Colour at the end of Wipe

[almcl]

Been greatly enjoying the new features of the latest StarTools development, but one aspect of the new facilities which I have not quite understood is the use of the colour button at the end of the wipe module.

My work flow is Open, Autodev, Bin, Crop, Wipe.

At the end of the wipe module if the colour button hasn't been invoked, a message pops up saying to use it.

What I haven't worked out yet is why?

The result is usually a hideous blue/green mess (see attached composite image of the Wipe results from my last four targets) but what should I do about this at this point? I know that by using a combination of ST's excellent tools &c I will eventually end up with a much cleaner result (I am having a bit of difficulty with colour at the moment but that's for another thread) but what should I be doing/deducing/changing in the Wipe module as a result of looking at the colour?

composite wipe1.jpg (166.5 KiB) Viewed 6001 times

[admin]

When StarTools processes your luminance and color separately, Wipe will actually processes and clean two datasets at the same time;
One is your luminance dataset (mono), the other is your coloring. Hence Wipe requests you make sure that both datasets are in order before proceeding.
If you find either one of them is showing bad results, you will want to make sure to figure out why that is and correct this. E.g. correct for dark anomalies, or remove stacking artifacts.
If you do not fix Wipe's color result, this will come back to haunt you when you reach the Color module.

E.g. of the four images you attached, only the image on the bottom right corner is should get the OK. The others will need further TLC.

Does that help?

[almcl]

Thank you, Ivo.

So, taking the two left images (and my apologies for not cropping out the amp glow, which in my haste to create the image I neglected to do), what should I be doing about the swathes of unusual colour? In the usual course of events the subject galaxy would be isolated/HDR'd/ contrasted &c and the background darkened, but is there anything that can be done prior to Wipe or at the Wipe stage to mitigate the problem?>

[admin]

Thank you, Ivo.

So, taking the two left images (and my apologies for not cropping out the amp glow, which in my haste to create the image I neglected to do), what should I be doing about the swathes of unusual colour? In the usual course of events the subject galaxy would be isolated/HDR'd/ contrasted &c and the background darkened, but is there anything that can be done prior to Wipe or at the Wipe stage to mitigate the problem?>

The very strong and very local gradients of those images to the left are highly unusual. I'd love to see the dataset(s) that harbor (or cause) them...

[almcl]

That's a most kind offer, Ivo.

You have a pm with a dropbox link to the stacks (I hope!)

Let me know if additional data would help?

Clear skies

Al

[admin]

Thanks Al - it's very helpful.

I had a look at the stacks and can indeed easily reproduce this. Something is very wrong with - as a guess - either the bias. darks or the flat correction. The signal is extremely uneven and also extremely noisy. I can see lots of correlated "noise" (e.g. clumps of pixels of roughly the same color). Lots of hot pixels too. What ISO did you use?

What do your stacks look like with just the flats applied? How about without the flats?

[almcl]

Thanks, Ivo.

The master bias (i'll put it and the master flat in the dropbox folder) is one I've been using for a while. I did change the flat stacking from sigma clip to average accidentally a while back, now changed back. I don't normally use darks, but was trying to reduce the amp glow a little. Everything in this set was shot at ISO 1600

I'll re-stack without the various calibration files and post... something I normally do while imaging and down loading. This usually results in heavy vignetting and some dust bunnies

[almcl]

Have tried stacking with and without bias and /or flats and also with flats, bias and darks for both the upper and lower left images.

All uploaded to the dropbox link sent earlier

[admin]

Have tried stacking with and without bias and /or flats and also with flats, bias and darks for both the upper and lower left images.

All uploaded to the dropbox link sent earlier

Many thanks for stacking and uploading Al!
The strange thing is that the stack with just the lights is actually the best one (evaluating LL stack) and looks the cleanest (except for dust donust and vignetting). The one with lights+bias frames actually has hot pixels and clusters introduced vs the stack with just the lights (did you maybe swap the names?). The one with just flats, similarly, shows a big light patch introduced (but does gets rid of vignetting and dust). Very strange.

The flats-only stack (with the introduced signal) is actually a really good laboratory demonstration of Wipe's ability to discern between gradient (e.g. celestial gradient, vignetting, uneven lighting) and faster undulating stuff (almost always nebulosity). It is refusing to Wipe away the introduced signal as it is correctly detecting it is undulating too fast in-and-out and therefore must be detail (you can increase Aggressiveness to override this apprehensiveness of course ). This ability is actually a key difference between sample-based implementations as found in PI or APP; Wipe is much more more careful with removing nebulosity in properly calibrated data.

I would redo the flats first and dispense with the bias for now. Let me know how you go!

[almcl]

Wonderful advice, many thanks, Ivo.

I was suspicious of the flats when I noticed the histogram was at about 25%, so I'll try recreating them, hoping the dust hasn't moved too much!

I'll also redo the bias files. If the temperature here stays low I might even be able to recreate some darks - my home made, external mount peltier cooler struggles to get much below ambient.

I did rename the files, but *think* I got the naming correct.

I'll start a separate thread on the on the colour of the galaxy NGC 2903.