Ha-OIII NB Techniques?

[viktorbravo]

How would I produce a synthetic green channel from Ha and OIII data, whereas Ha=Red:  Ha+OIII=Green: OIII=Blue then subsequently blend the channels as follows:

In Photoshop the  OIII/Blue would be blended using Multiply, the Synthetic Green channel from Ha and OIII data using Screen, Clipping Mask, and the Ha/Red Channel would use Normal.  In addition, there is a Hue/Saturation adjustment layer assigned as a clipping mask for each channel.

How would I begin to go about doing a similar process in PI, or should I just stick to Photoshop for this?

Thanks for any advice.

[pfile]

no idea about photoshop, i've never used it for astrophotography.

i made this narrowband M42 from Ha and OIII:



i simply used pixelmath to create the image by setting R=ha, G=oiii and B=0.3*ha+0.7*oiii, with rescale unchecked. before doing the pixelmath, i used linearfit to fit the ha and oiii to each other.

of course there's a heck of a lot of postprocessing after that...

[viktorbravo]

Thanks for the reply.  I figured that I would use pixel math to combine the image but I wasnt sure.  I guess I have to read up on linearfit.  Also, after pixel math what process do you use to adjust the color balance?

Also, you M42 displays almost realistic colors so I imagine its quite a bit of work to get a result like that.

Are you planning on doing a little tutorial? 

Thanks again, Vince

[pfile]

after the synthesis of the RGB, i did colorcalibration.  after that... well, to be completely honest i used curves to get the color where i knew it should be from previously doing LRGB M42s. this amounted to boosting the reds and blues (blues, a lot) and pulling down the greens, and finally increasing the saturation quite a bit.

most of the time spent working on this image was on deconvolution and related processes. my stars are not round and so i was experimenting with ways to bring them back into round. toward the end i did some MT thru a star mask with odd-shaped structuring elements to try to round them up but i was not entirely successful.

linearfit is pretty straightforward to use - just tell it which of the two images is the reference and drag the triangle to the other image. i thought there was a forum thread about this tool, but i can't find it now. i'm not sure it really matters which image is the reference but in my case i used the oiii image since it's histogram was more shifted to the right to begin with, probably owing to the fact that my oiii filter has 12nm bandwidth while my ha is 6nm.

[viktorbravo]

Thanks again pfile.  I tired pixel math and color calibration with various combinations of Ha and OIII.  It works, but its very difficult to produce the desired color balance.  I kept at it for about 8 hours and finally gave up as I could not get the color I desired.  Unfortunately, I'll have to stick with PS for this type of image as it seems to offer better control over the color.

[dhalliday]

I try various combinations in Pixelmath...
sqrt(Ha*OIII)
or .1Ha+.2OIII
etc etc....it depends on the signals you have...andits always a compromise solution..

Its never a winning game because the "new" third channel is dragging down the other two...but often there is no SII to use...
True you have to  fiddle in Curves to adjust the colors...
I often do a DBE run first to remove the worst of the mashup...

I have many such images on my flickr site..
None of them fabulous...
http://www.flickr.com/photos/daveh56/6733367753/in/photostream

Dave Halliday

[viktorbravo]

Thanks Dave.  I'll keep trying.