Page 1 of 1

Getting correct colours with Ha(L)RGB

Posted: Wed Sep 25, 2019 9:48 pm
by Rkonrad
I was fortunate enough to find some great data from a post in Cloudy Nights of M16. The imager took HA, R G and B. So as there's no luminance data, I used Ha as L in the compose module. The red colour is really off I feel.
What would be some techniques to make the ha data as naturally red as possible? Here are a few images.

This is the default scientific colour:
m16-scientific_default.jpg (406.4 KiB) Viewed 1614 times

This is scientific colour as well but with the stars sampled to get a proper white balance:
m16-scientific_stars_sampled.jpg (386.52 KiB) Viewed 1614 times
Finally this is the stars sampled version with green reduced ( a lot):
m16-scientific_stars_sampled-green_reduced.jpg (380.01 KiB) Viewed 1614 times
As you can see the red is not right. I also created a synthetic L via RGB, then blended it 50/50 with the HA to create an L for LRGB. It still looks similar though. Any comments are appreciated!

The data can be found here ... sp=sharing by David Ault.

Cheers Richard

Re: Getting correct colours with Ha(L)RGB

Posted: Thu Sep 26, 2019 9:53 am
by admin
"Correct" color is not really a thing when narrowband data is involved, except when the Ha is purely used as luminance (as you did). Even then, there needs to be enough signal in the RGB composite to support the luminance of the Ha signal, while the very different luminance/detail may definitely make the scene seems "uncanny"; the detail is roughly correct, the colors are correct, yet it doesn't seem quite right. This is simply because the luminance is so very different to visual L stack.

Some people use Ha for luminance, some people use Ha+R+G+B for luminance, some people use (R+Ha)+G+B.
Some people layer in Ha separately after processing . There are many, many techniques. And much depends on your goals and the data itself....

It appears in this case that the Ha exposure time is vastly higher than the RGB exposure time. This only really makes Ha suitable as luminance.
Doing so can yield something like this;

Re: Getting correct colours with Ha(L)RGB

Posted: Thu Sep 26, 2019 2:15 pm
by Rkonrad
I didn't get "something like this".... perhaps you meant to attach an image?


Re: Getting correct colours with Ha(L)RGB

Posted: Sun Sep 29, 2019 3:08 am
by admin
Apologies on behalf of my sleep deprived brain... :oops:
NewComposite.jpg (275.69 KiB) Viewed 1564 times
This is a simple L, RGB combine using Ha purely as L.
Emission coloring and star coloring is correct. The only thing markedly different is the luminance detail, which is much punchier and "different" to what you'd expect based on the visual spectrum luminance we're used to.