Star colour processing

[Russe]

Time and again I find myself in the quandary over how much colour stars should have in processed images. While I agree that if all stars are white, then the processing has been suboptimal, I wonder how far to take things. What do you guys do?

I add a starlayer in PS and intensify the colour twice to 150 and throw in one or two saturation boosts. Then I apply a Gaussian blur (2-5) and change the blending mode to colour. Many stars then turn out blue or magenta. Oh, and I run a HLVG filter over it to get rid of green stars...

[Pete Presland]

Tricky one! With my DSLR they are nearly all white

Sent from my GT-I9100 using Tapatalk 2

[Russe]

Tricky one! With my DSLR they are nearly all white

Sent from my GT-I9100 using Tapatalk 2

Unless you have a de-bayered camera, you'll be surprised about how much colour you get if you do what I've described above...

[Christopher Davenport]

Nice process, Russe, I will give your process ago.

My process is a bit clunky and a bit of a text book PS layer / mask technique.

I usually do slow layer ISO800, stars and galaxy core, which I push the saturation up on and get colour then curves to darken sky.

Second fast layer is done with ISO1600 or ISO3200 and placed / aligned on top with previous a slow ISO800 as a mask, to bring out / overlay the detail and colour on the saturated parts. (I use feathering in the mask, so that is does not look obvious.)

I get a lot of colour this way without a star mask as such, although I do use the slow picture as a mask. (so very similar to star mask.)

Simple but works okay for Galaxies and Clusters.

[Pete Presland]

Thanks, I will try that i did think it was just something DSLR users had to put up with

Sent from my GT-I9100 using Tapatalk 2

[Russe]

Nice process, Russe...

What I do is I copy my layer I chose for my stars and apply a "dust & scratches" filter to the upper one. Then I change the blending mode to difference. Then I select the lower layer of the two and do "select colour". This way I select my stars. Once I've selected enough, I click ok, expand the selection by 3 pixels and feather by 2 pixels (all under select). Now I do ctrl+J and voila - a star layer, ready to be used and abused...

[lensman57]

Tricky one! With my DSLR they are nearly all white

Sent from my GT-I9100 using Tapatalk 2

They end up mostly white because a DSLR does not have the necessary well depth to both capture enough data but also not to saturate the wells. Even OSC CCDs with small pixels can suffer from this from time to time though CCDs being 16bit make a difference and are less prone to this. To retain true star colour with a DSLR either the sub lengths have to be picked very carefully and then a lot of subs stacked or otherwise a mono CCD with RGB filters used. The other stage is the processing where from word go stars need to be protected by a star mask of somesort or a layer mask of a sort in PS, other dedicated AP software have their own stuff. Aestethically pleasing star colour also depends a lot on the subject, the field of view and the importance of the associated stars in the final process, overstaurating the colours as a matter of routine will not do with the exception of some star clusters perhaps such as M13.

A.G

[Leveye]

It's your picture and you're the artist here so it's all subjective. I use the magic wand in PS invert the selection,refine selection, feather about 0.5-1.5 and then use  color saturation at about 60-75 percent. Sweeten to taste as it may.

[glowingturnip]

I say lots of colour in them, so long as you're not adding any saturation artifacts - the colours are real, it's just our eyes aren't dark-adapted well enough to see them.

Being a newly converted Pixinsight freak, there's a few things I've been playing with in there - the new Masked Stretch tool gives nice gaussian shaped stars without clipping the cores, so they can then be saturated to give nice colour.  I don't like the effect of that stretch on other stuff though, so make a star mask, whack that on, and do a transparency merge onto a more conventional stretch.

The curves tool on saturation works great too, can boost the left side of the saturation curve (ie under-saturated) whilst keeping the right part of the curve (oversaturated) down.

Have been reading about a technique that I haven't tried yet (but will be trying very soon), that probably works in Photoshop too - split out the image into CIE Lab channels, take the b channel (which i think is lighter = yellow, darker = blue) and increase the contrast in it with an S-shaped curve, then merge the channels back together.

[Russe]

Will try that Lab channel bit.

[lensman57]

I say lots of colour in them, so long as you're not adding any saturation artifacts - the colours are real, it's just our eyes aren't dark-adapted well enough to see them.

Being a newly converted Pixinsight freak, there's a few things I've been playing with in there - the new Masked Stretch tool gives nice gaussian shaped stars without clipping the cores, so they can then be saturated to give nice colour.  I don't like the effect of that stretch on other stuff though, so make a star mask, whack that on, and do a transparency merge onto a more conventional stretch.

The curves tool on saturation works great too, can boost the left side of the saturation curve (ie under-saturated) whilst keeping the right part of the curve (oversaturated) down.

Have been reading about a technique that I haven't tried yet (but will be trying very soon), that probably works in Photoshop too - split out the image into CIE Lab channels, take the b channel (which i think is lighter = yellow, darker = blue) and increase the contrast in it with an S-shaped curve, then merge the channels back together.

This works in phtoshop as well, but you use the contrast tool to do it on the b channel very sparingly, according to Olly penrice " the only time that one is allowed to use this tool ".

Regards,

A.G