Narrowband processing (especially color combination) with Pixinsight

[mstriebeck]

Hi,

I used in the past Photoshop to combine and process my narrowband images. I recently found this tutorial how to do this in Pixinsight: http://lightvortexastronomy.blogspot.com/2013/02/tutorial-nebulae-pixinsight-workflow.html. I tried it out and here are the two versions that I came up with:

Photoshop: http://www.astrobin.com/86300/A/
Pixinsight: http://www.astrobin.com/86300/C/

(both using the Hubble palette).

I like the amount of detail, contrast and smoothness of the Pixinsight image a lot. But I like the color combination of the Photoshop image better then the Pixinsight version. In Pixinsight, I found it very cumbersome to play with the PixelMath process to try different color mixtures (way easier in Photoshop).

Are there any other ways (more WYSIWIG-like) to combine colors and see the effect in Pixinsight?

Thanks
     MarkS

[pfile]

+1 - not sure about needing wysiwyg but i feel like it's hard to arrive at the look of the 'modified' hubble palette where the green Ha signal comes out blue. if anyone has pointers along these lines i'd be grateful to hear them.

juan's 'natural' mapping works reasonably well for me, but i like the modified hubble look as well.

rob

[RickS]

You can use a realtime preview with the CurvesTransform to tweak the colour interactively after doing the colour combine.  The "H" curve is useful and you can also do magic with the "a" and "b" curves (helps to have a vague understanding of CIELAB).  If I want to work on specific colours I build a mask using PixelMath and the LAB a and b values.

Cheers,
Rick.

[RickS]

PS. the original narrowband images will probably make interesting masks to use with Curves as described above.  Also try the R, G & B curves if you like.

[pfile]

hoped there was more to it - have done stuff with the *a*b curves and never got there. guess i need to keep trying.

rob

[RickS]

Rob,

I found this book useful: "Photoshop LAB Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" by Dan Margulis.  The book is way longer than it needs to be for someone with a clue about colour spaces and you have to do some translation from Photoshop to PI but it did help me a lot when I started playing with LAB in PI.

Cheers,
Rick.

[mstriebeck]

I gave it a try and could indeed bring out more detail by stretching the red more and the green less. Unfortunately, it also makes my background very red'ish. I tried using my SII image (with clipped shadows and highlights) as a mask, which reduces the effect, but not completely.

[RickS]

Protect the background and dim areas with a mask if you're moving the central point of the a or b curve (which will cause an overall colour shift).  Extract the CIE L component of the image, use HistogramTransformation to clip the black point then apply as a mask.  Alternatively, use RangeSelection to make a suitable mask.

Cheers,
Rick.

[Josh Lake]

+1, Range Selection will be your best friend. With a Real Time Preview, you can really carefully hone your mask to just look at the features you want at the right smoothness. Play with it if you haven't before -- it's become a real image-saver for me.

[chris.bailey]

+2 for Range Selection. It can also help to make a Star Mask and combine the two in Pixel Math to try and protect the stars from wild colour shifts.

[pfile]

also one thing i learned from one of alejandro's tutorials is that you can multiply your L mask by the range mask to get a 'strong' mask which still respects the L data in the interesting parts of the image.

rob

[mstriebeck]

Thanks for the guidance - I will play with that.

[Josh Lake]

also one thing i learned from one of alejandro's tutorials is that you can multiply your L mask by the range mask to get a 'strong' mask which still respects the L data in the interesting parts of the image.

rob

Great tip, thanks Alejandro via Rob.