The Wizard and a couple of stars

User images created with StarTools.
Carles
Posts: 214
Joined: Thu Jul 30, 2020 5:06 pm

Re: The Wizard and a couple of stars

Post by Carles »

Stefan B wrote: Wed Sep 11, 2024 10:50 am I honestly don't know how to explain differently :lol: :think:
haha OK :lol: if you want me to try with your files, you can send it to me by email and I can give a try :)

I prefer this method than shrink or starless then add stars.. is more "natural" but most people use starnet++ or similar to remove stars. Mostly Pixinisight people xDD
Stefan B
Posts: 470
Joined: Sat Oct 31, 2020 8:59 pm

Re: The Wizard and a couple of stars

Post by Stefan B »

Hi Carles,

I did a quick process of the Ha part of the l-eNhance image and the NBAccent image. The Ha layer was processed in Mono with all the usual modules like Contrast, HDR, Sharp, SuperStructure (DimSmall) and Denoise. Stars were also deconvolved and shrinked.

The NBAccent image was only lightly processed since we only need the color information of it, right? Stars were neither deconvolved nor shrinked since my understanding of your explanation is that this isn't necessary. In SuperStructure I adjusted Saturation. The image was denoised in order to avoid chrominance noise.

Subsequently I loaded both images into layer and used 'Color of Foreground' for blending. This was the result:
Layer2.jpg
Layer2.jpg (558.45 KiB) Viewed 2962 times
I guess we can agree that as you predicted there's more detail in the layered image and stars are much more reduced compared to the NBAccent image. I think that this could be compensated when stars are shrinked in the NBAccent image and when you go through all the modules like Contrast and HDR AFTER accenting the broadband image with the NB data which I haven't done here but in my original NBAccent image. Anyway, your prediction holds true and the result is consistent with your Bubble Nebula comparison.

I also had a closer look at nebula regions and some stars.
Layer1.jpg
Layer1.jpg (280.85 KiB) Viewed 2962 times
In this magnification you can unfortunately see the oblong stars of my broadband image :oops: :lol: Just ignore them. But what you can also see are some stars that have spikes in the NBAccent image but not in the Ha image. As you would expect the nebula gets the blue spike colors although there are no spikes in the brightness/luminance layer. That's exactly what I meant with it being a problem when chrominance doesn't match luminance.

It becomes also obvious here:
Layer3.jpg
Layer3.jpg (315.6 KiB) Viewed 2962 times
The central star is pretty small with no apparent halo in Ha but get's a strange coloring after layering. Looking at it you can sense that something's off here even if not pixel peeping and without knowing how the processing has been done. Some of the smaller stars appear with color halos since the broadband stars are bigger and the color of the complete star surface gets inherited to the luminance layer.

I don't say the method doesn't have its benefits. And maybe you can fix most or all of the issues. An obvious cure would be to shrink also the stars of the NBAccent image in the first place so you significantly reduce the halo effect. The spikes problem doesn't arise with refractors anyway.

But in my eyes the original NBAccent image is much more straight forward in terms of processing and it looks more natural. As I've said, the image was my first go at the data and almost certainly didn't take me an hour to get finished. I am a lazy processor so that approach appears more convenient to me. But if you are a tinkerer your method can certainly be put to good use.

Maybe I will give it another try on my next image. I was going for CTB-1, the Popped Ballon, and already acquired 11 hrs of duo NB data:

Image

Can you see all those stars (deconvolved and shrinked) in the l-eNhance image? When I use a broadband image as base those stars will drive me nuts in processing. Also that thing is much fainter than the Wizard. You almost can't see anything in a single sub. So your layering might be much more straight forward than the conventional NBAccent.

Regards
Stefan
Carles
Posts: 214
Joined: Thu Jul 30, 2020 5:06 pm

Re: The Wizard and a couple of stars

Post by Carles »

Hi Stefan,

so we can agree it works :) Obviously the issue with spikes is one I don't have as I use refractor (and I think there's some sort of mask for the newt spider to make spikes go away ? check this out, unless you like spikes (i dont haha ) https://www.youtube.com/watch?v=ILd34isTl14 )

about darker areas and so on, of course I showed only the last result, i think, and not the Layer module operation, anyway, you need to check the "Brightness star mask mode" <--- there you can check sth like " dark when fg is dark " bla bla... toggle those and check if that goes away.

Another thing, personally, I only use Decon or SVDecon ONLY if that accentuates some details, otherwise I don't, as it makes all sort of troubles with stars, for example, on your H-alpha image, there is star artifact on the core, that you could kinda fix withing SVDecon and mask values. But otherwise, I wouldn't use it.

To my eyes, the final result looks much more detailed and contrasted than the original NBaccent only image. Also, remember you have control there.. you can not only test Brightness mask, but you can decide how much of that colour will go, clipping and so on.

For those specially bright stars that got darker, you can always use a mask and recover the brightness.

I know what you mean, for me when is an image "in the middle of the process" of integration, i go fast, and want to see result, I will get more picky when it's an image that I consider project done and it has many hours on it. Then, i'll do whatever I can to get the most of the data :)

Anyway i'm glad you tested and the results were as expected :)

I have tried that Balloon one haha but from Bortle 8 xD gathered 5h I think and barely has signal haha

Clear skies! :)

Carles.
Post Reply