Blue stars but no yellow or red stars

[mcbbcn]

Hi again,

I've calibrated, registered, combined and processed M31 following the NGC 1808 Pixinsight tutorial but when I increase saturation, I can see my blue stars, and stars that are supposed to be yellow or red are magenta, which may indicate too much blue signal.  I tried to use curves & levels in Photoshop but I can't get the stars to be yellow or red, they seem to be at the most magenta.  I don't know if it's because I'm lacking green signal, but I had to apply the Noise Reduction SCNR for Green average, because I had the same green artifacts that you could see in the NGC 1808 (2nd part towards the end).

By the way, when I tried to lower the blue, some of the most powerful stars start to turn green...weird!

Do you have any thoughts how to have a better star colors?

Here you have a pic of where I'm at.

Thanks for your help,

Miquel

[zvrastil]

Hi Miguel,

you're missing Color Calibration (often called White Balance in other softwares). I suggest to watch Color Calibration video from the collection of video tutorials by Harry Page, found here http://www.harrysastroshed.com/pixinsighthomenewbie.html. Other videos might interest you as well.

cheers, Zbynek

[zvrastil]

One more note. Don't try to fix already processed results with Color Calibration. Although it will probably improve the image colors, correct time to do it is after DBE and BackgroundNeutralization, but prior to Histogram stretch.

[mcbbcn]

Zbynek,

Thanks a lot!  I'll watch the video and try that.

I appreciate a lot the help!

Miquel

[mcbbcn]

Hi Zbynek,

After reviewing Harry's videos, I noticed that Harry does not seem to follow the linearize to non-linearize conversion that you see in the PixInsight video tutorial of NGC 1808.  So, I was wondering if this has any relevance at all.  The Pixinsight video tutorial in the PixInsight website seems to indicate that you can't apply reliable filter corrections if you don't convert the image to non-linearize, but I could be wrong because I'm still learning.

Please, let me know what you think.  I'm trying to figure out the right flow and I did not know if I could skip this step.

Thanks for your help,

Miquel

[zvrastil]

Well,

when it comes to image processing, math and rules behind it and there's difference between Vicent tutorial and rest of the world, Vicent tutorial is very likely correct :-). Or it does not matter. I'm not sure how it is in this particular case. I always do ColorCalibration on the linear image though. For example Deconvolution only have meaning on linear image. HDRWT or ACDNR work only on non-linear data. Some processes, like ATWT noise reduction can be done on both linear and non-linear images.
Maybe someone other will give you better answers.

regards, Zbynek

[Juan Conejero]

Hi Miquel,

To answer directly to your question: no reliable color calibration procedure can be applied to nonlinear data. Once you have stretched your image, you can only guess color corrections "by eye".

I was wondering if this has any relevance at all.

Yes, it is very relevant. For example, in your image the only solution is going back to the linear state, just before applying the first nonlinear transformation (usually with the HistogramTransformation tool). With the linear image, unless there are some instrumental causes for a color imbalance (for example, wrong filter usage, or improper image calibration and/or gradient removal), neutralizing the background with BackgroundNeutralization and applying ColorCalibration using the whole galaxy as a white reference should give you good results. This is clearly shown on the NGC 1808 video, and also on a recent video by Carlos Sonnenstein. Just two steps and you get an excellent color balance; it's just that simple: appropriate tools applied correctly make things easy. 

I tried to use curves & levels in Photoshop but I can't get the stars to be yellow or red, they seem to be at the most magenta.

This is just a manifestation of what I've explained above: inappropriate tools and techniques applied wrongly make things difficult.

[mcbbcn]

Thank you both!  I appreciate all the explanations and it's obvious that I have a ton to learn.  By the way, what is the best way to find out which tools are used for linear vs. non-linear images?  Is there some kind of reference?  Thanks for all your help!

[jtheios]

Thank you both!  I appreciate all the explanations and it's obvious that I have a ton to learn.  By the way, what is the best way to find out which tools are used for linear vs. non-linear images?  Is there some kind of reference?  Thanks for all your help!

Harry Page has a nice page on his website: http://www.harrysastroshed.com/pixuser/glossery.html

To that I would add:

Deconvolution: Linear only
ATWT: Linear or Non-Linear

Lately I've been using Juan's process for ATWT on linear images to reduce noise. It works beautifully.

-- John

[mcbbcn]

Thank you all. This is really good information!