Strange shades of colour on my DSLR images, and CCDs

[downingsimon]

Hello everyone,


it is great to be able to come on the forum for my first time. Thanks.


I am having some strange colouration issues on my DSS stacked lights of certain objects. Just so you know, I use a Skywatcher 8" Quattro and a Canon 650D modified camera, a Hutech LSP filter and Baader coma corrector, all on a NEQ6 pro mount, and an Orion autoguider using PHD. I live in what I would call a pretty light polluted area.


Essentially what is happening is that I will take the stacked image into Pixinsight, perform DBE on it and create a workable file. But as soon as I do a STF stretch on it after this, I can see this weird colouration in the background. Also, these colourations still appear, when I start to do normal processing, [after finishing with DBE and having reset the screen functions]. If I do not do DBE however, it appears fine when stretched.


So, after DBE and resetting, when I up the histogram on it, midtones only, straight away the strange shades of coluration begin to appear in the background. I believe that the 0.25 figure under 'mean' on the statistics is the limit for an image, so as to keep one from overdoing the background, but even at half of this level, I can still see the shades coming through. The problem with dropping the midtones that far, it means that I am not able to treat the object itself with as much midtone as it really needs. I have tried masking it and everything, it seems.


And here is the odd thing about it. This has only occurred really, on globular star cluster images like M3, or on images with relatively small objects in the frame. It seems as though when there is lots of background and not so many stars, it mainly occurs - not so much with objects with lots of nebulosity or a large object like M31. So, my Rosette nebula was ok, comparatively. And the only way to get rid of the strange colouration is to pretty much clip the black point, in order to rid myself of the visual defects. By the way, how does one determine when a black point is clipped or not clipped?! I still really struggle with this....   


I was hoping to attach a M3 image here, which is a classic case in point, but I cannot do so due to file size restrictions. Besides one needs a tiff to be honest, to see the problem issue really appear because the jpeg compression loses a lot of it. As that is so I have sent you a screen shot so you can see it for yourself. Maybe I am doing DBE wrong, but that said all the RGB 'peaks' under the histogram all line up during processing, and what I do with DBE seems to work with other objects like the aforementioned M31.


Incidentally, I do not use dithering on my PHD Orion autoguider settings - would that help? If so, what would be a good setting; default on it is 1, but the range is up to 100. By the way I know there are a few dust bunnies on this image, but they make no difference to the issue anyway. I have since cleaned up the sensor. Maybe I was simply using too few flats and biases with these images anyway; which I have improved upon since with more recent objects?


I guess, whilst I am at it, I would like to ask whether a CCD camera would solve these issues, and because I do not want to lose any FOV, I will need a large chip. Which camera would someone recommend, a Starlight Xpress SX-814 Colour or SX 25c? Something cheaper would do!!!   


I know I have asked a lot here, and I really apologise, but I thank you so much for your time in advance.


Simon
UK

[Jason Tackett]

Hey Simon,

I've seen something similar to this in my images, but not quite as dramatic. I also live in a light polluted area and I have found that some of my images with shorter sub-exposure lengths and high ISO will have very agreeable SNR after integration, but once I subtract the background with DBE, I am left with a much noisier image. This makes sense because the bulk of what I captured was light pollution and the little bit that remained was my precious signal, so the SNR of my target is still poor. In cases where I have very little signal buried in a mountain of light pollution I'll get some coloration after DBE when I use a whole lot of sample points since my images have color casts from my light pollution filter among strong light pollution gradients (this is getting me depressed thinking about it).

I've found that when I use a lot of sample points in DBE, I'll get complicated patterns in the modeled background kind of like yours. The remedy is, if possible, to use a very small number of sample points with DBE (maybe 4-5 / row) so the modeled background isn't trying to follow a complicated gradient since most light pollution gradients are gradual across an image. Also, check each of your DBE sample points to make sure that none contain stars since this will also cause artifacts in the modeled background. I see you are using a tolerance of 2 in the modeled parameters which means more pixels in each sample are being allowed into the model calculation so the impact of a star in one of the samples is greater than if the tolerance were lower.

I'm not sure if this is what is going on with your image (lots of light pollution & little signal), but hopefully there is some useful information here Oh, yes, dithering is highly recommended for the best possible outlier rejection and SNR improvement. I use the same autoguider with PHD, and I think I keep that parameter at the default, but I don't think lack of dithering is related to what you are seeing here. Also, I'm not sure if there is some incompatibility with the DSS output file and what PixInsight expects. There have been some threads on that topic I think.

Best,
Jason

[downingsimon]

Thank you Jason!

In an odd way it is good to hear that someone else has had an issue like mine. I actually tried reducing the selection points on the image as you suggested and it worked to a degree; it was certainly better than it was before! So I am happy about that and grateful to you. I think I will just have to accept that these sorts of images are hard for me with regard to the SN ratios. Maybe as the year ticks by I will get another shot at them and see how I do then.

Take care anyway. Maybe we will be in touch again some time?

Thanks again.

Simon

PS. I did read somewhere that a dithering setting of 12 is good, and I think PHD has it at just 1. So I may give the 12 a go soon, and see what happens. If this wretched weather clears up....