A little guidance please - correcting stars in image - GIMP

[Adam Y]

HI all - first off I extracted this star image from a picture of the lagoon nebula I am playing with in GIMP.  I used starnet ++ to get the starless version, then subtracted that from the full picture using layers in GIMP.  I am very happy with the way the nebula itself has turned out, and my thought was I'd fix up the stars - reduce the overall number, make some a bit brighter, remove weird colours and haloes etc, but I can't seem to get a handle on it - information overload.

 

I took a snapshot just as a jpg so you can see what I am talking about.  Zooming in should show what I mean in various areas.  Some stars are a weird blue with a halo, others have what looks like some sort of aberration, others are just a dull ruddy tone.  I'd like to reduce the number of the tiny dull reddish ones, and fix the blue ones to be a bit more white wtihout haloes, and correct the aberrations around some of the others.  I hope that's the right word, there are different colours around the edges.  Don't worry, i know better than to try to perfect a .jpg  :-)

 

I have my starless image as the base layer, this star field as the top layer, and my thought was I'd either use additive mode to combine the 2, or perhaps lower the opacity until i get a result I like.

 

I'd like a bit of guidance from you pro's.  I am sure there are other tools that are better, but I'd like to get my head round the basic concepts before I splurge.  Thanks!

[Adam Y]

... i could have sworn i added this as an attachment, let me know if i need to try again - perhaps i can't see my own attachments?

[TerryMcK]

Yes can see the image Adam

[Adam Y]

Thanks Terry - are you able to download it and zoom in to see what I am seeing?

[TerryMcK]

The jpg really isn't high enough resolution to see anything of note. Can you upload the full starless image zipped up? You should be able to upload 48MB in size.

[Adam Y]

Hmmm i'll have to put some thought into this - even zipped it is ~ 90 MB. 

[TerryMcK]

Have a look at https://www.transfernow.net/en 

You can upload the file and then post the link that is given.

 

I haven't used the service but it looks like you can do it for free

[Adam Y]

Excellent, didn't even have me create an account - I may do so at some point anyway, thanks!  here's the link, says its good until 31 Jan.  

 

https://www.transfernow.net/dl/20230124tFdYakYz

[TerryMcK]

I've downloaded it and can see the blue haloes around the stars. 

 

It's ages since I used starnet++ but I remember that there are some tweaks you can make to the settings:

In PixInsight you can change the stride from28 to 64, 32, 16 or 8 

 

I can't remember the syntax that you have to do it with the standalone version but there are examples in the installation directory of starnet and that may give you a clue. It may clean up the stars by playing around with stride.

 

It has produced a few strange artefacts too like this

 

Again playing around with the settings might help.

I must say it has produced a very rich star field and I can see what you mean about brown stars. What area of sky is it?

 

I don't actually use GIMP but had a play around with it in the past so not sure if there are any deconvolution tools in it. Essentially deconvolution sharpens images and is the opposite of convolution which blurs images. So have a play around with sharpening but don't do too much as it can look really odd if over done.

 

Here for instance is overdoing it

 

And the same area of your image without deconv

 

 

 

Then put that onto a blend layer the background being the nebula. Then play around with different degrees of blend to produce a pleasing image with a star reduction. There are other modes you can play with from Photoshop, GIMP being a sort of PS clone, I believe blend can help.

 

I have moved away from Photoshop for astro processing now and use PixInsight. It has a steep learning curve, is relatively expensive, but once you have bought it and learnt the fundamentals there is no more outlay (apart from buying some excellent plugins like StarXTerminator etc).

 

Maybe somebody who uses Photoshop for astro can chime in and make a few suggestions too.

 

[Adam Y]

So I did have some success in using wavelet decomposition to reduce the number of stars in the field, which seems to have helped.  Took out a good majority of those useless little brown stars.  There is a beautiful tutorial here which helped walk me through the steps and semi-understand what was going on.  (I have no background in this stuff, so its all Greek)

 

That is a good idea Terry, I did not think to look over the stride settings, instead just taking the default value.  I'll have a play with that and see if that is an improvement.  I don't think I actually did much sharpening in Siril or GIMP so it must be the output of  Starnet ++.  

 

I must say that just reducing the number of visible stars made a HUGE improvement to the overall picture, and I can probably live with it while I look for a way to reduce halos and deal with the fringes around some of the stars.  Here's a screenshot so far - it is the Lagoon Nebula, and just one I downloaded as raw files to use as a tutorial (I didn't take the picture):

 

[TerryMcK]

You are getting the hang of it from your latest image. I think downloading others images to practice upon is a great way of learning in the UKs cloudy environment this time of year. Especially if you can get the raw images that have already been aligned, stacked and calibrated.

[Adam Y]

Thanks Terry - even better, I had to download all the flats, biases, darks and calibrate and align them too!  My next challenge is going to be to try to capture this comet c2022/e3 ztf - using my wife's Nikon DSLR.   

 

Love this forum!

[Adam Y]

I think I know what may have happened on this picture - when I processed in Siril, I used deconvoluting which had an overall nice effect however I was not zoomed in enough to notice that the default settings put some wicked rings around the stars.  By backing off on the settings before applying the deconvolution (and monitoring in zoom), I got the effect I wanted without the rings and halos.  Current working theory anyway based on latest trials.