›

White-clipped reds after continuum subtracting and then merging Ha data into RGB

Jeffrey Nibler
in PixInsight
I have been using the method from the tutorials to integrate Ha data into my broadband data, specifically for Ha regions within galaxies, and using continuum subtracting and all the related aspects of it (noise reduction, background subtraction, etc). 

I have many, many hours of Ha data for my current target and the Ha regions are very strongly donut-shaped on this galaxy. They have a hollow(ish) core and very strongly defined outer ring. What's happening is that strongly defined outer donut portion of the Ha regions is so strong in signal that it is white-clipping once it's merged in with the RGB data. 

I could play around with the factor I use for blending the CSUB Ha image into the RGB image so I get less Ha strength, but that would reduce the detail I get for the lighter, broader regions of the Ha and I really don't want to lose that. What I really need here is the ability to do HDRMT on the CSUB Ha data before I merge it into the RGB data but you can't use HDRMT with linear images. I have tried using range masks on the CSUB image to mask the brightest portions of the Ha data and then pull them down with curves adjustment and while I can see that working in the CSUB image, it still isn't enough to prevent white clipping when I then merge it in with RGB. 

After merging, when I do my initial stretch, I have tried STF and GHS, and when using GHS, I have tried adjusting the Local Intensity to get less of a stretch on the brightest parts but it's just not enough. 

Any tips on this would be greatly appreciated thank you!

Comments

  • Jeffrey Nibler
    For example, in this video: https://www.adamblockstudios.com/articles/blending-halpha-using-nb-colourmapper-after-continuum

    .... Adam mentions something about "in this workflow, I didn't do special steps with HDRMT" but the video was all done in linear mode so I'm wondering what special steps could be taken to reduce contrast like HDRMT can?

    I've been playing around with the NBColorMapper Script version of the Ha-RGB merging in the meantime.
  • BlockHeadBlockHead
    Yes, you can certainly stretch the Ha and blend it in (using screen) when both the Ha and RGB are non-linear. This is another technique. 

    You can both colorize the Non-linear Ha and blend it into the RGB using the ImageBlend script. This is the best way to go for this particular situation.

    -the Blockhead
  • Jeffrey Nibler
    Thank you Adam, I will give that a shot. 
  • Jeffrey Nibler
    Here is what I came up with. What do you think about the Ha blending on this? Do you think there is room for improvement? You can still see a bit of white clipping but it's a lot less than I was getting before. The subject is Holmberg II. One of the most challenging parts of this, is getting the luminance blending right. It's an art to do normally, but when you add Ha to the mix, it's even more challenging. I believe I am using the approach you had suggested to me a while back which is to blend the Ha with the Lum before integrating the Lum with the RGB. The trick there being getting the ratio right - for the Ha regions, you have brightness from the lum, brightness from the RGB image, and brightness from the Ha image. This makes it very difficult to avoid clipping. 

    The process I followed was to perform the continuum subtraction on Ha the fully remove the background. Next stretch RGB, stretch Lum, stretch Ha. Slightly flatten Ha with HDRMT and then pull down a bit of brightness with curves. Merge ~0.25x Ha with Lum via pixelmath. Use Channel Combination technique to integrate lum with RGB, then use image blend script to merge Ha with the LRGB. 

    Please don't hold back on critique! Nit-pick away. The two things I would like to improve on here are that the Ha regions look a bit too solid/dense, and there is still a bit of white clipping happening. I'm open to all other critiques, relating to the Ha merge or anything else. 

    Here is a link to the full size image:

    Here is a very small screenshot of the Ha after continuum subtraction, background removal, and some flattening. I can upload the full image if desired. I tried to attach it to this post but it looks like maybe the attach a file link isn't working. 

    Thank you again for your help!
Sign In or Register to comment.