aim not to perfectly align the star tracker on the  North Celestial Pole so that the stars move across the sensor during the taking of the exposures. (Only 19 in the case of the lower segment as clouds rolled in.)  This may help to remove what is called ‘colour mottling’ – variations in the colour of the sky background.  In fact, the alignment was quite accurate as the streaks as seen in the top section of the image below were only 30 pixels long – a drift of 1/10th of a pixel per second.  (So the stars would not have been significantly trailed in each 15 second exposure.)

For the first time I used a simple method to remove them.  The ‘Select Colour’ selection tool was used and, first, the blue colour selected with a low ‘fuzziness’  level.  The selection was expanded by 2 pixels so that the blue streaks were fully covered and the image painted over with the background sky colour selected from the image.  This removed the blue streaks as seen in the middle section of the image below.  A similar technique was used to remove some red and green trails and ‘splodges’ to give the final result seen in the lower section.

Enhancing the nebula image and adding the stars back into the image

The local contrast of the nebula image was increased and the areas away from the Milky Way selected and smoothed with an application of a Gaussian Blur with a radius of 40 pixels. (This latter edit could not have been done working on the initial image.)

The stars image was brought back, copied and pasted over the nebula image and the two layers flattened using the ‘Screen’ blending mode.  The brightness of the Stars layer was adjusted using the ‘Brightness Contrast’ tool to give a good blend giving the image called ‘Summer Triangle -1’

Though I could easily make out the constellation patterns observing by eye, these bright stars do not stand out in digital images in the way that they did when film was used.  Then, halation within the film’s emulsion layers scattered their light and made them appear larger – just as brighter stars are shown in star atlases.  To simulate this film like effect, the brightest stars making up the constellation patterns were selected: the stars image was processed using the curves function to leave the brightness of the brightest stars but reduce that of the fainter stars until only the brightest remained.  A Gaussian Blur was applied to them which makes them larger but reduces their brightness so their brightness was brought back using the ‘Brightness Contrast’ tool giving an image called ‘Bright Stars’.

The Summer Triangle image was brought back, the Bright Stars image copied and pasted over it and the two layers flattened using the ‘Screen’ Blending mode to give the result, ‘Summer Triangle -2’ shown at the top of the article.  An annotated version is shown below.

This image has an empty alt attribute; its file name is ST9-814x1024.jpg