Suggestions for handling color data in galaxies?

Wade Hilmo

I am looking for some help improving my handling of galaxy color.

I am satisfied with my image capture gear and techniques.  I am comfortable with my skills up through creating integrated masters for each channel.  But once I get past what I consider the technical part, and have good masters, my processing limitations become apparent pretty quickly once artistic decisions become important.  I feel like I can manage processing luminosity reasonably well, but I have not been able to grasp how to handle color well.  In particular, I have a terrible time giving blue data a proper treatment.

An object that shows my limitations more than most others is M31.  People consider it a good target to start with when learning, presumably because it's so bright and easy to find.  To me, it is easy to get an image of M31 - but it is also one of the most difficult objects to process and get *right* due to both the color and sheer dynamic range of the luminance.

Over the last few days, I've been going through some of the new features in the latest PixInsight release and wanted to try MCC.  I picked some M31 data that I collected a couple of years ago and have been reprocessing it from scratch.  I have finished the RGB master as a starting point, and done some basic preparation.  It's at that point where I need to start making decisions that have ramifications later.

I've attached a sample of where I am at.  This is the integrated master RGB.  In looking at the HistoryExplorer, and here are the steps that I have done since integration:

•     BlurXTerminator in "correct only" mode to clean up stars in the corners
•     ImageSolver
•     SpectrophotometricFluxCalibration
•     MultiscaleGradientCorrection
•     SpectrophotonetricColorCalibration
•     SCNR (to remove green noise at 50% strength)
•     HistogramTransformation to apply autoSTF
•     IntegerResample do down sample by 4x
•     Save as a JPG at 80% quality

This is the point where I kind of throw up my hands.  I am comfortable processing most aspects of it, but my main goal for this image is to learn something about handling color better.  Every time I process M31, I go into it hopeful and come out disappointed, specifically with the color.  It's hard for me to express why I am disappointed, except to say that the color wrong, especially in areas that should be blue.  I suspect that my inability to quantify the problem is related to my inability to fix it.

I have been rewatching your M83 processing session this morning, as I believe that it comes the closest to handling this kind of data.  I suspect that I am getting off track pretty early in my processing for color, and it just gets more apparent in later processing.  I understand the effect of luminance on saturation, but I have absolutely zero intuitive feel for it.  I've done lots of trial-and-error, but feel like I've hit a dead end, so I figured that I would ask here.

I'm wondering if you have a recommendation for content I could review that does a deep dive on processing colors in galaxies.  The attached sample shows where I starting (after integration).  I acquired it with an 80mm refractor at F/4.8, with an ASI2600MC Pro.  The xisf is huge, at 1.2GB, but I am happy to make it available if you (or anyone else here) wants to take a look at it.

I have purchased your Fundamentals course, and have a current subscription to Horizons.

Thanks for any suggestions,
-Wade

BlockHead

Your image does look like pretty good data to me. I can certainly see the bluish OB associations that you can boost in terms of color contrast. Yeah, there are many different approaches. As I mentioned in a recent e-mail I will be making "Deep Sky Academy" in which I hope to address questions exactly like this. 

Unfortunately there are a great many things happening right now. On this side of my site, I am trying to change it, and I need to update it since PixInsight changed some stuff. So I will get there...I do eventually fulfill my promises (I did with Comets and later Narrowband content). 

On suggestion I have is to create a color image to blend with the Luminance that is really really ugly in terms of brightness...but represents all of the colors. Too many people try to keep the color image looking good...and it does not need to. Only the blended result matters. In addition, I demonstrate additional ways to boost the contribution of color with respect to ImageBlend. You can take and image and multiply it (soft light or even screen) by itself. And this image can be used in part or whole to enhance the original. Finally... you might also consider utilizing Masked Stretch for the color image also for bright objects like this. 

M31 in particular is very bright and also has faint stuff. It is more difficult than M83 to process.

M83 is only one instance... there are no objects in general that stand in for a typical example. Instead... the way to approach processing images is that you are looking for image ATTRIBUTES (features)... bright, faint, saturation... etc. These attributes/features is the way to navigate processing. Don't look at galaxies as a type... decompose any object as a set of features/attributes. This is the way I intend to teach the course.

-the Blockhead

Wade Hilmo

Thanks for the response.

This particular image is from my wide field rig, which has the OSC camera.  My normal imaging rig has a mono camera with filters.  When I'm processing OSC data, I am already in the habit of extracting the luminance and processing it separately from the color.

And your point about galaxies as a type, versus features and attributes is well taken.  I don't image many reflection nebulae, so most of my experience with blue features is in spiral galaxy arms.

Your "Deep Sky Academy" series sounds excellent!  But please don't take this as pressure.  I know that you have other priorities.  There is no rush on my end.

Thanks again,

-Wade

BlockHead

OSC color is already a detriment... there is a loss of contrast right off the top for a number of reasons. 

-the Blockhead

Wade Hilmo

How so?

I've been under the belief that modern OSC cameras and drizzle integration are on just about equal footing with mono cameras for strictly RGB imaging.

BlockHead

That requirement for properly drizzled data is a killer.

