Luminance sub frames

[Adam1234]

Hi all, I had a question on luminance sub frames. 

So I've currently got 90 x 60s (total 90 minutes) sub frames each of R, G and B (of the Crescent Nebula if anyone interested). I'm now taking some luminance subs - should I go 90 mins total to equal each of the R, G and B, or 270 mins total to equal the total RGB?

Cheers

Adam

[Adam1234]

Either way, I think I'm stopping with the Crescent tonight, its going to be getting too close to the street light soon, and my guiding seems to be playing up for some reason 

Edited November 4, 2020 by Adam1234

[wimvb]

The general guideline is to use much more luminance than rgb, so you can get enough data to enhance detail and weak signal. You use RGB to provide colour to the L image. But the crescent is different, because it’s mainly an emission nebula. Especially if you have light pollution, you should consider to go for Ha rather than L.

[Adam1234]

Brill thanks!  I got a couple hours of Ha on the Crescent, might get some more tomorrow then. 

So Ha better as luminance for emission nebula what about other nebula like planetary/reflection etc?

 

[david_taurus83]

Planetary tends to be emmission as well but reflection is mainly broadband. The Iris is reflection for example. You need to be careful when using Ha as Luminance around broadband though. Take Orion for example, if you used Ha as Lum it practically removes the Running Man as that's reflection.

Edited November 5, 2020 by david_taurus83

[wimvb]

2 hours ago, Adam1234 said:

So Ha better as luminance for emission nebula

Ha enhances the red from RGB. You can also blend it into luminance. But because most of the information is in the narrow band Ha data, luminance (broad band) doesn’t necessarily add much quality wise.HaRGB may be all you need. The Ha provides structure and detail, while the RGB provides colour in the stars.

Reflection nebulae are broad band targets which don’t gain from using NB filters. Eg, the Pleiades (m45) are imaged with traditional LRGB techniques.

[ollypenrice]

A a general rule I find that the processing is made easier if you match exposures, so 3 hours of luminance for an hour each of colour. However, it is possible to shoot far more than that, say 6 or 9 hours of luminance for an hour per colour, but you may have to work on adding the L in iterations so as not to bleach out all the colour.

As Wim says, the main place for Ha is in the red channel because it is red. It won't pick up any blue signal so it will dim the blue down if applied as luminance. The classic Ha as lum 'look' is pink nebulosity with big blue star halos. Not good. Once you've added it to the red you can try a bit as luminance but I limit it to about 15% opacity when I do use it that way. You can add OIII to both green and blue channels, too.

Some people make an HaOIIIOIII image (called HOO) and then add RGB stars to it. This strikes me as an illogical approach. I make an RGB image with a gentle stretch to get the stars small and colourful and then lighten the red with Ha and the green and blue with OIII. It avoids the clunky artificiality of blending in the RGB stars and preseves any broadband blue in the RGB layer. This gives a better rounded colour space.

Olly

[Adam1234]

Thanks for all your responses, that's helped alot and given me the info I needed! 

1 hour ago, ollypenrice said:

You can add OIII to both green and blue channels, too.

Thanks for that tip, I was wondering whether you could do that and was going to ask that question next, now I don't need too 😀

 

Thanks!

[rickwayne]

It's worth spending extra time on luminance because of the physiology of our vision. We discriminate fine detail largely in greyscale, so the visible structure of your image will be much improved by more luminance integration time. This reduces both what we commonly think of when we refer to "noise" in a digital image -- small artifacts analogous to grain in a film image -- but also smoother, more finely-delineated detail in both small structures and in sweeps of tonal variation. Noisier RGB data are tolerable, in fact some imagers commonly blur the RGB a bit to reduce grainy chrominance noise.

[ollypenrice]

24 minutes ago, rickwayne said:

It's worth spending extra time on luminance because of the physiology of our vision. We discriminate fine detail largely in greyscale, so the visible structure of your image will be much improved by more luminance integration time. This reduces both what we commonly think of when we refer to "noise" in a digital image -- small artifacts analogous to grain in a film image -- but also smoother, more finely-delineated detail in both small structures and in sweeps of tonal variation. Noisier RGB data are tolerable, in fact some imagers commonly blur the RGB a bit to reduce grainy chrominance noise.

I'm listening but not convinced. Once the L is added to RGB it ceases to be greyscale and becomes colour, yet we still see the improved level of finely resolved detail.  My own explanation for why this happens is different. The L layer has, per unit time, about three times the signal of a single colour layer because it simultaneously captures R and G and B. Because it has three times the signal it goes deeper on the faint stuff and can take far more sharpening and contrast enhancement. (Sharpening is contrast enhancement but on smaller scales that what we consider contrast local enhancement.) I think that's why it is effective.

I quite agree that a greyscale rendition of an image will allow us to discriminate more detail but the problem is that LRGB is not greyscale. It's RGB. The L is absorbed.

1 hour ago, Adam1234 said:

Thanks for all your responses, that's helped alot and given me the info I needed! 

Thanks for that tip, I was wondering whether you could do that and was going to ask that question next, now I don't need too 😀

 

Thanks!

If you want to be really crafty begin by making an L Ha-R G B image and save it.

Now add OIII to the green channel and save this as OIII to green.

Go back to the L Ha-R G B and add the OIII to blue, calling it OIII to blue.

Now, in Ps Layers, you create 3 layers thus:

OIII to green

OIII to blue

L Ha-RGB.

This lets you adjust the green-blue balance of the OIII in real time while seeing the result.

Olly