IC5070 - Star Processing Attempt.

[silentrunning]

I've been developing my own star removal/reduction process and thought I would post the results.

I'm working on my recent limited data set of IC5070, part of the pelican nebula. This is first image I figured I would try and do something with the stars, I've never bothered in the past.

I've looked at various things out there like Stratton and Starnet but ultimately I wanted something that is guaranteed to identify stars at all mags and not all sorts of other things. The only way to do this is by using star catalogue information and Pixinsight has an incredible script called MaskGen that will take a plate solved image and generate a star mask based on things like magnitude range using catalogues such as GAIA which go extremely faint. On this image I generated three star masks for the magnitude ranges bright-15, 15.1-17 and 17.1-19 so I I included probably all of the stars in my image. I then loaded these and the original image into Affinity.

By generating star selection layers based on the star masks and applying Affinity's Inpainting tool to the underlying original image I was able to remove all the stars without the sort of artefacts that Dust and Scratches creates and reintroduce the three levels stars by turning on and off the layers.

To each star layer I can apply any processing I choose such as minimal blurring to reduce the size of the stars. I only needed this on the brightest layer for the result I was after, although I'm far from finished.

I have not applied any other processing but with the stars removed I should be able to achieve for example some clean nice Hubble pallet effects and edge sharpening without effecting the stars.

The images show the result for combinations of layer visibility.

1. This is the original image

2. All star layers turned off...

3. The faintest mag 17-19 stars reintroduced, they are there!

4. With the 15-17 mag stars added in...

5. With the brightest stars added in with a minimal blur applied to keep them nice and small...

 

Here's the full image with the stars removed...

 

Edited June 3, 2020 by silentrunning

[Annehouw]

Thanks for the nice write-up!

Looks like inpainting works very good. I have a copy of Affinity Photo. Time to give this a try.

 

[Datalord]

7 hours ago, silentrunning said:

Pixinsight has an incredible script called MaskGen

Where is this script? I can't find it in my Scripts menu?

[silentrunning]

The script can be found here with some excellent others...

http://www.skypixels.at/pixinsight_scripts.html

[MartinB]

Brilliant!  Very interesting John thank you.  I've taken notes!  Marvellous image as well.

[Datalord]

On 04/06/2020 at 08:22, silentrunning said:

The script can be found here with some excellent others...

http://www.skypixels.at/pixinsight_scripts.html

Those scripts are brilliant! Except... The generated mask is off by 3-4 pixels, which means it is in the wrong place for all the stars. So close... 😕

[silentrunning]

52 minutes ago, Datalord said:

Those scripts are brilliant! Except... The generated mask is off by 3-4 pixels, which means it is in the wrong place for all the stars. So close... 😕

Ummm interesting, it's bang on for me. Were they all shifted or randomly? Did you put the date of the acquisition in as it takes into account proper motions?

[Datalord]

41 minutes ago, silentrunning said:

Did you put the date of the acquisition in as it takes into account proper motions?

Hmm, sort of. It's multiple nights stacked, but it is at maximum one or two days off. I guess I have to play around a bit.

[silentrunning]

56 minutes ago, Datalord said:

Hmm, sort of. It's multiple nights stacked, but it is at maximum one or two days off. I guess I have to play around a bit.

A year or two ain't gonna make much difference. Something else must be effecting the result. Were they all shifted by the same amount in the same direction?