Star colours in M13

[old_eyes]

I have a query about star colours in M13.

Image taken Friday 2nd July. Equinox 80ED on unguided CG5-GT with one shot colour Atik 16ic. Mount unguided so limited to 1min subs. 14 x 1 min subs with darks and bias - no flats. Assembled in AstroArt and tweaked in Photoshop Elements.

I am pretty pleased with the image, but I notice that a lot of the M13 images on SGL show lots of yellow in the stars. This ismage seems to be very blue. I have tried going back and redoing the stretch, keeping a careful eye on the histograms for the diffferent channels, but it always comes out this colour (or develops some awful overall colour cast).

Is this reasonable for M13? Is it the spectral sensitivity of the camera so that blue dominates? Or is it something I am doing in the processing?

Any advice gratefully received.

old_eyes

[roundycat]

the problem almost certainly lies in your processing. Did you do all of the post processing in PS Elements? It is very easy to stretch the data overall without keeping an eye on what is happening to the white point. Once you start to saturate the stars in any channel all hopes of getting the correct colour are out the window.

You can do this with a levels stretch if Elements does not have curves. Just use the mid point slider and move it left to about 2.0 then do it again up to three or four times. The deeper your exposure the less you will have to stretch. The further you move the mid slider in one go the more you risk accidentally clipping the white point, something you MUST NOT DO.

A couple of mild visits to Hue/Saturation (Image-Adjustments) will improve whatever colour you have.

Dennis

[Earl]

My m13 is very much the same, I lost colour which was showing in single images.

Lots to learn

[themos]

Once you start to saturate the stars in any channel all hopes of getting the correct colour are out the window.

It's worth checking that most stars are not saturated in the subs, too. Your M13 core looks like it hit the roof.

[old_eyes]

the problem almost certainly lies in your processing. Did you do all of the post processing in PS Elements? It is very easy to stretch the data overall without keeping an eye on what is happening to the white point. Once you start to saturate the stars in any channel all hopes of getting the correct colour are out the window.

You can do this with a levels stretch if Elements does not have curves. Just use the mid point slider and move it left to about 2.0 then do it again up to three or four times. The deeper your exposure the less you will have to stretch. The further you move the mid slider in one go the more you risk accidentally clipping the white point, something you MUST NOT DO.

A couple of mild visits to Hue/Saturation (Image-Adjustments) will improve whatever colour you have.

Dennis

Thanks Dennis.

Elements does have curves, but it is not as flexible as in full PS. you can't add additional control points. In some ways that helps because it forces you to use a number of relatively gentle stretches rather than trying to blast it in one go.

I thought I was focusing on bringing up the mid range without saturating in an attempt to avoid burning out the core. I'll try again - even more gently.

Is there a way from histograms that you can identify when you are clipping the white point? As I watched the histogram with each stretch step it didn't look like there was much going on at the top end.

I am guessing from your comments that you are confirming there should be more colour in the image?

old_eyes

[old_eyes]

It's worth checking that most stars are not saturated in the subs, too. Your M13 core looks like it hit the roof.

Good point themos. I am pretty sure not as I had to do a lot of stretching to see the cluster well at all. However, when I get back to my computer (on road at the moment) I will check.

I think the core brightness may be caused by my over-enthusiastic stretching of a relatively small total exposure (see my response to roundycat).

I'll have another go at the image at the weekend and see what can be done.

old_eyes

[themos]

In the histogram display it's worth checking if there's a "logarithmic" option. That way, you get to see something in the chart even if it's a thousandth of the peak.

[roundycat]

first off you should never attempt to stretch the whole thing in one go.

I am not familiar with Elements but if you cannot add extra control points (the full version allows you to use fourteen) then it is really no better than levels, just using a different display.

The problem with relying on the histogram to check for clipping, especially the white point as it often contains very little data - just a straight line along the bottom of the graph, is that you cannot really tell if there is clipping or not. remember the histogram is just a bar chart that shows you the number of pixels at a given value. If you have lots of pixels with a value of, say, 10-20 levels (in 8 bit), then showing pixels with values of 220-255 will hardly register if there are only a few. Hence the straight line along the bottom of the graph.

I would suggest taking the raw data and open it in PS. Go to Levels and set 0/2.0/255 and ok. Do it again. Do it again. Three iterations should start to show meaningful detail unless your exposure is woefully short. If you can check the max pixel level in a 16bit program to see if it exceeds 65,000. If it does you are in trouble but it is unlikely. One minute at f7 is not pushing the boundaries.

If your histogram starts to look spread out with a sizeable lump near the left centre you are getting somewhere. If not do a fourth Levels stretch.

At this stage it might be worth visiting Image-Adjustments-Hue/Saturation and increase the saturation by +20. Then repeat. As with stretching don't try to do it all at once.

I am certain that your image has colour in there somewhere. There may be something here that helps Astrophotography

Dennis

[Shibby]

Hi old_eyes,

I hope you don't mind but I had a quick tweak in Photoshop, just to show that you have actually captured the colour differences between the stars, so with some careful processing as advised by the other posters you should definitely be able to improve the image.

Lewis

[old_eyes]

Hi old_eyes,

I hope you don't mind but I had a quick tweak in Photoshop, just to show that you have actually captured the colour differences between the stars, so with some careful processing as advised by the other posters you should definitely be able to improve the image.

Lewis

Lewis,

Not at all. Thanks for taking the trouble. I plan to follow roundycat's suggestions, but any chance of a clue as to HOW you tweaked the image? It would be much appreciated.

old_eyes

[old_eyes]

f

I would suggest taking the raw data and open it in PS. Go to Levels and set 0/2.0/255 and ok. Do it again. Do it again. Three iterations should start to show meaningful detail unless your exposure is woefully short. If you can check the max pixel level in a 16bit program to see if it exceeds 65,000. If it does you are in trouble but it is unlikely. One minute at f7 is not pushing the boundaries.

If your histogram starts to look spread out with a sizeable lump near the left centre you are getting somewhere. If not do a fourth Levels stretch.

Will do Dennis

old_eyes

[Shibby]

Hi old_eyes,

I aligned the channels then boosted the saturation. A sneakily easy way to align the channels (if you use Photoshop) is to use the "Auto Colour" function, tho this can have other undesired effects so it'd be best to play with curves/levels.

I think I also reduced the contrast a little, which seemed to improve resolvability of the stars near the core.