M31 - Andromeda Galaxy 6 Part Mosaic - Updated

[Taman]

The original version of this was posted back in January. It only has 8 x 4 minutes exposure time across the whole image, so getting a half decent result from it was a bonus, so I thought.

Here is what I did to get a much better version. This may not be the correct way, but it worked for me. Hopefully it will encourage others to experiment with their own images. 

I discovered by accident that Siril can do photometric colour calibration. This seems to be a secret process, as a Google search revealed that lots of people use it (in PixInsight), but they don't seem to talk about it much! I also read an article on creating a synthetic luminance layer from an RGB image and using it to enhance an image. This looked very promising for my project.

I started off by stacking the 6 parts in DeepSkyStacker. I also tried this in Siril, but DSS gave better results. Photometric colour calibration was then performed in Siril. This plate-solves and adjusts the colour of the entire image based on the real colour of the stars, looked up from a database. Fantastic, if it works!

After a slight curves stretch, the 6 calibrated images were stitched together using Microsoft Image Composite Editor (ICE) to create the mosaic. ICE also does a good job of adjusting slight colour differences between the images.

Next I used Starnet++ to remove the stars. The original and star-less images were then opened in Photoshop and one subtracted from the other (Image - Apply Image - Subtract), to create a separate stars only image. A grey-scale version of the star-less image was also used to make a synthetic luminance image. (A "real" luminance image is taken with a mono camera using no colour filters).

Masking, noise reduction and careful curves and levels adjustments were used on the "fake" luminance, to extract as much contrast, detail and sharpness as possible. The high pass filter was used to sharpen the dust lanes.

The RGB version of the star-less image was then processed with levels and curves to bring out colour and contrast, also being careful not to overdo the bright central core. I found that the calibrated colour works well on the stars, but results in a very brown Andromeda Galaxy. This may be correct, but is not nice to look at! A tiny adjustment to the red channel in curves fixed that and also had the effect of increasing the blue which gave a pleasing result. The colour image was then blurred slightly using the dust & scratches filter at 12px.

The luminance image was added on top of the colour image as a layer, with the blending option set to luminosity. I then played with the saturation of the colour layer and the opacity of the luminance. The luminance has the effect of muting the colour, so the saturation will need to be increased more than usual.

The last problem was the stars. I increased the saturation to bring out the colour and decided to reduce the impact of the stars in the final image. This may not be to everyone's taste, but it really made a difference. This was done by selecting and feathering the stars, then using the minimum filter at 1px. The effect was too much, so I reduced the opacity of the filter to about 75%. 

The stars were then added back to the colour-luminance composite, by using the Image - Apply Image - Add function in Photoshop and setting the opacity to 80%. The stars are still there, but are not so prominent. The image also still plate-solves!

Details: 6 part mosaic. Each part 8 x 240s, with darks, flats and bias. Taken with the C11 at f6.3 and Eos 550D.

Here are the two images for comparison. Gentle critiquing welcome.

[Annehouw]

Nice! You even got the dust clouds in M110!

[MarkAR]

Both really good, I think I like the second one better.

I think the only thing you might want to consider is trying to tame the brightness in the core a little.

[Taman]

34 minutes ago, Annehouw said:

Nice! You even got the dust clouds in M110!

Thank you!

Yes, I've also noticed an extremely faint tidal tail between M110 and Andromeda. I wasn't sure if it was real or something from the luminance layer, but it really is there!

[Taman]

39 minutes ago, MarkAR said:

Both really good, I think I like the second one better.

I think the only thing you might want to consider is trying to tame the brightness in the core a little.

Good suggestion. I'm viewing it on a large monitor and it's difficult to judge the overall brightness.

[vineyard]

Beautiful. How long did all that take you if you don't mind me asking? I prefer the second image - the stars are more restful on the eye, and also seem more "individual" (each one a little solar system), whereas in the first one they seem just like stars (not sure if that makes any sense!).

[Taman]

28 minutes ago, vineyard said:

Beautiful. How long did all that take you if you don't mind me asking? I prefer the second image - the stars are more restful on the eye, and also seem more "individual" (each one a little solar system), whereas in the first one they seem just like stars (not sure if that makes any sense!).

Thanks Vineyard! 

I'm glad you agree with me that the star reduction made it a lot better.

The whole thing start to finish was around 8 hours. Just putting it through Starnet++ to remove the stars took well over an hour because of the 350 MB file size!

I really hope to get some more exposure time on this, perhaps increasing the size of the mosaic. It's a work in progress! 😀

[callisto]

2nd looks good.....as said, with the M110 dust clouds also 👍