Hi folks,

you see me a bit nervous as I fear to embarrass myself .

I had asked you to submit some RGB images of yours some days ago. This was to test some findings I made playing around with the results of my first steps in narrowband imaging with that 12 nm Omegon filter pointed to the Pacman Nebula. In the final Ha-RGB of the Packman I found again a color variation that acompanies me in all images of red emission nebulas I have made so far with the K5 and the K3ii: what is red in the images of the astromod DSLR guys varies in tones from blueish magenta to redish magenta in my images. These sometimes ugly colors annoyed me most of the time and I shifted them a bit to the red side in the past to come near to the color tones the astrocam guys achieve.

For fun I now isolated the extremes of these color variance in the Pacman Ha-RGB and set up a "filter" in Photoshop. Having tested this "filter" (I will call it SNB filter for "software narrowband" in the following) on many of my older captures and having compared it to the bicolor images in the internet I am quite sure that it is able to separate OIII emission areas from H-alpha areas and illustrate them in the popular bicolor narrowband colors. I hesitated until today to disclose this hypothesis but after I have applied it to the images you sent me and found the same principle I felt I have to do this. But before I start to discuss it in the big cruel astro world I decided to discuss it with you friendly guys here in our familiar group (and also because all was made with a Pentax cam!).

This is what came out with a click of the mouse without much need to fiddle about:


Please keep in mind that the underlying RGB colors in the image above were aquired from a tiny, blured images I had made in the past so the color transitions are very undefined......

Having that filter which in fact is no more than a correction layer that can be switched on and off I am able to apply it to all images I like to without changes. It works on nearly all of my emission images - may they be a first prestretches of linear files or final images. On a prestretched Astro Pixel Processor stack the effect is a follows - out of the box:


Please note that in the gif above nothing more than stack -> background calibration -> single stretch -> apply SNB filter was made.

With half an hour of processing in Photoshop something like this comes out:
More resolution:

Some more samples of "raw" images and older, processed images where I just applied the SNB without doing much more:
NGC 6888 RGB (processed):


SNB applied:


Original RGB (processed that time - from my early K-5 days):


SNB Filter applied:


This is the Great Wall in Cygnus from a few weeks ago. Just stack, balance, prestretch, SNB filter:


This is the Veil Nebula complex (just some curves):



And finally the Western Veil alone (RGB):


...and SNB filter applied:


My theory is that the more bluesish magenta in my original RGB data is a mixture of OIII and RGB with OIII dominance and the redish magenta is the result of the opposite mixture. Another theory is that the reason why we have not seen this kind of "single shot OSC narrowband images" before maybe is that most of the "serious" astroimagers shoot with astromodified cameras. There the H-alpha may be so dominant that the color variations melt together in red tones.

Conclusion: unmodified one shot color cameras can generate bicolor narrowband images without physical filters - IN ONE SHOT!

Puh- now it's out there! I would like to open up a discussion with you on that sentence above. What do you guys say?

Cheers
Pete


P.S. Again thank you for the heap of images you sent me! I saw the effect in nearly all of them but the colors mostly were a bit different. So the "filter would have to be adjusted to fit for those colors, The toal difference should come from different preprocessing and stacking software and data treatment.

P.P.S. Strange that I find out something new for my stock Pentax K3ii the a day after my new ASI1600mmpro was ordered. Now it is here and had its "first light" last night (esp. to fewayne: later more..... in your corresponding thread).

THat's pretty cool! Do you think you will release your filter setup?

Yeah, I think so. It is a pretty simple trick, just an adjustment layer that can be placed above any RGB image layer. But before I do that I think there must be a critic discussion about it - mainly because I may be "biased" .

I compared my image results with those narrowband bicolor pictures availlable in the internet and the distribution of the differnt emissions is very comparable up to identic. But as we say here - maybe "the wish is the father of my thoughts"

I wonder if the Ha portion can be tweaked to toggle between red like it normally is and then the gold color that many bicolor ones show. The Oiii does fall into blue already.

Yes it can. The „filter“ has two parts that can be switched on and off individually. One is for OIII an one for Ha. If the latter is switched off or with lower opacity the result is red/magenta and blue. The red can be finetuned then.

....a red/blue version could look like that:




Of course the colors should/must be fine tuned. I like the blue more faint and brighter, others want it to smash. The same with the red. The above image is just the filter applied on a more or less unprocessed image (just standard auto DDP stretch in APP). The best effect comes when the RGB stars are taken back or even are eliminated, then stretch, noise reduction etc. After apply the SNB layer. Finally blend in the original RGB stars on top with original color.

Don't have the background to comment with any authority. Looks like pretty interesting stuff, though -- I would love to hear how the physics work here. I started to think that it's not that complicated but when you start factoring in the spectral response of the sensor, then put the Bayer matrix in front of it, then demosaic in RAW processing, and use that to drive RGB channels, there's many a slip 'twixt the cup and the lip, as the old proverb says.

Maybe I disappoint you because I have not thought about the physics. I did a purely empirical approach. I postulated that the blueish magenta represents OIII (cyan-blue in SHO-palette) and the redish magenta represents Ha (golden in SHO-palette). Next I isolated the two colors and tinkered a color correction layer for each of the two that led to the aimed target colors - et voila!


These are what I call the origin colors. The color on the right side represents OIII regions and the left one Ha-regions in my untouched RGB images.

Sorry, forgot the target colors. These colors result when I apply the correction layer to the origin colors above:

Ohh wow, I like both the red/blue and the yellowish blue versions depending on the target. That's amazing! I think it would be dependent on the target as to what one would be best.

Pete, as always your posts are a real pleasure.



Perhaps we can dig a little deeper here.

Are you familiar with r-Spec / diffraction gratings or the similar? Normally this would be used on Mono camera.

This will only work if you have the hardware at hand (can't do this when the photos are taken by others.)

I would propose the following way to get a better idea what is actually being picked up by each of the RGB channels by the camera.
1. Use a broad-band light source and run a spectrum of each of the RGB channels in the camera and generate spectrum graphs.
2. Use a mono camera (with no IR or UV blocking) on a broad-band light source with the narrow band filter used and generate spectrum graphs.

These graphs will give you very important data about:
1A. How much overlap there is in spectrum between R,G, and B and channels. This is an important consideration when doing channel subtraction when the target frequency is in an overlap zone.
2A How much the narrowband filter passes / blocks out of the target band. This is a biggie because many narrowband filters will not block infrared and uv, or will not have complete blocking at all visual frequencies. This becomes more significant as the target bandgap gets more narrow or has non-ideal transmittance at the target frequency - that is, when the total illuminance in the target band is decreased. - If there is a lot of 'leakage' in the Infrared or visual non-target frequencies, it can produce unexpected (and potentially misleading) results.


Thanks again for your hard work and beautiful images!

Just to say it......I disclosed a .psd file with the SNB filter and a description here a few minutes ago: Narrowband colors w/o hardware filters for stock DSLRs - DSLR & Digital Camera Astro Imaging & Processing - Cloudy Nights


The direct way to the file is here: Dropbox - pete_xl_selective_colors_SHO-share version SNB filter.psd


If somebody likes to play with it I'd be happy to get a feedback.


Scotty (VoiceOfReason) will share his action with the filter soon. He tries to implement a starmasking function into the filter!


Cheers
Pete

I’ll try it out Peter
Usually I’m all in when experimenting/testing, but with this I’m still crawling on the floor not ready to stand up, not even when I’m holding on to something

Here's the actions for it. I'm working on getting the mask right for stars, so expect more to come. Dropbox - NarrowbandOSC.atn