LRGB Imaging

[Simon Pepper]

Hi all,

I have been a mono imager now for a long time and when it comes to NB imaging, I like to think I have got this down, however when it comes to LRGB imaging/processing I can never seem to get happy with my results. I am sure the main issue is I always seem to have gradients or rainbows over my image and it makes processing difficult to get it where I want it. I have looked on YT for videos but I cant seem to see a decent process. My usual workflow is as follows: crop, blurx on each channel noisex, then I combine RGB, but its at this point I see these issues, magentas, greens, blues, browns is this normal (I know it is for me)? Even after SPCC nothing changes for me that I can see its still very much rainbow colour. How far does one go at desaturating and processing the RGB image am I missing a step here? How far does one go with the Lum image? Im after that dark space background not clipping, RGB stars and dusty grey dark nebula mine always seems to go brownish, but these colours below are abismal! 

Filters are Antila v Pro LRGB and the image below is just combined RGB.

Thanks 

 

Edited October 6 by Simon Pepper

[CraigT82]

Have you tried any of the gradient correction tools in PI? There’s ABE, DBE and gradient correction itself - that I know of (there’s prob more I’m new to PI) 

[symmetal]

Yes, with NB SHO processing, quite often I find I don't need to do any gradient correction, but RGB seems to make slight gradients very noticeable.

GraXpert is very good at removing gradients with just 1-click. It can be used stand alone or as a PI plug-in though the stand alone version needs to be installed to use it from PI. The stand alone version can be downloaded from here. Click on Releases\Latest on the right hand side to get the latest Windows installable, or other, version. Incorporating it for use within PI is well documented here.

PI's DBE is also good though requires a lot of work placing control points over the image. Seti Astro's Auto DBE script turns this into a 1-click process too, though you'll need to mask out the dark nebula areas before using it. Scroll down to find it listed, along with a handy video on how it works.

If gradients are minor then gradient correction can be applied after the RGB image has been created in ChannelCombination, while bad gradients may benefit from correcting them on each channel individually beforehand. A final correction may be needed after ChannelCombination too for any residual effects.

Alan

Edited October 7 by symmetal

[Simon Pepper]

Hi,

I used SETI Astros ADBE on the image, but I do it at the linear stage when all channels are separate perhaps I need to be doing it once combined to RGB... Interesting that you also see gradients more on RGB images perhaps its just normal! I have spent hours on this image now and I am still not happy with the results. My latest iteration thats destined for the recycle bin. I will try gradient removal once combined to see if that helps

Thanks 

[Pompey Monkey]

There is no reason not to use gradient correct at any point in your workflow 👍

[symmetal]

After using ChannelCombination to create the RGB image, it will still be in a linear state. It's possible that applying Gradient correction on the RGB image is no different to applying it to the individual channels beforehand. Gradient correction is usually applied before background neutralization, so likely it isn't possible to cross reference gradients between channels in the RGB image anyway.

Only when using LRGB combination does the image need to be non-linear, (stretched) beforehand. Applying gradient correction again after stretching, may be beneficial as the scale of the gradients will have been stretched too, and could be easier to remove. Applying it too often during processing will likely cause a loss of actual information though, so needs caution.

Alan 

[ollypenrice]

I'd be wary of using BlurXT on the RGB layer for processing the nebulosity. I'm not looking for detail in the RGB, I just want intense colour and low noise. Detail and signal depth will come from the luminance.

I'd fully process both the L and the RGB separately with these priorities in mind, de-star them both and then combine them. I'd then do a final process on this stalrless LRGB.

Finally I'd do a BlurX version of the RGB just for the stars and apply that as the final star layer. I see no point in using L for the stars because it just dilutes star colour.

Olly

[Vroobel]

I use the BX and Gradient Correction before other steps if I'm going to use the channels as masks. The NX may affect your colors if you work on them before or after applying it, just try. 

I follow the processing presented perfectly on the PI authors' tutorial series on YT - who knows better then them? 🙂

 

Edited October 8 by Vroobel

[Elp]

2 hours ago, ollypenrice said:

I see no point in using L for the stars because it just dilutes star colour.

Isn't one of the reasons for using a luminence filter is to reduce UV/IR star bloat?

[ollypenrice]

21 minutes ago, Elp said:

Isn't one of the reasons for using a luminence filter is to reduce UV/IR star bloat?

Perhaps so if your blue and red filters are not cutting off the UV/IR and your optics are not well corrected. Many filter sets do seem to have a luminance filter which cuts off the UV/IR a little short of the B and R.

However, the L filter lets through far more light and will more readily over-expose stellar cores and bleach out colour. I've always found it preferable to use RGB-only stars where possible. Modern star-removal software makes bloat-control much easier since bloated stars can be isolated and given a contrast high enough to drop the bloat below the level of the background sky. The individual is best advised to experiment, I guess.

Olly

Edited October 8 by ollypenrice
Clarification

[Pompey Monkey]

1 hour ago, Vroobel said:

I use the BX and Gradient Correction before other steps if I'm going to use the channels as masks. The NX may affect your colors if you work on them before or after applying it, just try. 

I follow the processing presented perfectly on the PI authors' tutorial series on YT - who knows better then them? 🙂

 

I see any tutorial on post-processing, i.e. after creating the masters, as guidance. There is always more than one way to skin a cat*. This seems to be especially the case now that the A.I. tools are so powerful and have multiple use cases.

*Apologies to the cat-lovers.

 

[Simon Pepper]

Very interesting I had not thought of processing RGB this way. I have always processed each channel separately. I will combine them before processing and ignore Blurx to begin with. I gave it another go last night and it’s getting better there is clearly an art to LRGB processing it’s definitely harder to obtain the results I want. I’ll post it up later, but will also try @ollypenrice flow…

[Pompey Monkey]

38 minutes ago, Simon Pepper said:

 ...there is clearly an art to LRGB processing it’s definitely harder to obtain the results I want. I’ll post it up later, but will also try @ollypenrice flow…

It's not easy, and something that I still haven't really got my head round. For me, I think being an engineer with almost zero art ability can make it doubly so.

But to pinch a phrase from the cycling world (which @ollypenricewill know all too well), "If it doesn't hurt, you're not trying hard enough." 🤣

[Simon Pepper]

Here is what I have come up with much improved but seems a little soft still not sure I will ever be happy with this one! I will try all the suggestions above thanks for giving me some ideas to try

[TiffsAndAstro]

Fantastic image. Doesn't look soft to me.