And you also know that the colors have a crossover between filters- decreasing color contrast.

Fine...I am probably behind in my assessment. My experience to date has been the colors simply lack contrast - I think one issue is the sampling. Take HII regions of a galaxy. If you image an HII region with a medium to short focal length they disappear a bit with OSC. Why? Because only one or two pixels have that red information. It is so easy to loose this in the shuffle of registration/drizzle since the nearby pixels are of the other colors...and you need to jump several pixels to the next red one. This is a loss. In terms of spatial information it isn't too big a deal. But if the colors are changing on small scales...they get mixed more with OSC. This happens less I feel with mono- where each pixel is doing its thing in each filtered image. 

-the Blockhead

Wade Hilmo

Thanks!

James Mauck

Wade,

I can see good detail in your image, and I’d love to work with the data!

If you haven't seen it already, Adam has an excellent GHS walkthrough titled "GHS and M42" in Stretch Academy, which I believe highlights similar challenges with M31, I.E., protecting the bright galactic core and overall colors.

I have a write up of my OSC workflow if you are interested for comparison. Let me know.

Best,

Jim

Wade Hilmo

Thanks for the pointer to the GHS Walkthrough.  I try to follow all of Adam's content, but I am behind and haven't yet watched the Stretch Academy videos.  I will check it out in the next few days.

As for sharing the original XISF file, I am working on it.  I have run into a bit of a problem in that my imaging site's coordinates are shared in the FITS headers, and I cannot get PixInsight to either edit or remove them.  It looks like they are being automatically written into the file with their original values when I save it.  Even if I remove them, they are re-added.

I've started a thread on pixinsight.com to see if there is any way to strip out these (and some other) headers.  I am sure that I could strip them out by converting the file to a different format and back, but I am trying to preserver the processing history (as short as it is).

Thanks,

-Wade

James Mauck

Here is the process to remove location headers that works for me...

In PI, select File | FITS Header menu item while the target file view windows is highlighted. Locate the highlight each SITExxxx header (SITEELEV, SITELAT, SITELONG) and click Remove button. Alternatively, edit (redact) the undesired values. After all are removed/edited, click Apply (Square) button, and followed by File | Save As to preserve the FITS header changes in a new filename. Close out open file, and re-open the newly saved file to confirm location header changes are preserved.

Best,

Jim

Wade Hilmo

It is not that simple.

If you read the topic over on pixinsight.com that I posted above, you can read the story (which is still unfolding).  The gist of it is that if your XISF file has an astrometric solution, there are a number of headers that will be written to the file with values from that solution.  Those will overwrite any changes that may have been manually made to the headers before saving the file.

Most of the headers are harmless, but there are 4 of them that contain latitude and longitude coordinates to six decimals.

Before raising the issue, I changed the coordinates to all zeros after the decimal, so that they had precision only to the nearest full degree.  PixInsight will silently replace them with the original six digits of precision when you save the file as an XISF (again, only if there is an astrometric solution, which my file has, since I am trying to include the history with application of MGC, SFCC and SPCC).

Currently, there are four ways to avoid this:

1. Remove the astrometric solution, re-edit the headers, and re-save the file without the solution.
2. Do step 1 and then re-run ImageSolver to add the solution back.  If the coordinates were not in the headers before ImageSolver runs, then they will never be added to the file.  Of course, this will all show in the history.
3. Uncheck the embedded data in the XISF save option.  Of course this removes stuff that I was hoping to share with the file.
   
4. Alternately, save the file in a format other than XISF.

 

James Mauck

Ah, I see now, Wade...I did not follow through reading your thread.

Thank you for explaining further!

Wade Hilmo

Sorry for the delay in sharing the file, but I wanted to wait for the thread on pixinsight.com to play out.  At this point, it looks like that's the case.  I've made my comments to the PI folks, and they may or may not make do anything about it.

In the meantime, I have saved the file as FITS (with sanitized site coordinates) and uploaded it to my OneDrive.  I'll leave it there for about a week before taking it down.  I don't have a lot of storage space on OneDrive, and this file takes up a fair amount of it.

You (or anyone) may download it from here.

If you have any problems accessing the file, please let me know.

-Wade

James Mauck

No worries, Wade.

I have downloaded the zip and will start working it this weekend.

Best,

Jim

James Mauck

I made a quick run at it. I've uploaded the processed XISF here:

https://drive.google.com/file/d/1NSNVA_h9PHWw4ALKfws2dNvJzLhMb26b/view?usp=sharing

I leave it up there for a bit, let me know if you're able to download it.

Hopefully, my image inserts work below...

Most of the processing is my normal workflow.

Here are comments on my workflow:

I attempted a precise gradient correction in DBE by selecting background samples far away from the galaxy halo.

I attempted to preserve colors in the first major stretch using GHSs Colour blend, which is fully described in Adam's Stretch Academy. I do GHS processing on starless image as it is much easier.

After GHS, I did light sharpening on extracted luminance and added back with LRGBCombination.

I applied CT to my artistic eye, highlighting a warm core and blue/red nebulosity boosts.

Added back lightly stretched and color preserved stars.

Best,

Jim