Processing for Cat's Eye Nebula?

[KuriousGeorge]

Hi folks, I'm starting processing on the Cat's Eye Nebula (attached). I believe the core in the 32-bit stretched .fit file is preserved. If so, is there some way I can say "If the pixel value if greater than X then multiply by Y" ?

If something already exists, that would be great! Just trying to apply a simple stretch and fix the core. It seems difficult to do the opposite (i.e., keep the core down and stretch the rest).

[aworonow]

An exponential stretch can be expressed as 1-(1-$T)^n, where n > 0 and $T represents the pixel values in the image that you drop the triangle on. So, in PixelMath,

if ($T>a, 1-(1-$T)^n, $T)

where "a" is you lower limit of intensity for applying the stretch. You can define the values for a and n in the symbols tab.

Alternatively, you could adjust the stretch using CurvesTransformation.

cheers. Alex

[KuriousGeorge]

Thanks Alex!

A few quick questions:

1. I enter the above equation in PixelMath > RGB/K. I get the error "Unknown function identifier: if ($T>a, 1-(1-$T)^n, $T)
2. What is the range for "a" for a 32-bit monochrome .fit file?
3. What is "n" for no effect and n for maximum effect?

[pfile]

"if" should be "iif"

rob

[Cosmick]

If something already exists, that would be great! Just trying to apply a simple stretch and fix the core. It seems difficult to do the opposite (i.e., keep the core down and stretch the rest).

HDRMultiscaleTransform is designed for this purpose. It will usually work pretty well after transferring the default STF to the image via dropping the STF icon onto the HistogramTransform process.
For small scale features reduce the number of layers and for large scale increase the number of layers. To reduce the impact on the darker areas you can use the built-in Luminance mask or you can make your own.
It can take a bit of trial and error to get the best combination of stretch and HDRMT settings. I often stretch more than I normally would do, in preparation for HDRMT.

[KuriousGeorge]

Thank you Mick. I didn't know about that feature. It works really well. Please see attached.

[Cosmick]

Thank you Mick. I didn't know about that feature. It works really well.

No worries 
Also works well on galaxies with bright cores. I often also add together the HDRMT processed image and the one with only a stretch applied. This allows fine control over the amount of HDRMT by varying the percentages - 50/50, 67/33, 25/75 etc.

[KuriousGeorge]

Thanks Mick. Good idea. Do you also do that % add in PI? If so, what feature? Normally I would use a PS luminance layer for that for real-time % opacity preview.

[pfile]

you can either use pixelmath to blend 2 images (Image1*0.5 + Image2*0.5) for instance, or create a mask (for instance, 50% = "0.5" in pixelmath expression, "create new image" ticked, and apply to image, then use the resultant image as a mask). then with the mask applied to Image1, apply the pixelmath expression "Image2" to Image1.

or, you might use an image derived from the main image as the mask (for instance, a luminance image which has been smoothed out slightly) when applying Image2 to Image1. or you might first multiply that smoothed luminance by some amount before using it as a mask. masking is computed in 'real time' so if you make changes to the mask image those changes will be reflected the next time you do an operation to the masked image.

rob

[KuriousGeorge]

Thanks Rob. I think I need to gain a bit more experience in PI to do that quickly and efficiently.