A bit of fun with a rose up close

[wimvb]

the Rosette Nebula with limited data. Due to the target going out of sight from my location, and clouds covering that part of the sky when we had clear-ish nights, I never got around imaging this nebula with a blue filter. So instead of waiting for next season, I decided to combine the H-alpha, Red and Green data into a tone mapped image. So here are two mappings: HaRG and RHaG, just for fun

Technical details:

12 x 4 minutes H-alpha, 8 x 5 minutes Red and 8 x 5 minutes Green. As always I used the MN190 and ASI294MM to collect the data.

Processed in PixInsight

Enjoy

[wimvb]

One more variation. This is also a HaRG map. The difference is mainly that I used linear fit before combining the data and no colour calibration afterwards.

[Rodd]

The middle image looks almost like a SHO.  Very nice.  While I do like "natural palettes"--meaning RGB or NB that are made to mimic RGB  (such as HOO), I am always thankful I captured SII when I decide to do it.  Your middle image looks like SHO.  Very nice.

I just read a thread where it was opined that blue is unnecessary, because Lum.  Using that logic, I suppose  contains all of RGB and to get blue, one just has to subtract G and R from lum.   I suppose this is true for each color.  Not sure I am convinced.  It makes sense, and certainly saves a night (or 2 or 3) of imaging if true.

[wimvb]

Thanks, @Rodd. The "no blue is necessary" theory is just that, a theory. If you have L, red, and green, you can theoretically re-create blue by subtraction. But that only works if the red that you collected is just as strong as the red component in L. Same goes for green. If for whatever reason the red component in L doesn't match the red collected signal, you either subtract too much or too little. Either way, you are not left with a clean blue signal after L - R - G. Also, any camera induced signal (noise, dark current, amp glow, pedestal) will have to be accounted for. That's why this only works in theory, but rarely in practice.

Mind you, the same is essentially true when we do red-continuum subtraction, where we subtract red from H-alpha. That's why in the Pixinsight documentation/processing example a correction factor is used, and the median red signal is not subtracted: Ha - c*(R - med(R))

Anyway, in the image I posted here, I combined the data I had, all in good fun, with no aspiration for being "correct" in any way.

[Rodd]

10 minutes ago, wimvb said:

c*(R - med(R))

What is c.  I use Ha-(r-Med (r))*.05..........I fiddle with the multiplier as needed.  .05 may be the right number but it could be off by a factor of 10.  What is the correction factor?  Is this a new revision?  I have not seen it before.

[wimvb]

57 minutes ago, Rodd said:

What is c.  I use Ha-(r-Med (r))*.05..........I fiddle with the multiplier as needed.  .05 may be the right number but it could be off by a factor of 10.  What is the correction factor?  Is this a new revision?  I have not seen it before.

Your .05 is my c. I tweak this value, but most of the time it's between 0.3 and 0.5. That is with a 7 nm Ha filter and different gain/offset values for the red filter and Ha filter.

[Rodd]

2 hours ago, wimvb said:

Your .05 is my c. I tweak this value, but most of the time it's between 0.3 and 0.5. That is with a 7 nm Ha filter and different gain/offset values for the red filter and Ha filter.

I see. A matter of position. Yeah, .3-.5 is more like it.  I have had pretty low factors. Probably depends on the signal strengths. 

[wimvb]

47 minutes ago, Rodd said:

I see. A matter of position. Yeah, .3-.5 is more like it.  I have had pretty low factors. Probably depends on the signal strengths. 

Yes. And signal strength depends on exposure time, gain settings, bandwidth of the filters, etc. That's why even Vicent Peris won't attempt a more rigorous solution .