Adding Ha to RGB

[gnomus]

I was looking at the sensational photographs taken by Fabian Neyer - http://www.starpointing.com/index.html

I noticed that in a number of his images he uses both Lum and Ha (as Lum) combined with RGB filters.  I am hoping to get an Ha filter in the not-too-distant future.  If one follows the Lum + Ha approach, do you just stack the Ha and Lum data together to give one 'Luminance' image (which is then added to the RGB combo), or are Lum and Ha added to the RGB as separate luminance channels?  (And if so, is this done in a particular way?)   

I am always intrigued to see the differences in the amount of each channel that people collect.  For example, in http://www.starpointing.com/ccd/ldn1251.html Mr Neyer uses 4.3 hours of Red compared with just 2.7 hours of Green.  How do folks decide how much of each channel they should endeavour to capture?

[MarsG76]

Hi mate,

Using halpha as luminance can make your images look more detailed, and helps in a huge way when imaging from light polluted skies.... There are a number of ways to use halpha narrowband frame in your images, one way is to replace the red channel with your halpha, another is to mix your into red and another yet is to use Ha as a luminance channel on your RGB exposures.

As with everything there is a trade of to using halpha as luminance, it makes your image look washed out.... The reds look pink.

The way I like to use halpha in my RGB shots is to mix 30%-50% halpha into the red channel. Try both, or various combinations and judge with your results to what suits you... With the luminance channel again experiment with ha as luminance and adjust the transparency with or with out IR luma exposure...

During editing the trick is to try different combinations with the data you captured until you have something you like... The brute force approach...

I use photoshop for the majority of my Astro image editing.

[Freddie]

There's a thread related to this in Imaging-image processing.....   which may be of interest

[carastro]

I tend to do the same as MarsG76, use Ha instead of Luminance of the target is rich in Ha.  As said it does tend to make the reds pink but there are ways around this:

http://starizona.com/acb/ccd/software/ps_hargb.aspx

Carole 

[KyleStoke]

Hi Gnomus as Mars has said there is a few methods, as per these pics first one is Ha as lum only, second pic is Ha mixed with red

[KyleStoke]

[steppenwolf]

I prefer to blend Ha into the red channel rather than simply use it as a quasi luminance channel.

My workflow goes approximately like this:-

I extract the red channel from my original RGB image and create a new image containing just the red data. I then blend the Ha with the red data in the new image by pasting the Ha on top of the red and change the opacity to a percentage that is high enough to just show the extra detail contained within the Ha data without swamping the original red data. I then merge these two new channels then copy and paste them back into the red channel of the original RGB Image. Next, I paste the Ha data on top of the RGB data and change the blend mode to Luminosity - again, I adjust the opacity of the 'luminance' channel until the image starts to reveal more detail then flatten the data.

I do this iteratively so that the Ha content increases in small amounts until I achieve the result I am seeking.

[ollypenrice]

Adding Ha as luminance is anathema to me. Ha is red. If you add it as luminance your green and blue channels will be altered to be illuminated as if they were red. This is simply false. They are not red. They are not even remotely red! Imagine doing this to an LRGB image of the Pleiedes in which lots of outlying blue nebulosity makes for a fine spectacle. If you illuminate this in Ha, in which none of the nebulosity is visible, you will kill it stone dead. So what is wrong for one image remains wrong for another.

Yes, there is red in this image, but it is ERE (extended red emission) and not Ha. Illuminate this image in Ha and you will get stars and not much else. How can this be right?

In Photoshop I add Ha to the red channel in Blend Mode Lighten. Where the Ha is brighter than the red it brightens the red channel. that seems honest enough!

And then I take off my high hat and do try a bit of Ha as lum because it will, on Ha dominated targets, hoik up the contrasts and pull down the star sizes. But I keep the opacity very low indeed. 15% or so.

Olly

[KyleStoke]

I think the proof is in the pudding on the two images i posted although both are fairly roughly processed to my eyes the one where the Ha is mapped to red is far superior

[ollypenrice]

I think the proof is in the pudding on the two images i posted although both are fairly roughly processed to my eyes the one where the Ha is mapped to red is far superior

In many ways yes, it is superior. But your background sky is no longer neutral in the second image. You say 'mixed with red' but this mixing can be done in many ways. At regular intervals in processing it's a good idea to measure your background sky values. Personally I like parity in RGB and a value around 23 per channel in Ps. Some people like to have the bue rather higher.

Olly

[MarsG76]

Adding Ha as luminance is anathema to me. Ha is red. If you add it as luminance your green and blue channels will be altered to be illuminated as if they were red. This is simply false. They are not red. They are not even remotely red! Imagine doing this to an LRGB image of the Pleiedes in which lots of outlying blue nebulosity makes for a fine spectacle. If you illuminate this in Ha, in which none of the nebulosity is visible, you will kill it stone dead. So what is wrong for one image remains wrong for another.

Yes, there is red in this image, but it is ERE (extended red emission) and not Ha. Illuminate this image in Ha and you will get stars and not much else. How can this be right?

In Photoshop I add Ha to the red channel in Blend Mode Lighten. Where the Ha is brighter than the red it brightens the red channel. that seems honest enough!

And then I take off my high hat and do try a bit of Ha as lum because it will, on Ha dominated targets, hoik up the contrasts and pull down the star sizes. But I keep the opacity very low indeed. 15% or so.

Olly

I agree with you also, that's why when processing its a try different methods see what looks good for you in the final image.

[KyleStoke]

In many ways yes, it is superior. But your background sky is no longer neutral in the second image. You say 'mixed with red' but this mixing can be done in many ways. At regular intervals in processing it's a good idea to measure your background sky values. Personally I like parity in RGB and a value around 23 per channel in Ps. Some people like to have the bue rather higher.

Olly

Hi Olly, I agree with you, the image above was a quick reprocess with the red blended to see what it came out like and if you zoom in its quiet easy to see the blue is slightly misaligned. 

When you say 23% per channel do you mean using the info button? as on the image above i have values around R = 20 G = 22 B = 36.

I presume you just adjust the balance using levels on the specific colour you want to bring up or down

[ollypenrice]

Hi Olly, I agree with you, the image above was a quick reprocess with the red blended to see what it came out like and if you zoom in its quiet easy to see the blue is slightly misaligned. 

When you say 23% per channel do you mean using the info button? as on the image above i have values around R = 20 G = 22 B = 36.

I presume you just adjust the balance using levels on the specific colour you want to bring up or down

I measure using the Eyedropper Colour Sampler tool set to a radius of 3x3 pixels or higher. Then, yes, the values appear on the info panel. I put four samplers on in different parts of the image. You can adjust in various ways. If  the top left of each colour channel histo peak are not aligned then I'd start there in Levels by cutting back the black on the one(s) too far to the rigtht. Then you can use levels or curves or even the colour balance tool set to shadows.

Or if you have PI you could run Background Neutralization, though I never find I need to after DBE.

Olly

[KyleStoke]

I measure using the Eyedropper Colour Sampler tool set to a radius of 3x3 pixels or higher. Then, yes, the values appear on the info panel. I put four samplers on in different parts of the image. You can adjust in various ways. If  the top left of each colour channel histo peak are not aligned then I'd start there in Levels by cutting back the black on the one(s) too far to the rigtht. Then you can use levels or curves or even the colour balance tool set to shadows.

Or if you have PI you could run Background Neutralization, though I never find I need to after DBE.

Olly

Thanks for the info unfortunately i don't have PI but its on the list of things i will buy over time