Star colour?

[jimjam11]

Attached is my second attempt at guided imaging, of NGC2403. It is a stack of 7x300s lights with flats, darks and bias.

No matter what I try I cannot get richly coloured stars. They always come out white or close to it. I have tried manually boosting star colour and also Noels astro actions, but I still dont get the amazing colours I see from others. In particular I seem to struggle with red stars.

I generally process by a few levels stretches to lighten and balance the histogram, followed by 3-4 histogram stretches, generally with an S shaped curve. 

The stars look nice and round to me, so I think the guiding worked well?

Is this because I am imaging through an LP filter?

Am I doing something wrong at the stack stage which is preventing me from getting decent star colours, or am I not collecting enough data (I thought I should be able to get coloured brighter stars with minimal data)?

Or am I being too crude in my processing?

TIA,

StarColourTest2.tif

[Starlight 1]

Are you useing DSS  if so set the saturation up to 20 or 25% as Dss allways put at 0.

[Starlight 1]

Ok this is your photo , I put it in Dss and move the LRGB down a bit and saturation up to 25%

[Thalestris24]

Hi

Looks like the focuser tube might be intruding from the left...

Louise

[Starlight 1]

Hi

Looks like the focuser tube might be intruding from the left...

Louise

You are right Louise , did not see that I was pointing out it do have star colour.

[Thalestris24]

Someone else had same problem recently with a 130pds and reducing coma corrector so it was kindof on my mind!

[jimjam11]

Someone else had same problem recently with a 130pds and reducing coma corrector so it was kindof on my mind!

Yes, I commented on that thread. I thought trying to sort out my star colour would be achievable without requiring a saw!

[Oddsocks]

The white stars are to some extent normal, the star core is so bright it quickly saturates the three RGB pixels to 100% and when the pixel data is reconstructed 100% in the three RGB channels = White.

There are so many different ways to restore the colour afterwards that everyone has different ideas, depending on the tools at their disposal, and the quality of the image in the first place.

As you are using PS and Noels Actions then the easiest way is to restore the stellar colour early on in the processing, make just a gentle stretch of the raw data to keep the background light and balance the three colour channels as perfectly as possible to align the the left side of the individual channel histograms with each other and try to arrange the widths of the histograms to be the same width, in the histogram view look at the superimposed view, where the three channels are shown overlapping each other, to judge how the channels are lining up, remember to press the recycle/refresh view icon, top right in the histogram window, each time you make a change to update the histogram display.

Once the channels are perfectly aligned then go in to Noels actions to run the increase star colour tool.

As Noels action runs you will be presented with a couple of interaction windows, the important one is the Gaussian Blur window, use the preview screen to zoom the display to maximum and find a big bright star, adjust the Gaussian Blur setting while looking at the star core and increase the setting for blur until you just start to see a dark donut appear, then back off the blur setting by one pixel.

Then carry on with the action to completion.

The first run will begin to bring a little colour into the core, now run the tool again, and again, and again, reducing the size of the Gaussian Blur setting each time.

When happy with the outcome you should zoom right into the image at pixel level and look at the brightest saturated stars, there are likely to be colour artefacts present in the stellar cores, to deal with these you can use the PS colour replacement tool. Sample an individual pixels colour in the stellar airy disk at the unsaturated boundary between the star and the sky background and then brush this into the star core overwriting any colour artefact present.

Then you can return to levels and curves etc to finalise the image.

Sounds like a lot of hard work, and it is, but when you have less than perfect data because of the use of light pollution filters in poor skies you don't have a lot of choice.

If you want to improve on this then you can take a soft stretched, colour boosted layer and then blend it with a harder stretched copy with no colour boost in layers etc....loads of possible variations on this theme.

These days I pretty much do everything in PixInsight with selective masks but prior to this I used PS and Noels Actions so I processed your data in PS and Noels Actions only.

For your image I lowered the stretch, ran Noels Actions, Increase Star Colour, five times using Gaussian Blur successively with a setting of 5.0, 4.0, 3.0, 2.0 and 1.0, then touched up the stellar cores of the brightest stars with the colour replacement tool, finally re-stretched and resized for web, took about fifteen minutes so rather rough-and-ready and it would have been more accurate with the un-stretched raw data as a starting point, but it gives you an idea of what is lurking in your image waiting to be revealed.

