Horsehead and Flame, HaRGB combined ("MartinB method")

[SteveL]

After reading about MartinB's method for combining Ha into/onto RGB data for his Cone nebula image a while back, I was itching to have a play with some of my own data.

Erik Larsen ( Flickr: eriknlarsen's Photostream ) has very nicely allowed me to use his RGB data in this combined image, and to post the results. His data was 11x300s, taken with a modded Canon 500D through an Orion ED80. Many thanks for allowing me to show this.

So, I took Erik's RGB Horsehead and Flame RGB data

and the Ha data that I took on Christmas Day

Threw them both into Registar, aligned the Ha data to the RGB data, saved them out.

Now using MartinB's method, I load the RGB data into a layer, then duplicated that RGB layer on top with "soft light" mode. This increases the contrast, but also drops the black point, so I used some levels/curves to bring that back up a little.

I then pasted the greyscale Ha data into a layer on top, set the green and blue channels to black (only the red was left), then set that layer method to "lighten" to it only affected the reds.

After some careful levels and curves on the duplicate RGB and the Ha layers, it got somewhere close to what I wanted. The stars have a slight red halo from the larger Ha stars, which should be easy to fix later, and maybe I over did it with the Ha lightening

This is the result. Lots of lovely reds, but no loss of contrast like previous attempts of using just an additiona Ha layer, plus the other colours are not washed out. I am currently digging out all my previos Ha RGB images to see if I can improve on them with the same method.

(Click on image for larger version)

[steppenwolf]

set the green and blue channels to black (only the red was left),

Not sure quite what you mean in this stage - can you expand?

A very interesting technique by the way

[SteveL]

Not sure quite what you mean in this stage - can you expand?

In PSCS2, I pasted in the grey Ha data (which is RGB grey) on top, then selected the channels tab, selected only the green channel, "select all", right clicked and "fill" with black, repeat for the blue channel. This leaves only the red channel with anything above 0, and so when used as a lighten layer, it only affects the red data on the layers below it.

I make any claim about this technique, it was all picked up from MartinB's cone nebula thread earlier

[ncjunk]

Its a great image but, just to be overly critical, missing some of the fine detail. I like the way you have added those to differing sets of data though.

[MartinB]

Glad it worked for you Steve, certainly helps with a very tricky problem.

[SteveL]

...but, just to be overly critical, missing some of the fine detail.

I think thats more due to the mismatch in pixel resolution between the Ha and RGB data sets than the process itself, althugh I did have to use a fair amount of noise reduction in the Ha layer to prevent the noise from showing. Registar had to increase the size of the Ha data quite a bit to match the RGB, so each pixel became 2x2 or 2.5x2.5 (maybe more), and so any noise at all looked horrible.

If/When I ever get a clear moonless sky, I`ll be out for the same FOV in RGB, and then I`ll return to this proces with matched data.

[RichieJarvis]

Very nice Steve - thanks for the testing of the technique - might give this a go with the rosette later in the year....

[resonator77]

superb images Steve,

what set up you used?

[SteveL]

superb images Steve,

what set up you used?

Here is the original post for the Ha data that I took:

http://stargazerslounge.com/imaging-deep-sky/93855-horsehead-flame-christmas-day.html