Settings in the compose module v 1,8

[fmeireso]

Hi all

This is more a question then a problem. Maybe Ivo can jump in here.

What is actually the best way to process Ha Luminance and OSC color? Most of you know i use a mono DSLR with a Ha filter for luminance and i use an unmodded DSLR for taking color.
Many times i used L,RGB in compose module but lately i got good results L+Sunthetic L from RGB, RGB. On this occasion i loaded the NB accent File also with the Luminance. I got good result, still i think that is not the way to go.

Comments very welcome

[admin]

What is actually the best way to process Ha Luminance and OSC color? Most of you know i use a mono DSLR with a Ha filter for luminance and i use an unmodded DSLR for taking color.
Many times i used L,RGB in compose module but lately i got good results L+Sunthetic L from RGB, RGB. On this occasion i loaded the NB accent File also with the Luminance. I got good result, still i think that is not the way to go.

Unfortunately, there is no one "best" way to combine narrowband and visual spectrum data, but there are more legitimate and less legitimate ways. Much depends on your goals, the object and the characteristics of the data you acquired.

There are many different ways of combining the signal, but you should always try to understand what is happening (or not happening) to the various types of signal, and what the end result means. If you can explain what you are looking at in the end result and why things have the coloring and detail that they do, then you're on the right track.

It sounds like you want your Ha to dominate the detail in your image (valid for some datasets and objects, but not so much for others); you would indeed do this by importing Ha as luminance. Adding RGB as well as luminance to Ha doesn't make a whole lot of sense from a signal processing point of view; they are two fundamentally different signals with wholly different characteristics. E.g. calling the combination "luminance" would not be true (and Tracking, decon, etc. would be drawing nonsensical conclusions about noise levels and PSFs, etc.). You can really only create a synthetic luminance and add it to a "real" luminance stack if the R, G and B add up to be the exact same wide spectrum as L.

What would you define as a "good" result?

[fmeireso]

As a good result i would propose the IC 434.

viewtopic.php?f=11&t=2752

So i am only talking about emission nebulae, forgot to mention this in my first post. Now the horsehead i tweaked a bit in GIMP but that was only more saturation and playing a bit with contrast and stuff like that. Actually i could have boosted the saturation in Startools also, i just did'nt

[admin]

As a good result i would propose the IC 434.

viewtopic.php?f=11&t=2752

So i am only talking about emission nebulae, forgot to mention this in my first post. Now the horsehead i tweaked a bit in GIMP but that was only more saturation and playing a bit with contrast and stuff like that. Actually i could have boosted the saturation in Startools also, i just did'nt

Please keep in mind that the horse head nebula and surrounding complex is very dusty (lots of dust filtering light other than red). It is a region that appears noticeably redder than less obscured H-II areas (which will appear more purple/pink in the visual spectrum).

For emission nebulas with strong Ha emissions, my general advice would be to acquire sufficient visual spectrum signal, and use Ha for pure luminance.
This way you have guaranteed color information where Ha detail exists, yielding correct visual spectrum coloring.

If you don't have sufficient visual spectrum signal for whatever reason, and you really want a visual spectrum-ish rendition, you may want to use the NBAccent module instead to bulk up the visual spectrum data with artificial red (or Balmer series pink/purple).

If you don't have sufficient Ha signal (for example when imaging a galaxy, where only really knots tend to show up), then here too the NBAccent module would be the better choice.

I would not use both Ha for luminance and the NBAccent module, as the luminance and NB accent stretch will differ and will interact incorrectly, causing reddening/modification of other areas as well.

[fmeireso]

Thanks very much Ivo,

This pretty much confirms most of my assumptions. But since i took an experiment with Nbaccent and Ha as Luminance and got a descent result on the Horsehead i just got confused about. Allthough, and i mentioned this in another thread, it seemed not right.

About taking color, do you think or know that a modded DSLR would be a benefit over a unmodded DSLR? Since indeed sometimes the Ha is weak in an unmodded DSLR which might compromise the coloring, i am guessing...

thanks

Freddy

[admin]

Thanks very much Ivo,

This pretty much confirms most of my assumptions. But since i took an experiment with Nbaccent and Ha as Luminance and got a descent result on the Horsehead i just got confused about. Allthough, and i mentioned this in another thread, it seemed not right.

