Pixinsight narrowband combination

[Ryderscope]

Some advice please on workflow options in PI when combining Ha SII and OIII. I have been fiddling with palette combinations (using pixelmath) with my data from IC2872 and am not happy with results so far. Before I get too far down track though I was wondering if there is a general consensus on workflow options. For example, would it be best to ...

a) Process each of the Ha SII OIII masters separately for noise reduction, gradient removal, stretch, enhancement etc, before colour combination and further processing; or

b) Do the colour combination first of (using pixelmath) and then follow on with the usual workflow of gradient removal, noise reduction, colour correction etc.

Many thanks in advance.

Rodney

[RickS]

Hi Rodney,

Here's a rough outline of my NB workflow.

Lum: I combine the initial linear masters to create a synthetic luminance. Depending on the data I might do that in a variety of ways. If there's little extra detail in the Oiii and Sii I might just use Ha as luminance. A couple of other options are to do a noise weighted integration of the Ha/Oiii/Sii masters (no rejection) or LinearFit Sii and Oiii to Ha then sum a third of each with PixelMath. Then I'll do decon, gradient removal, noise reduction, stretch, etc.

Colour: First gradient removal, if needed, then I remove most or all of the stars from the linear masters (you can skip this step but it solves the problem of magenta halos very neatly.) Then I'll do a noise reduction on each filter (typically very heavy on Oiii and Sii if they are noisier.) For the colour combine I typically LinearFit the Oiii and Sii to the Ha and then do a simple 1:1:1 colour combine. Next is a stretch and heavy chrominance noise reduction with ACDNR. Then I'll adjust the colour, often quite dramatically, mostly using the ColorMask script and Curves (R, G, B, a*, b*, Hue are all useful) and also boost the saturation if needed (controlled with a clipped Lum mask.)

Finally combine Lum and colour and do any final processing... Voila!

The ColorMask script is here: http://pixinsight.com/forum/index.php?topic=7751.0
I may be slightly biased but I reckon it is an indispensable tool for NB colour processing in PI :)

Cheers,
Rick.

[niharika]

Rick, can you please elaborate on your star removal technique in pixinsight.

Regards
Raki

[Ryderscope]

Wow! Thanks for the detailed response Rick. Clearly I will need to find a quite room and digest this information. Will return later.
:thanx:

[RickS]

Hi Raki. The trickiest part is generating a good star mask which takes a bit of parameter tweaking and maybe combination of a few different masks with the PixelMath max() function. Once the mask is applied then a few applications of MultiscaleLinearTransform to delete the small scale layers followed by a few applications of MorphologicalTransformation Erosion will clean up most of the stars. Sometimes it is necessary to CloneStamp out remnants of larger stars. TGVInpaint seems like it might be helpful as well but so far I haven't been able to make it behave as I'd like.

Ha usually comes up nicely but Oiii and Sii tend to look a bit more ragged. In the end it doesn't matter a lot if you're willing to tidy up a little with CloneStamp.



No probs, Rodney. Happy to answer any questions. I'm not sure I have the one true way to do NB but so far it works for me :lol:

Cheers,
Rick.