Coping with off colour stars

[roundycat]

Hi all,

I am taking the liberty of posting this recently worked out method of getting rid of cyan stars when doing HaGB colour. If this is in the wrong place or out of place someone please delete it.

http://www.dens-astropics.org.uk/page%2017.htm

Dennis

[MartinB]

Well I've had a good look Dennis. I need a bit of practice sorting out the stars. The main problem I encountered was over saturating the red stars, I found it helpful to add in a bit of enlargement with the maximum filter at the actual stretching stage before fine tuning with the combined image. Also very interesting to see how you do very similar things in a completely different way to me!

When you mention blending the red with the Ha by pulling the Ha to the top and using lighten shouldn't it be the red layerl with the chunkier stars that is at the top?

The image I've come up with is very rough and ready and could be improved with a bit of practice, I'd even forgotten to properly align the new red combine I did!

If I was doing this image again I think I would process each channel seperately. The Ha could have handled a greater stretch than the G and B channels. Some of the Ha detail was lost because I didn't stretch the Ha red channel enough. You can see this with the comparison of an HaHaRGB I have done previously and the HaRGB using your technique. The more I think about it the more I think you may be able to boost the stars in the Ha channel using the white point adjustment and maximum filter. I guess you've done this and found it doesn't always work.

Not using an Ha as a luminence obviously helps the red saturation but it is harder to bring out the detail.

So here they are, the first is my original image and the second is a crude version of the Dennis star enhancement technique

(click to enlarge)

(click to enlarge)

[centroid]

Well it seems to have worked Martin.

I haven't read Dennis's procedure, as I don't Ha as yet. But the effect of bringing out the missing background stars, looks exactly like the application of unsharp masking on a median filtered (or similar) image.

Dave

[MartinB]

Actually Dave the 2 images I posted might be misleading. The first image doesn't actually need the star enhancement technique since it has full RGB data as well as Ha. The second image is using a minimal amount of red data to enhance the star field and avoiding the cyan colouration you get using just Ha alone as the red channel. The stars are more forward in the bottom image purely as a result of processing inconsistencies between the 2 images. I don't think you would do away with the red channel if you had the data, but this shows that you can save yourself a fair bit of time by not having to bother collecting much red. I only used 3x150 secs of red binned x2 and I think the star colour looks ok

[centroid]

Ah, ok that explains what I'm seeing then, and yes, the star colours do look very good.

Dave

[roundycat]

Martin, thanks for the wake up call! I checked with a PS book and did some simple tests and the layer order for Red over Ha or Ha over Red is unimportant. The simple fact is that with Lighten blend mode the pixels in the blend layer or the underlying layer will be lightened to the brightest value of the two.

Back to the web page now to amend that bit of PS ignorance.

I have avoided max and min filters as much as possible because wherever you have fine detail in a nebula it will be modified by the filter unless you filter stars on a separate layer. I have done this with high settings just to have a look and didn't like what I saw.

The whole idea of the second stage, ie, High Pass filtering of the nebula and subsequent covering up with the 'soft' star layer, is to enhance nebula detail, especially dust lanes and so forth. This is why I use a red layer with not much nebula detail, so it doesn't mask off any detail that I may have managed to photograph. Masking the nebula from a well exposed red layer will not work because the mask will never be good enough to include the nebula and exclude all the stars in the nebula. If you miss a lot of the stars in the nebula in this way you end up with off colour stars just over the nebula area.

Dennis

[MartinB]

The whole idea of the second stage, ie, High Pass filtering of the nebula and subsequent covering up with the 'soft' star layer, is to enhance nebula detail, especially dust lanes and so forth. This is why I use a red layer with not much nebula detail, so it doesn't mask off any detail that I may have managed to photograph. Masking the nebula from a well exposed red layer will not work because the mask will never be good enough to include the nebula and exclude all the stars in the nebula. If you miss a lot of the stars in the nebula in this way you end up with off colour stars just over the nebula area.

Dennis

Yes, I agree. I have played around using star layer masks on top of nebula layer masks but you can nearly always see the joins. Often in the area of nebulosity the star colour isn't that pronounced and you can get away with it.

Anyway, the next time I do an Ha neb I will give the technique another go, like all processing techniques it takes a bit of practice to get it right and every image will be different.

[roundycat]

just a further note Martin to your mention of the possibility of 'extra' stretching of the Ha data. I found the bright parts of the nebula (especially the Rosette) tend to saturate and posterise when doing this. It's that age old problem of wanting to brighten part of an image without touching any other part of it when they are both the same brightness.

Dennis

[MartinB]

I guessed you'd had a problem! So, a smidgeon for red it is then.