Steve Cannistra's Method and PI

[jerryyyyy]

I have used this laborious method to convert H-alpha and O-III into an RGB image.  I have a bunch of new data from M20... may not be the best target as is mostly H-alpha, but as a learning exercise is there an equivalent method in PI? 

[Carlos Milovic]

Do you have any link to a description of that method?

[pfile]

http://www.starrywonders.com/bicolortechniquenew.html

[dgbarar]

Hi Jerry,

This is the PI thread where the bi-color blending method is discussed:

http://pixinsight.com/forum/index.php?topic=2777.msg18768#msg18768

More specifically use the following formula in PixelMath:

iif(OIII>0.5,1-(1-Ha)*(1-(OIII-0.5)),Ha*(OIII+0.5))

You will use the image generated from PixelMath expression above along with your original Ha and OIII images to construct your Steve Cannistra bicolor image.

I have only done one image this way, did not take much in away of data, or take much time to properly process.  Attached is an example

Don

[pfile]

so i have exactly zero experience with photoshop.

just the other day i was thinking about this problem and i noticed that cannistra uses the multiply and screen operation in photoshop to merge the ha and oiii. there's also a thread here where the photoshop blend modes are translated into equivalent pixinsight pixelmath expressions:

http://pixinsight.com/forum/index.php?topic=2409.0

now, that's all well and good, but i do wonder how to combine the pixelmath expressions to come up with the same thing. i guess there has to be some combination of the screen and multiply operations on the two images. i'll have to experiment.

to be honest i did try this once in PS and i didn't like the results. generally speaking i've sometimes used OIII for green and some combo of HA and OIII for the blue and liked that better. still, i've never been entirely happy with my bicolor efforts.

[jerryyyyy]

I also posted the question over at CloudyNights and got this magic formula:

Red = Ha
Green = Oiii
Blue = Oiii + 34.5%HA

I did the CS5 methods and the PI calcultion was much easier and I thing got a better result. 

Posted on Astrobin:

[dgbarar]

Hi Jerry,

Your image is very nice and I like the colors.  But the blending methodology that you used is not the Cannistra synthetic green that you requested in your original post.  Go through the PI threads that pfile and I suggested.  It will provide you the methodology that you were originally seeking.

Don

BTW.  Interesting that we both chose the same subject matter for our first bicolor using the Cannistra method.

[Geoff]

Nice pic Jerry.  I had a look at your astrobin gallery and your latest pics are a big step up from the earlier ones. Keep up the good work.
Geoff

[jerryyyyy]

Hi Jerry,

Your image is very nice and I like the colors.  But the blending methodology that you used is not the Cannistra synthetic green that you requested in your original post.  Go through the PI threads that pfile and I suggested.  It will provide you the methodology that you were originally seeking.

Don

BTW.  Interesting that we both chose the same subject matter for our first bicolor using the Cannistra method.

Yes, I got the original method and did it step by step (did it once before), works but not as good as this I think.  Thanks

[jerryyyyy]

Nice pic Jerry.  I had a look at your astrobin gallery and your latest pics are a big step up from the earlier ones. Keep up the good work.
Geoff

Thanks, Pfile has been a big help.  It also helps to do everything in PI and master one application.. jack of all trades and master of none? 

One other issue that has helped me is just collecting more and more images... eventually you get enough signal even in the SF Bay Area. 

I am working on a couple nebula and I think that I get better results with 1 (gasp) 60 minute image than two 30 minute images.  I am not sure the theory works on this, but to my eye it appears the way to go.  I have had a friend here help tune my Atlas so now I can get these long exposures.  The tuning of an Atlas mount is another black science.