Colour balancing - help needed!

[edarter]

I'm looking for some advice regarding colour balancing please!

I have started playing with Starnet++ in an attempt to allow me to push the processing of my targets without the worry of screwing up the stars, watching a few videos on this has also highlighted some things I was missing regarding colour balancing. On my starless image, the first thing I'm now doing is to bring in the black point on each R,G and B separately until its just at the start of the data.... all good so far and I'm doing this in small steps as per Steve's book Dark Art or Magic Bullet. What is generally noticeable though is that a) the white point is usually to the right of where the data starts and b) Green and blue histograms are narrower than the red. 

For the histograms should I adjust the white slider for green & blue to try and make the histograms approximately the same width as the red? I have tried it with an image of the Eastern Veil I'm working on at it looks ok colour balance wise (though loads of other issues manly down to my rubbish processing skills!) but I'm not sure if this is best practice or not.

One last question - adding the star only layer back in, the generally accepted way seems to be a new layer with blending mode set to either screen or linear dodge add. However, I found that setting blending mode to lighter colour seemed to nicely reduce the 'in your face' nature of the countless stars in the Eastern Veil area. Again, is this ok or am I inadvertently screwing something up that will become more obvious as my processing skills develop?

Thanks
Ed

[newbie alert]

You can colour balance using levels and adjust the separate red green and blue channels... I think you're talking about the mid point rather than the white, although with a starless image you can bring that in to make a brighter image.. remember when Steve wrote the book starnet didn't exist I dont think

With the blackpoint I try to give myself abit of space rather than bringing  it up to the foot of the histogram.. but I'm more of a narrowband imager, than a broadband and like to use carbonie's tools

[edarter]

Yep, that's how I'm doing it, by using levels. I'm just not sure if I should also be bringing in the right hand slider for green and blue channels to make them approximately the same width histogram as the red.

[wimvb]

I don’t use Photoshop or Gimp, but one thing to note about histograms and how the work.

histograms show the distribution of intensity levels. The peak is the most common intensity level in the image. In astro images with a background, such as star fields and galaxies, the peak of the histogram indicates the average background level. The width of the histogram is often considered a measure of noise. But if you think of it, a wider histogram means that there are more dark pixels and bright pixels. The thing is, we don’t notice the dark pixels. But we do notice the bright pixels. So, if for example the histogram for the red channel is wider than that for green and blue, there are more bright red pixels than green or blue, and the background will show a red cast. Even though, if you measure the average background, it will be neutral. To get rid of this colour cast, you need to adjust the widest histogram. One way to do is, is by aligning the right hand slope of the histograms.

pixinsight has a tool to adjust this type of colour cast. It is called scnr. scnr = subtractive chromatic noise reduction.

[edarter]

ok, so if I'm understanding you correctly, rather than widening the histograms for green and blue (and therefore potentially adding noise) I should be narrowing the red one?

Thanks
Ed

[wimvb]

Yes, that's the idea. But also, if you bring in the black point for red, it will shift the histogram to the left, aligning the right side edge with that of green and blue

[edarter]

Yep, already aligning the left hand edge for each of the channels, sounds like I was going in the wrong direction though by stretching the green and blue to match the red. I will now try reducing the red instead!

[ONIKKINEN]

Consider giving SIRIL a try. SIRIL is a free astro processing software that is very simple, effective and easy to learn (as there aren't that many functions). Color balancing with SIRIL is as easy as a couple of clicks with the "normal" mode and will work most of the time much better than manually fiddling with sliders in PS. The photometric color calibration tool however works every single time and makes the colors as real as they can be, since the data is from actual photometric surveys. With photometry the only way colors can turn out wrong (is there such a thing? Maybe not) is with very noisy data that the photometry does not apply well to for some reason.

In my opinion pretty much all of the YouTube tutorials about processing entirely in Photoshop are to be avoided because they make no sense at all. Photoshop is a powerful tool after linear processing (backgrounds, colours, stretch) has already been done.