Typical NB processing with PixInsight

[topboxman]

What are the typical processing steps using PixInsight for NB images. I can't seem to find a good tutorial that's consistent for NB images. There are several different methods but I don't seem to do them properly.

The following link shows sample processings of Bubble Nebula. The details are there.

http://peternagy.smugmug.com/Telescopes/Nebulae/15823498_3vBdTQ#!i=2137856533&k=ztLNFbB

If you see the first two NB images, notice the colorless parts, these are mostly from Ha. I mapped as follows:

R = Sii
G = Ha
B = Oiii

When I do LRGB combination, the result is almost all green. I believe the majority of green was dominated by Ha. After stretching with Histogram Transformation, I run SCNR and it resulted what you see at my web site. It strips out a good portion of Ha and showed colorless parts that seem to defeat the purpose of capturing Ha. I believe the yellow parts are from Sii and the blue parts are from Oiii.

What are the typical Pixel Math for NB processing? What's a good Pixel Math to begin with?

My first NB image at my web site is unusual processing. It includes Ha, Oiii, Sii, Red, Green and Blue. I made it all up and got strange and decent results. I know there's no right or wrong way. I used Harry's Adding Ha to RGB method for all broadband filters. For example, Ha was added to Red, Sii was added to Green and Sii was added to Blue using Harry's tutorial or Vicent Peris method. I mapped them to:

Red = SiiGreen
Green = HaRed
Blue = OiiBlue

I know this is crazy.

The second Bubble Nebula at my web site is purely NB, no RGB were included. It seems to have less color than first Bubble Nebula.

Do my NB processes of Bubble Nebula look correct to you?

Also at my web site, you will see other processes like HaRGB, Bi-color using Ha and Oii, Ha, Oiii and Sii.

Thanks,
Peter

[Nocturnal]

Hi Peter,

creating an RGB image out of 3 mono images regardless of their content should be pretty trivial, especially when the signal levels are comparable. When you used the LRGB tool I assume you unchecked the L channel, right?

I haven't yet tried this myself but correcting color balance issues would seem to be a similar task regardless of what is mapped to R, G and B. I would run Background Neutralization to get the background gray. Then do ColorCalibration with some bright stars to balance the other channels although you may not like that result. I think the key is to balance the channels to get a pleasing result. Perhaps using the histogram tool on individual channels is a good option.

[topboxman]

Yes, the "L" is unchecked.

NB processing is a totally different animal than RGB processing. RGB processing is a piece of cake. NB seems to require tweaking depending on signal strengths of each NB channel. I have a lot to learn about NB processing.

Peter

[Nocturnal]

Hi Peter,

well there's no sense arguing if you feel your RGB skills don't apply here and I think they do. I'm sure someone else will jump in and help out. I'll be interested to read the rest of the thread and learn from it.

[topboxman]

Hi Sander,

The thing is that all three channels are properly balanced according to Histogram. Ha is mapped to green channel and dominates heavily over Sii and Oiii and makes the the whole nebulosity green. At this stage I believe this is normal. I have read that people play with hues to change from green to to more desirable Hubble palette color. I have never mess with Hue controls when processing RGB images so I am unfamiliar with Hue controls under Curves Transformation.

When I use SCNR to remove green, it pretty much removes most of the green and makes them somewhat colorless and semi-transparent. I think I need to play around with the SCNR parameters.

I am surprised there is not much PixInsight tutorial for NB processing. I am sure the processing is a little different from DSO to DSO but some kind of good guidelines would be nice.

Peter

[topboxman]

If you would like to have some fun processing my images, I uploaded a zip file of Bubble Nebula containing the following:

Calibrated and stack of:

21 x 15 minutes of Ha: Good seeing condition
31 x 15 minutes of Oiii: Below average seeing condition
12 x 30 minutes of Sii: Average seeing condition
8 x 10 minutes of Red: Ok seeing condition
8 x 10 minutes of Green: Ok seeing condition
8 x 10 minutes of Blue: Ok seeing condition

The zip file is 115 MBytes.

http://www.sendspace.com/file/y2aji9

Click only on "Click here to start download from sendspace" link at bottom of web page. Do NOT click on any "DOWNLOAD" or "Play Now"buttons.

Thanks,
Peter

[Nocturnal]

Hi Peter,

well I certainly would like to give that a try. Thanks for uploading.

[topboxman]

Note that I didn't include a stack of Luminance because the seeing was so bad that not only the stars were badly bloated but it barely showed nebulosity. The red channel showed a lot more nebulosity than Luminance. Also RGB were unbinned at 1x1.

Peter

[Nocturnal]

Hi Peter,

I don't think L is often combined with NB because that negates your expensive filters. Why capture these tiny slices of the spectrum only to then combine them with the entire visible range as a brightness modifier? That doesn't make sense. So please don't feel you need to upload the L for my sake. I wouldn't use it :-)

[topboxman]

I mentioned the "L" channel because I included RGB FIT files inside the zip file I uploaded. I figured you would wonder why I didn't include "L" if you want to process it with RGB in addition to processing NB.

Peter

[Nocturnal]

Gotha Peter, thanks for explaining.

[viktorbravo]

FWIW, on bi-color Ha and OIII images I use Pixel Math as well to create and combine the channels.  Also I found that working with the linear images is pretty restrictive in getting the colors that I wanted so I played around with some things.

Then I stretched and completely processed the mono images before combining and the amount of stretching in each image will dictate the strength in each channel produced in Pixel Math, giving you a lot of layers of control, almost too many.  I also found that for bi-color, the very high signal areas in each mono image need to be toned down somewhat before combining and I use HDRWT for that.

I assume that after combining linear images in PM one could also use the live preview in HT and adjust to taste before making them non-linear, producing the same effect.

I have no idea if this is the correct way to do it, or if there even is a correct way.  I am not an expert in any stretch of the imagination as I just started using PI, so I am just learning this by some trial and error. One thing is for sure, you have lots and lots of control and that can be both a good and a bad thing.

There is lots of trial and error to stretch each mono image to taste but I've been pretty happy with my results so far.  Of course YMMV.

Here is my latest effort: http://www.rattlesnakeobservatory.com/astro/ic-1805/

My first effort: http://www.rattlesnakeobservatory.com/astro/nebulae-3/m-8/  Need to re-visit this one.

The beauty of color NB is that there is no correct color balance so find a method that produces something that you like, and stick with it.