Here's why you get magenta stars in the Hubble palette

[RickS]

I was just playing around with my first three narrow band subs of the night and combined the unregistered Ha, OIII and SII subs into a colour image. Because they are unregistered you can see the red (SII), green (Ha) and blue (OIII) stars in little groups rather than superimposed on top of each other.

The Ha signal is the brightest, followed by OIII and SII which is typical for most objects. That means that the SII data has to be stretched a lot more than OIII which is stretched more than Ha to get an even colour combination. This bloats the red and blue components of the stars more than the green which is shown nicely in the attached crop. Register the images and voila... magenta star halos.

I obviously have some spare time on my hands while ACP does the hard work of driving my scope. Hope this was of interest to someone :)

Cheers,
Rick.

[PRejto]

Yes, and thanks!

Now if only there was an easy solution!

Peter

[RickS]

Peter,

I have two solutions but neither is trivially simple:

Use deconvolution to shrink the SII and OII stars to match the Ha stars better
Take some short RGB subs and replace the NB stars with nicely coloured RGB stars

I like the second option but it is a fair amount of work.

Cheers,
Rick.

[Tandum]

Rick, have a go at tone mapping.
http://astroanarchy.blogspot.com.au/2009/11/power-of-tone-mapping.html

[RickS]

I've been meaning to give it a try, Robin, but I haven't found an easy way to do clean star removal in PI yet. I might have to spend some more time on it...

Cheers,
Rick.

[pluto]

I process each channel separately in Startools and do a cheeky reduce on the star size in the SII image if it's really different, depends on the target though I've found :D

[Spookyer]

Hi Rick, I have just been working on a narrowband image. I have got the colour of the neb how I like it but the stars are a blue/green colour. When I use colour correction it turns the stars white which is fine but it then turns the neb a monochromatic red as well. How do I get around this?

Brett

[RickS]

G'day Brett,

You could use a star mask to protect the neb while you tweak the stars. If I don't add RGB stars I often mask the stars and desaturate them to while.

Cheers,
Rick.

[Spookyer]

I tried a star mask but couldn't get it to work. I could only get the big stars and there are a lot of little ones in the shot that were not being selected even with playing with the settings. I am not sure how to just select the nebulosity and avoid the stars?

The online tutes for pixinsight are good but when your image doesn't match the sample and you get undesirable results from following the process it can be a bit hard to know what to do.

[RickS]

Making a good star mask can take some effort and tweaking of parameters.

One useful trick is to copy your image and hit it with a big HDRMultiscaleTransform (maybe 4, 5 or 6 layers and 4-6 iterations) and then build the star mask from that image.

Play with the scale, structure growth and smoothness. If you're working with a linear image then the midtones value is important (probably a good idea to stretch a copy and then HDRMT as described above).

Sometimes I build a star mask and then do fine tuning with MorphologicalTransformation to grow or erode the mask.

Unfortunately, every image is different and it's not possible to give recipes that work in all cases. You have to play around and experiment.

I'm happy to have a go at building a mask if you want to put a FITS file somewhere I can download it.

Cheers,
Rick.