After reading Vicent Peris brilliant post  (a must read for any Pixinsight nerd : https://pixinsight.com/examples/M31-Ha/) about how he processed his Ha Andromeda image, I decided to try this on one of my data set. Luckily a few weeks before Vicent's post I shot M81-M82 in Ha and was fiddling about integrating Ha in my LRGB set and more specifically how to inject Ha continuum substracted image in R and B (for Hb). I was already aware of Ha continuum substraction as I already used it in my Gecko project (https://www.astrobin.com/y2napd/) with my friend Victor (https://astrovirusblog.wordpress.com/) .
Victor also add used this technique for his Triangulum Galaxy work (go read his great post here : https://astrovirusblog.wordpress.com/2019/10/19/triangulum-galaxy/

Where Vicent post gets great is where he introduces two techniques to restore stars in the Ha substracted image (for all people unwaren you get star residuals artefacts after the continuum substratction).

After reading his post I decided to try it in order to post a "True Ha" with good looking stars of the M81/M82 pair.

Data set : 40*300s of Ha  + 90*60s of R. Both data set shot under SQM18.3 ... Thin Ha data but still some usuable data.

Processing

[*]

H data vs. R data : we can distinguish some distinct Ha structures in M81 as well as M82

H image (STF, 40x300s)

R image (STF, 90x60s)

The first step consits in continuum substraction of the H. To do so I used the usual PM formula. Q value is data and object dependant. I settled at 1.3 in this case but YMMV.

The resulting image is leaving a lot of artefacts in stars as you can see below

H image after continuum substraction (close up)

So far so good, this the usual technique used where you inject this image in your R (and L and B) data either at the linear or non linear stage

The next step is that you need to restore stars to get a standalone "True Ha" image.
Vicent used two techniques, one being truer to the data and consisting in restaoring Ha stars with R stars.
I diverged from this and used my original Ha stars for restoaration. I found this approach easier when dealing with noise profile compared to Ha stars.

As a first step I built the smallest possible star mask from my original H image

I applied the mask to the H continuum substracted image and then used PM with the following formula. With the mask active I only target stars artifacts and I apply only 5% of the original Ha stars intensity

H image after Ha stars restoration.



From here I used the deconvolution process after building a PSF. Note that I should have done this as an initial step in the workflow but I wanted to try decon on reconstructed stars

H image after decon of restored stars



To finish this little Pixinsight training session, I streched the image with HT.

I hope to do True Ha surveys of galaxies and nebulaes using this technic in the future.

Axel