M51 - help on colour balance, sky colour, general processing!

[graemekite]

Hi - first of all, a note of thanks, I've been lurking on this forum for a while, and it's a great source of help and information (and inspiration!).  Now, for some specific help.  I've been imaging a while, starting with a camera pointed through the eyepiece, moving on to a DSLR camera body at prime focus, and now, a cooled CCD monochrome Atik 383L plus.  I started with monochrome, but have now added LRGB and narrowband filters.  This is my first serious attempt with LRGB, and I've spent WAY too long trying to figure out the processing.

I don't think this is a bad image at all, but there are specifics that have given me grief. 

- it looks over processed

- the colour balance doesn't look right, (esecially when you look at the colours of some of the stars)

- I had real trouble with noise in the sky background, particularly evident when overlaying the L layer onto the RGB, which I've rectified by bringing the black point up to the point of clipping the histogram on the left - leaving me with a jet black slightly unnatural sky.

- (I know the guiding is a bit out - don't get me started!).

The basics are of the image are:

- Atik 383L plus, on a Skywatcher 200, f5.  Guided with PHD (I won't do the detail here).

- 31 x 600 sec of Luminance, in 600 sec subs, x1 bin

- 15 x 200 sec Red subs at x3 bin

- 15 x 240 sec Green subs at x3 bin

- 15 x 320 sec Blue subs at x3 bin

- Stacked as 4 separate stacks in DeepSkyStacker with darks, flats, and dark flats applied.

- No processing applied to the RGB stacks in DSS.  Minimal stretch and curve in DSS on the L stack.

- post processing in GIMP (2.9.1 development version with 16 bit colour support)

- RGB opened as layers in GIMP.  Layers coloured, aligned, and merged. RGB histogram peaks aligned.  Black point moved up to left of histogram.

- Used curves on RGB to bring up detail (using the master channel).

- rescaled the RGB image to match the L image.

- L processed in GIMP with a small amount of levels and curves.

- Paste the L image to the RGB image.  Align.  Set L layer to 'value', and voila!

- a bit more tweaking on levels and curves in the LRGB to reduce background noise.

A couple of processing points I'm not sure on. 

- my RGB histogram peaks were not aligned, despite my attempt to get this right with the different sub exposure lengths.  I tried to align using levels in GIMP.  Is this correct?

- The R histogram was much fatter than my G, which was fatter than my B.  Do I need to correct, and if so, how?

- Do I need to do further adjustments with the colour balance tool, or hue/saturation tool?

- any hints for application of curves?  I found the best way was to anchor the curve to the right of the histogram, and then bring it up about half way along the curve.  If I don't anchor it, then the sky background gets brought up as well.  If the sky background comes up too much I start to get a colourful, brownish, mess in the background.

That'll do for now, all comments, criticism, hopefully help, appreciated!  Cropped JPG attached.

[Ewan]

Hi,

I cant help with Gimp as i have no knowledge of it.

First off don't use DSS for anything other than stacking, leave the processing to the right programs if possible.

I have only ever tried 2x2 binned but when possible i won't bin, i appreciate imaging time is precious though.

If you look at an image of say The Rosette, the histogram (red channel) will be 'fatter' than the other channels because it is predominantly red in nature, so maybe take a look on the web at other images of the same target to get an idea of the 'standard' colour but in the end it's upto you.

As i use Photoshop i used t+he HLVG on your image  got this.

It's very good at sorting the 'greens' out.

Someone with more knowledge can probably help you out more though.

All the best.

[FunkyKoval35]

Hi. You can try this tutorial. It is using PS, but may help you a lot.

[graemekite]

thanks both, I learnt some useful stuff on setting the gray point, and the difference between linear and non-linear stretching.  I also came to the conclusion that my RGB subs are way too noisy for some reason, and that is causing me issues.  I will have to go back and reshoot them at some point I think before spending too much more time on this one.

One question - should a 'standard' workflow always look to align the peaks of the RGB histograms?  I haven't been able to work out how the different contributions from RGB in a subject affect the relative positions of the peaks (left to right), the relative breadth of the peaks, and the relative heights. 

thanks!

[FunkyKoval35]

Answering your question: you need to do color balance. In PS you need to use color sample tool(new panel will show up) and click on dark(background) and light(star) region of your image. After doing this you will see two sets of three values(color levels for each channel - RGB). Now, you need to set each channel value on the same level - background to min, star to max value. So if you have levels, for star R=230, G=215, B=222 you need set all three values to 230. For background you need do the same thing, so if you have R=50, G=43, B=60 then you set all three values to 43. This should align the peaks of RGB histogram for your picture.

P.S. Sorry for my English

[ollypenrice]

Bin 3 is pushing it, I'd have thought, and in a fastish scope isn't going to be necessary. You risk hitting saturation on the stellar cores, making star colour processing difficult.

I don't see why your backlground should be noisy since you have the data. Do you have a problem with LP? Using black clipping to crop it out damages all your precious faint data so there must be a better way.

When stretching, keep an eye on the all channel view of the three histograms and use the black point slider (without clipping the pedestal!) to keep them aligned in the top left of the peak with each iteration.

Olly

[graemekite]

thanks -  I think that answers the colour balance question.  There seems to be a few ways to do it - one of the links above describes using the gray point tool to set the gray point by clicking in the sky background, and that worked pretty well on another image I'm working on.

I will stop binning x3 for colour and switch to x2. 

I do have some LP, although the subject was pretty high so it shouldn't really have been a problem.  I think I will re-shoot the RGB at some point, but these short summer nights are not much help and M51 is starting to dip a bit in the west now.  Another time maybe!

[opticalpath]

With the exposure times you used, at 3x3 binning I think it's likely that a lot of the stars will be saturating, so you're losing some colour information. At this focal length, 3x3 star images will also be rather small and blocky with sharp gradations of brightness between adjacent pixels, so you get hot spots in each colour rather than a smooth gradation from centre to edge of each star image. That can mess up the star colour when you combine RGB channels.  Sometimes slightly blurring each colour channel can help a bit but binning 2x2 would be a better choice. 

Did you calibrate your RGB exposure times by imaging a G2 (Sun-like) star?  If not, you'll probably find it helpful to do that (as described here for example:  http://www.kellysky.net/White%20Balancing%20RGB%20Filters.pdf)

You're dealing with two separate colour balancing issues:  1) the imbalance caused by light pollution, which often adds red and green to the sky background, and 2) imbalance caused by different sensitivities to R,G and B of your camera + filters.  Whilst it's quite possible to adjust unbalanced R/G/B images, getting the correct relative R,G & B exposures to begin with makes life easier and reduces interaction between the two effects as you stretch the colour channels.

Adrian

[graemekite]

***

Did you calibrate your RGB exposure times by imaging a G2 (Sun-like) star?

***

Yes - I did attempt to do this, although obviously not very successfully! I say that - I assume my histogram peaks should be aligned if I've done G2 calibration properly?

thanks.