Being a resident of South Africa, home to the bushman tale of the Milky Way, the story about a girl from an ancient race resonates deeply.  The legend says she created the Milky Way by throwing the cinders from a fire into the sky to light hunters’ paths home.

While parts of the Milky Way can be visible throughout the year in South Africa, if you want to photograph the centre of our galaxy, you must plan to shoot anywhere between March and September. This is when the Galactic Centre can be seen overhead at night. Unlike in the Northern Hemisphere, South Africa is fortunate to capture the Milky Way under excellent star gazing conditions when it is at its most spectacular.

I was inspired by the fabulous images of the Milky Way centre here, so I decided to take the plunge and see if I could capture the vast beauty of the southern milky way and galactic centre.

I planned a family holiday over the new moon to the dark Mountain Zebra national park to combine a wildlife trip with evenings to take my Milky way Mosaic.

The image was planned with a combination of telescopius  and NINA, with I used for my image accusation. After a lot of planning and framing, I settled on a 90° frame of 40 panels with a 20% overlap. 



I acquired my images with a Samyang 135mm F2.0 ED UMC lens at F4, built with custom 3d printed rings and a myFocuserPro2 stepper motor-based focuser.  The evenings were a crisp -5 C most evenings under very dark skies.

Using NINA’s advanced sequencer, I could fully automate the acquisition of the frames from my ASI 2600MM Pro. Each panel consists of an LRGB set of 5x60s for L and 3x30s each for RGB.

The frames were all stacked and calibrated in Pixinsight using the WBPP script. I used the fabulous https://www.astropy.org/ tools to write a small python script to update the “OBJECT” property of the FITS headers of the frames. By grouping all the individual panels under the same “OBJECT”,  WBPP could stack and calibrate each of the panels individually.

I used the PhotometricMosaic Script by John Murphy to build the Mosaic. The scrip produces excellent results in stitching and calibrating the individual frames. The LRGB panels were all combined individually to create four monochrome frames. Finally, a colour image was produced from the RGB frames, combined with the noise-reduced and processed L image.
Processing such a large image, over ½ gigapixel, is quite challenging, and some imaging processes like noise reduction would not be complete on such a large image. To overcome this, I wrote a small python script to split the source image into four panels that could be noise reduced individually and then joined back to form a single image again.  Lastly, I used the excellent Generalised Hyperbolic Stretch Script by David Payne and Mike Cranfield to stretch the image. The stretch produced a nicely stretched image without blowing out any bright areas.

The resulting image is quite hard to share in its full resolution. I think only a large print will do justice to the ½ Gigapixel. The one shared here is ¼ of the full resolution, and the revision D gives you a sense of what the complete resolution image components look like.

This is by far the most ambitious project I have tackled. After three nights of accusation and weeks of editing, I am very proud of how the image turned out, especially when you view the details of the image at full resolution.  

Please comment or DM me if you want any insight on the scripts of other parts of the workflow.