 

[ollypenrice]

Good stuff, Oddsocks.

Olly

[jimjam11]

The white stars are to some extent normal, the star core is so bright it quickly saturates the three RGB pixels to 100% and when the pixel data is reconstructed 100% in the three RGB channels = White.

There are so many different ways to restore the colour afterwards that everyone has different ideas, depending on the tools at their disposal, and the quality of the image in the first place.

As you are using PS and Noels Actions then the easiest way is to restore the stellar colour early on in the processing, make just a gentle stretch of the raw data to keep the background light and balance the three colour channels as perfectly as possible to align the the left side of the individual channel histograms with each other and try to arrange the widths of the histograms to be the same width, in the histogram view look at the superimposed view, where the three channels are shown overlapping each other, to judge how the channels are lining up, remember to press the recycle/refresh view icon, top right in the histogram window, each time you make a change to update the histogram display.

Once the channels are perfectly aligned then go in to Noels actions to run the increase star colour tool.

As Noels action runs you will be presented with a couple of interaction windows, the important one is the Gaussian Blur window, use the preview screen to zoom the display to maximum and find a big bright star, adjust the Gaussian Blur setting while looking at the star core and increase the setting for blur until you just start to see a dark donut appear, then back off the blur setting by one pixel.

Then carry on with the action to completion.

The first run will begin to bring a little colour into the core, now run the tool again, and again, and again, reducing the size of the Gaussian Blur setting each time.

When happy with the outcome you should zoom right into the image at pixel level and look at the brightest saturated stars, there are likely to be colour artefacts present in the stellar cores, to deal with these you can use the PS colour replacement tool. Sample an individual pixels colour in the stellar airy disk at the unsaturated boundary between the star and the sky background and then brush this into the star core overwriting any colour artefact present.

Then you can return to levels and curves etc to finalise the image.

Sounds like a lot of hard work, and it is, but when you have less than perfect data because of the use of light pollution filters in poor skies you don't have a lot of choice.

If you want to improve on this then you can take a soft stretched, colour boosted layer and then blend it with a harder stretched copy with no colour boost in layers etc....loads of possible variations on this theme.

These days I pretty much do everything in PixInsight with selective masks but prior to this I used PS and Noels Actions so I processed your data in PS and Noels Actions only.

For your image I lowered the stretch, ran Noels Actions, Increase Star Colour, five times using Gaussian Blur successively with a setting of 5.0, 4.0, 3.0, 2.0 and 1.0, then touched up the stellar cores of the brightest stars with the colour replacement tool, finally re-stretched and resized for web, took about fifteen minutes so rather rough-and-ready and it would have been more accurate with the un-stretched raw data as a starting point, but it gives you an idea of what is lurking in your image waiting to be revealed.

Sat1.jpg

Thank you for taking the time to produce such a detailed response, much appreciated. My version of Noels actions didnt prompt at any stage, but I changed it to do so. I assume you specify the blur radius for each of the blur stages (I count 5 within the action)?

[Oddsocks]

In my copy of PS (CS6) and Noels actions I only see five input screens for the Increase Star Colour tool, and the Gaussian Blur -  Pixel Radius interaction window only appears once so I guess I must be using a different version of Noels Actions to you, mine is V 1.6.

I ran your original TIF image again and have posted screen shots for each step as it appears in the action sequence.

In total I ran the Noels action for Increase Star Colour fifteen times before I was happy and then used the colour replace tool on the cores of just the brightest stars to wipe away the colour artefacts.

This time I began with a weaker initial stretch to tone down the blue stars before making a final stretch again to reveal the galaxy core.

Spent longer this time, 20 minutes instead of fifteen, and thought it came out a bit better but with your original raw data as a starting point and a bit more time you should do much better than me.

I have attached the resulting jpg below.

Noels Actions, Increase Star Colour, Interaction screen sequence:

1: Save Selection

 

2: Contract Selection:

 

3: Gaussian Blur:

 

4: Fade:

 

5: Hue and Saturation:

 

 

[jimjam11]

How strange, I am running the same versions, purchased last week. I manually tweaked the actions to include prompts and will see how I get on.

[Robp]

Thanks for that Tutorial Oddsocks.

I tested the method out on a couple of images I recently processed and it brought out the star colour nicely