An approach to subduing Halos?

[prusling]

Hi all, my first post here so be gentle!

My Oiii filter generates pronounced halos around the brightest stars and I'd like to be able to correct these through processing. I've seen a few approaches described but I can't seem to be able to apply them effectively, and I resorted to Google to see what other approaches there may be.

I chanced upon a description of what may be a novel approach (in Pixinsight as it happens) and I'd like to see whether we can achieve a similar process in StarTools. The full details can be found at http://astrodoc.ca/wp-content/uploads/2 ... -STARS.pdf but in summary it involves subduing the halos by by blending the Oiii data with halo-free data from another filter. I assume this would involve use of the Layer function. Apologies if this is already a known technique - if so, please can someone point me to a description of the process.

I can provide sample stacked Oiii and Ha images if anyone can spare a few minutes to give this a go.

[admin]

Hi and welcome to the forums!

There are many ways to accomplish this and much depends on the dataset and cause of the halos.
Do upload a dataset and we can see what technique is best suitable!

[prusling]

Hi Ivo, thanks for the reply. I've uploaded stacked samples of Ha and Oiii images to https://www.dropbox.com/sh/mjnku45jyxrx ... FSIea?dl=0 which should be accessible; I didn't take any Sii. Many thanks

[admin]

Here is one possible solution;

With Tracking off, first create a mask with the offending halos. It's usually quick & easy to click on their borders while "Flood Fill Lighter Pixels" is selected as brush mode.

StarTools_018.png (1.39 MiB) Viewed 9734 times

Next, in the Heal module, create a plausible background where the stars + halos once were.

StarTools_019.png (1.39 MiB) Viewed 9734 times

Next, launch the Layer module, set Layer mode to Overlay, click Undo > Foreground to put the previous image (from the undo buffer) into the foreground layer. Select "Where composite is dark, use background" as the Brightness Mask Mode.
You can subtly change the effect by varying the Brightness Mask Power, as well as the Mask Fuzz.

You'll end up with something like this;

StarTools_016.png (657.61 KiB) Viewed 9734 times

Does that help?

[prusling]

Many thanks, Ivo, that certainly removes the halos but it's a shame that it removes the associated stars as well. I'll keep trying with the 'substitution from another filter' approach and post the results.

[prusling]

Here's a trial using a similar mask, with the (stretched and saved) Ha as the background, blended in where the stretched Oiii foreground is light; I've set the mask fuzz at a roughly optimal point but there are still some artefacts from the halos. Would you recommend any alternative Layer settings to improve this?

I assume I'd need to undo the stretching before combining the de-halo'd Oiii version with the original Ha in the LRGB load for onward processing?

Many thanks for your assistance.

[prusling]

Here's a full process based on the blending of the Oiii halos with Ha signal (green interpolated)

The workflow I've used is as follows:
- Open the stacked Ha, manually Dev it to mid-stretched sky background, save it as "Ha stretched"
- Open the stacked Oiii, manually Dev it to match the Ha sky background, create a mask on the halos flood-filling the lighter pixels, use Layer function opening "Ha stretched" as background, blend using mask mode "where fg is light use bg" and adjusting mask fuzz to suit, remove stretch in Dev (tracking had been left on), save it as "Oiii less halos"
- Process original stretched Ha and "Oiii less halos" as usual

I've found that it is key to match the sky backgrounds so that when the Ha is substituted for the Oiii halos it is easier to match and there are less residual artefacts.

Keen for feedback on this approach

[admin]

Definitely a valid approach as well. I love seeing user come up with their own gems.
There are som any different ways! Another approach could be to use the luminance mixer (Red/Green/Blue Luminance Contribution) in the develop module.

Load, say, Ha as red and OIII as blue using LRGB module.
Using it, you can use just remove the luminance contribution per channel (in this case blue and green).

Then in Layer module, select Desaturate fg (Average), Cap Mode Normalize. This is the halo-free luminance frame.
Click Copy.
Click Undo > Fg; this will yield (with the current settings) the luminance frame with halos.
Click Paste > Bg
Click Copy, Click Paste > Fg.
Now we're ready to create a Mask with the halos in it. Reverse the mask.
Back in the Layer module, you should now have a halo-free luminance image (you can use some Mask Fuzzing if you wish).

All that said, halos are best prevented during acquisition of course.

[prusling]

Thanks, Ivo, please could you post the log from this approach

[prusling]

Credit should go to Ron Brecher (astrodoc.ca) for his article on this technique