LP filter as luminance with DSLR/OSC

[smokey688i]

First off, a huge thank you to Ivo for a wonderful program...I have thoroughly enjoyed getting some great results in this growing hobby of mine in large part due to your tireless work on StarTools.

I image mostly from my suburban backyard red/orange zone, and I've seen Ivo mention the technique of using a light pollution filter while taking luminance subs with a DSLR/OSC, then no filter to capture correct color, and lastly combine these in processing. I've searched around for some results of this technique, and haven't come up with anything. Once the weather clears here I'll give it a go myself, but does anyone have any feedback/results on this approach they wouldn't mind sharing?

[smokey688i]

Step 1 completed, 19 full-color frames in this one of M31, thought I'd share this as a step by step to see what improvement can be gained. Plan is to add ~20 subs through a dgm npb filter as luminance and combine in the LRGB module.

https://flic.kr/p/Baz1im

[admin]

A nice image to begin with - looking forward to the non-LP/LP combined version!
(remember you can go deeper now with the LP filter in place)

[KD0NPT-Astro]

Smokey, I use the Hutech IDAS LP2 for my LRGB imaging in a hot zone. It is one of the better color balanced ones out there at this time. They also make a clip-on version for DSLRs. I usually do mostly 10 and 15 minute subs with it.

http://www.sciencecenter.net/hutech/prices/filters.htm

There are images on my web site, still in it's youth, taken through it. One of the four street lights nearby is 25 feet from the NW side of my dome.

http://www.kd0npt-astro.net/index.html

[smokey688i]

Shot a decent set with the filter in place last night, but I'm having some issues combining them. I tried loading the unstretched LP filtered stack directly into LRGB as luminance, and the color image as all three channels, but after stretching and wiping it looks like an over-saturated mess - the galaxy comes out very green, and the background is a noisy purple and red soup. Going to process the luminance stack separately, then load it into LRGB to see if that helps, but I have a suspicion using a nebular-type filter as the DGM NPB isn't going to give me good results for this target. Am I off track here?

[smokey688i]

Ok, better result by processing the luminance first. Still needed to use the "Cap Green to Yellow" option in the color module to keep so many of the stars from looking green...I know it's hard to critique the result without knowing the processing steps I used for each, but I don't see much improvement over the original result. Perhaps the difference will be more noticeable for a dimmer target that is dominated by more ha/oiii nebulosity, where the DGM NPB is better suited.

Version 1, no LP filter used

andromeda small.jpg (482.42 KiB) Viewed 8012 times

Version 2, with DGM NPB filter as luminance

NewLRGBComposite small.jpg (364.65 KiB) Viewed 8012 times

[admin]

Very interesting! As an alternative to LRGB combination, have you tried the following technique?

http://forum.startools.org/viewtopic.ph ... 2532#p2532

E.g. process RGB and luminance separately and combine them using the above technique in the Layer module.

[smokey688i]

Better results this time by preprocessing both color and luminance, then layering them with the Color of Fg option in the the layer module. I also took more care while processing the luminance file to focus on the detail I wanted, and heavily denoising the color file. I'm pretty happy with this one...but side by side with the original not a significant improvement.

Version 3

2 new color only 2138x1408_filteredx2 small.jpg (240.81 KiB) Viewed 7974 times

I think part of the problem is my LP filter stack with the DGM NPB was only 5 min subs (longer was blowing out the blue channel), and just by looking at the initial autodev stretch the visible detail was not any better than with the original color stack following the initial stretch - so I wasn't gaining much in detail with the LP filter in place. I think I'll reserve the DGM NPB for Ha/Oiii rich nebulas and give this approach another try there...

I am going try something else with Andromeda once it clears...I was limiting my original color subs to 2 mins to avoid the histogram being too far right. This was at 1600 iso. For my camera (Nikon D5100), I should not be introducing read noise if I reduce the ISO to 800, 400, or even 200 according to http://www.cloudynights.com/topic/40818 ... mal-noise/. (D7000 has the same sensor as the D5100) If I lower the ISO, then I should be able to take longer exposures, which should allow me to capture more of the fainter detail (especially when using nikonhacker's true dark current firmware mod)...As I understand it, lowering the ISO a few steps will essentially give me more tiers of light intensity per pixel that can be recorded, which will allow for the fainter details to be detected. Of course, too low an ISO will start to dump some of the photons detected, so the question will be how low is too low an ISO. Am I making sense here?

[admin]

Better results this time by preprocessing both color and luminance, then layering them with the Color of Fg option in the the layer module. I also took more care while processing the luminance file to focus on the detail I wanted, and heavily denoising the color file. I'm pretty happy with this one...but side by side with the original not a significant improvement.

The main reason to separate Luminance from Color is to achieve correct coloring, while being able to go deeper detail -wise. Depending on how bad your light pollution is and how effective your filter is in mitigating it, you may or may not see any enhancements in the detail. Being able to achieve 5 min subs is quite long as far as exposure time goes, whereas in the more severely light polluted areas that could definitely white out most of your image without LP filter.

I am going try something else with Andromeda once it clears...I was limiting my original color subs to 2 mins to avoid the histogram being too far right. This was at 1600 iso. For my camera (Nikon D5100), I should not be introducing read noise if I reduce the ISO to 800, 400, or even 200 according to http://www.cloudynights.com/topic/40818 ... mal-noise/. (D7000 has the same sensor as the D5100) If I lower the ISO, then I should be able to take longer exposures, which should allow me to capture more of the fainter detail (especially when using nikonhacker's true dark current firmware mod)...As I understand it, lowering the ISO a few steps will essentially give me more tiers of light intensity per pixel that can be recorded, which will allow for the fainter details to be detected. Of course, too low an ISO will start to dump some of the photons detected, so the question will be how low is too low an ISO. Am I making sense here?

You perfect sense! In most cameras, ISO is a post-capture gain, e.g. a simple multiplier (in digital terms) of the signal already captured (typically before it goes to the A/D converter though!); it does not make your sensor more or less sensitive. It merely manipulates the signal after capture to take advantage of the full dynamic range of the A/D converter. As such, try staying close to your model's native ISO, which indeed may be lower than 1600 ISO. So, yes, this will give you a larger dynamic range to 'fill' before saturation occurs.