Help with Swan nebula LRGB processing in PixInsight

[ThomasM]

Hi,

I'm working on an image of the Swan nebula (M17) taken with a robotic telescope using PixInsight. The exposure times are 6x60 secs (Clear) and 6x120 secs in each of RGB, all unbinned. I've applied the following steps so far:

1. Registered and stacked the frames

2. Applied a careful Multiscale Linear sharpening to the L channel

3. LRGB combine with equal weight to each colour channel

4. Histogram transformation using the Screen Transfer Function. I kept the colour channels linked then increased the blue to achieve a more pleasing colour to the nebula.

Unfortunately I find red halos around the bright stars which are enhanced when I use the curves tool. The end result is posted below. I'm sure there must be a good explanation for the halos? I checked the R-band image and the stars look fine apart from the brighter stars are slightly over-exposed.

All advice appreciated!

Thomas

[MattJenko]

The halos might well come from overstretching in the R channel because of when you did the combination of the channels. I would get all the various channels processed a bit first before combining, so use DBE on them all to remove gradients, and then the LinearFit tool to match up the RGB. Once you add them together using ChannelCOmbination, have a look at the halos and you can either manage them as an RGB image, or use MorphologicalTransformation on the specific channel in question to reduce the bloat before combining the channels. 

Once you have the RGB, do some ColourCalibration before a stretch and look at MaskedStretch for getting tighter stars. You can blend these into a normal HistogramStretch for the best of both worlds. De-noise and tidy up.

For the L, it is similar, but this is where you want to do the sharpening, once you have removed gradients and gotten it nicely ready. You can then apply the L to the RGB via the LRGB combination tool.

[andrewluck]

It looks like either your focus or seeing conditions were not as good for the red images. If you measure FWHM using the FWHMEccentricity script are the numbers comparable across the colours?

Try doing an RGB combine without the luminance and see what it looks like. I avoid any processing of the RGB frames before combination as processing individual colours makes colour balance very difficult later.

Pixinsight has a star halo reduction script that may help here although I've never got the hang of it.

Andrew

[ThomasM]

Matt, Andrew, many thanks for your suggestions.

Andrew, the median FWHM measured from each RGB image is 5.30px (red), 5.33px (green), 5.16px (blue). For comparison it's 4.84px for the luminance (after sharpening).

I've combined the unprocessed RGB images then applied the Auto Stretch STF (unlinked this time). Finally I applied the curves as used in the previous image.

The colour balance isn't quite right but the stars look better.

[andrewluck]

Stars look a lot better. Next step will be Background Neutralisation followed by Colour Calibration and that should sort out the colours.

Andrew