About taking color, do you think or know that a modded DSLR would be a benefit over a unmodded DSLR? Since indeed sometimes the Ha is weak in an unmodded DSLR which might compromise the coloring, i am guessing...

thanks

Freddy

Glad that helped Freddy!

For astrophotography having a modded DSLR is definitely of benefit, particularly if you want to chase extra detail and interesting structures in "red"; so much detail is available in Ha! And you can always temporarily "unmod" your DSLR by simply putting a luminance filter in front of it (which puts back the Infrared cut-off).

[Mike in Rancho]

I would agree with the general concept, but, Freddy you may be an unique use case?

I think normally it's a stepping stone progression. DSLR, modded DSLR, and then maybe into astrocams. The modded DSLR opens up the red side to be more like an astrocam. And with removal of that red-side filter, of course there's still the bayer filters and the sensor QE graph itself that determines the response, which still usually falls off on the right side, though that's true of all OSC. But while they will pick up Ha now in broadband, the common usage of such mods tends to be with a duoband filter.

I believe most Ha-mod Canons are still cut from 400-700, so I'm not sure an L or UV-IR would change anything? My full spectrum D5300 of course needs an L or UV-IR regardless, in broadband, otherwise the far blue and far red can mess things up.

With Freddy already having the mono-mod Canon though, I wonder about the cost-benefit of a third DSLR.

I might just get a red filter (if not a full LRGB set) and do some extracting and recompositing.

[fmeireso]

With Freddy already having the mono-mod Canon though, I wonder about the cost-benefit of a third DSLR.

Yes, that is the purpose of my question. I just wonder if an modded DSLR would benefit just for taking color instead of with the unmodded one. I am sure not going to change my workflow for emission nebulae , this means taking ha with the mono and color with well a color DSLR OSC.

It is my tihinking, a mono DSLR or any mono cam with ha filter takes more nebulosity in just because there is no bayer matrix. In the mono every pixel of the sensor is recording data, with color only the red works.

[Mike in Rancho]

Yes, that is the purpose of my question. I just wonder if an modded DSLR would benefit just for taking color instead of with the unmodded one. I am sure not going to change my workflow for emission nebulae , this means taking ha with the mono and color with well a color OSC.

It is my tihinking, a mono DSLR or any mono cam with ha filter takes more nebulosity in just because there is no bayer matrix. In the mono every pixel of the sensor is recording data, with color only the red works.

It's also the filter Freddy. Blocking everything else sets up that contrast so that you are only getting Ha. Duobands with modded DSLR (or OSC astrocam) work quite well, and debayering is often pretty darn good, if not fully equivalent to mono. Also your astrocam sensors these days are really good and have better response, though maybe there are newer DSLR's (or mirrorless, more likely) now that have things like the 571 in them?

EDIT: Hey Freddy I was thinking (I know, look out! ) -- it seems your cameras are a 600D and 800D, so you already have different pixel scales, like 3.7 and 4.2, which must then be scaled to match by the stacking program (I think PI?).

What would be good then is to see if anyone has some RGB-no filter calibrated subs they could share from an APS-C Ha-mod Canon, at a similar focal length and type of refractor, and of an Ha-dominant nebula target that you already shot or could shoot. That way you could test to see if, and by how much, the color response differs when you composite in your intended (but difficult) Ha-as-L fashion.

[decay]

What would be good then is to see if anyone has some RGB-no filter calibrated subs they could share from an APS-C Ha-mod Canon, at a similar focal length and type of refractor, and of an Ha-dominant nebula target that you already shot or could shoot. That way you could test to see if, and by how much, the color response differs when you composite in your intended (but difficult) Ha-as-L fashion.

Very cool, Mike Hehe, at least as I had a similar idea and I tried to combine my two data sets of M1 in Freddy's Ha-as-L fashion. I used the duoband (l-enhace filter) stack as L and visual stack as RGB in compose (Canon EOS 2000Da full spectrum). But the result was very disappointing . But of course, that may be due to my limited compose skills and due to the very poor data quality. And I used Ha and OIII so maybe it's worth another try and to use only Ha.

I could share the datasets if this would be of any help and someone would like to have a go (Freddy, Mike?).

Best regards, Dietmar.