NGC 7822 / Ced 214

[drmikevt]

This image represented my first ever start to finish SHO image captured from my backyard.  I'm very happy to finally be at this point.

I know many people prefer a less colorful interpretation of the SHO palette, but I like these colors to better differentiate the filter lines. 

Larger one http://www.astrobin.com/301671/?nc=user.

70x600s
SV80ST
MyT
QSI 683wsg-8

Comments and constructive criticism appreciated.

[Niall Saunders]

Hi,

I know many people prefer a less colorful interpretation of the SHO palette, but I like these colors to better differentiate the filter lines

Well, given that the human eye cannot make these interpretations of colour itself, then any representation (within reason) has to be accepted.

You have worked hard to achieve a fine image - and I certainly would not ne criticising it on any basis of 'colour interpretation'. Your processing of structure (both large-  and small-scale) shows that you have worked hard with your data (or at least worked hard to acquire high-quality RAW data in the first place). Similarly, your treatment of the inevitable noise present in all of our un-processed images shows a good (but sympathetic) understanding of the processes available to you within PixInsight.

As usual though - when faced with images far superior to my own efforts, I do like to remind you that you 'could have done better', if only to console myself in my personal failures 

Well done!

[Nocturnal]

I like the colors you came up with. What was the initial channel blending you used? I've been working on NB imaging since about April and getting this color mix has been a bit of a challenge. Thanks!

[drmikevt]

Sander

Thank you!  I've been struggling myself.  I'd like to get to a place where I can dial in the 'turquoise and gold' motif that I see some people produce, but that (currently) seems most easily done in the 'that other program'. 

For this (and all my processing) I follow Vincent's advice:  SHO maps to RGB in Channel Combination.  Then, DBE, BN and Color Calibration using an area of the nebula that is well represented by the 3 filter lines for white (without structure detection, of course).  I usually will select 3-4 different areas of 'background' and 2-3 different 'white' areas and then combine them in various ways using Color Calibration.  It is sometimes amazing how different the hues and contrast can be when using different previews.  I will tile all the new Color Calibrated images on the screen and compare them to select the one with the richest color and contrast.  I try hard to get the best color possible from the get go.

Then, for stretching, I'm starting to experiment Masked Stretch instead of HT for color.  The image above, though was stretched using STF settings transferred to Histogram Trans.

The real work comes after stretching with the Color Mask script.  This is also something I am just learning, but seems to be the key to adjusting colors for NB image with any sense of control.  Simply following the steps on this page, https://pixinsight.com/forum/index.php?topic=7751.0, work very well to get to a better image, and to help understand how the script works.  I have found it is important to use previews to get zoomed-in views to make sure you are not over-saturating anything.  Then, I will do some further color correction using Curves with and without a masks.  I have not had great success with the AIP scripts. 

Below are 2 more posts.  The first is the data stretched with STF (Target Background .28-3 and Shadows Clipping at -2 to -1.5) and HT.  The second is the same data (preprocessed a little differently, but still) put through Masked Stretch at .28 target background and .03 clipping percentage, and then given a small S-curve.  In both cases, no other color manipulation was done outside of Background Neutralization and Color Calibration.  Both were stretched to put the peaks at around .28.  As you can see, there is a dramatic difference in the two results.  In the first case, using just BN, CC, STF and HT, you get pretty close to the image I posted.  How are you trying to do it?   I am just about to start working with the MS stretched result to see what I can do with the much more intense colors.  I'll post results in a few days. 

I'm still learning every day, and you certainly know tons more then me, and I don't mean to tell you things you already know, but I hope this helps.
Mike

[drmikevt]

STF and HT

[drmikevt]

Masked Stretch and S-curve

[drmikevt]

Here is a complete re-working of the data using quasi tone-maps and the SHO-AIP script followed by Color Mask and non-masked curves.  I was shooting for the elusive blue/gold palette which seems to be a struggle for most without using that other program.  Start to finish processing in PI.   Some of the stars have a bit funky and I plan to eventually replace them with RBG stars.  The down-scaled version below is a bit darker than the larger one on A-bin, http://www.astrobin.com/301671/C/

Enjoy
Mike

[Nocturnal]

Thanks Mike for taking the time to post so many details! I have some NB data I'm going to try to process this way and will let you know the results.

[Nocturnal]

I think my main surprise is that you used a straight SHO->RGB mapping to get started. What I've read so far is mixes of 80-20 and 20-80 for the SH->RG. This doesn't make a lot of sense to me as you are affecting the intensity of one color based on the intensity of another color. It rather defeats the purpose of capturing those channels separately before. Then again I also think that when I see Ha-RGB type combinations or where Ha is applied as L on a color image of some sort. It all sounds a bit more like alchemy to me

[drmikevt]

Yes, I have been doing a lot of thinking about how to represent NB data.  More or less, you have a choice between trying to represent the data in the most 'true/scientifc' way, in order to show how the data from different filters differ, or you choose to make a pretty picture and decide to let go of differential color information. 

But, this assumes we agree on what a pretty picture is.  As I've said, I think the first pic is quite nice and I like the full range of colors (despite it having a number of sig. processing issues).  However, I'm also excited that I've figured out how to get to the gold/blue palette that is so popular (and pretty). 

I should say that I think I have noticed that SHO to RGB mapping goes best (for me) when there is very little data manipulation prior, which is to say there is no attempt to match the channel intensity, and the combination is in the linear stage.  You just deal with the raw data (assuming, of course, that you have reasonable data from all 3 filters).  However, when going for the blue/gold palette, it seems quite important to first stretch the 3 channels to a similar level before combining, and then I've seen the best results (so far) with the SHO-AIP script using about 50/50 S/Ha for red, 50/50 Ha/O for green and 120% for O3 in blue (with a pre-made, combined Lum) (and then following that with various Color Masked curves).  My workflow is evolving quickly and I hope to post instructions for achieving blue/gold using only PI soon.

Mike

[Nocturnal]

Yes, lots of opinions on how to combine data and what is pretty etc. I think combining channels rather defeats the purpose of having narrow filters to begin with so I'm resistant. Put it this way, if you could buy a filter that has notches at both S and Ha, would you buy one? It could make your Red channel in one shot. I realize that having the split masters then gives you freedom to combine in any way you like but it seems that at least some people don't really understand what happens when you combine channels. I mention Ha-RGB again because that is an example where a dim Ha section would knock down a bright GB section if Ha is applied as L. It really makes little sense to me

I look forward to your upcoming instructions!

[drmikevt]

Another reprocessing.   Certainly, there are issues, but the purpose was to work on star removal, dynamic compression and color balance/hue.  Star were removed for the SHO images but added back with a synthesized Lum.