Pixinsight - New Process - ArcsinhStretch

[PKay]

Latest process for PI.
It can be used instead of histogram transformation or similar.

Here is a quote from PI:

"Arcsinh stretch is a method of stretching image data while preserving the orignal color (i.e. the original R,G,B ratios) in every pixel of the image. Color is retained from the very faintest object to the very brightest unsaturated star. Contrast this to a traditional gamma curve stretch where the brighter objects become desaturated (i.e. bleached) during the stretch."

Anyway I tried it on NGC2237 (poor noisy data - Moon out) and the difference was quite marked.

Two images attached: The 'more' red one used the ArcsinhStretch process.

Only thing I noticed was the stars 'went with it' or became red as well.

[sil]

Isn't this central regions here meant to be much bluer? Really looks like this has only been applied to the red channel only. I haven't seen or tried arcsinh stretch yet myself.

Looks like progress towards the approach I use based on Geralds technique for enhancing nebulosity without pushing stars.

Out of interest have you tried using the new Photometric Color Calibration process after this to see if you get the stars back to true?

[PKay]

Hi Sil

Sorry bout not replying...NBN satellite internet is hopeless.

Yes I have tried the PCC but can't seem to get it to work.

Still at the trial and error stage with PI, and really have no idea what I am doing most of the time.

[RickS]

I tried ArcsinhStretch on an example of PCC calibrated data (VdB 149/150) and the result looked reasonably colour accurate. The result was too saturated, even for my tastes, but a blend with a HT stretched version produced a nice result.

Cheers,
Rick.

[sil]

Thanks Rick, My computer is busy with other work so i havent tried myself :(

My interests have been side tracked with building the Plumgeek Spirit Rover (one handed a challenge but got it done) and seeing if i can paint a model car for a gift. I get sidetracked easily by my interests and hobbies :(


Interested your feedback re saturation, I suspected it would "overprocess a little" and do that being a cosmetic stretching. But glad you feel it doesn't unbalance the colour correction. The process I take uses a lightness channel mask in it and this may push my source image enough to get a better lightness mask for my own workflow though. Ways and means :) Give me a simple one click solution and I'll turn it into a 1,000 step monstrosity for a 0.1% gain :)

Peter, I hope you're enjoying your PI learning curve. Take lots of notes and try lots of things, try to work along with tutorials with your own files, it really helped me adapt what was being said and done to my own understanding and personal goals for my data. It was a while before I found out that the exact same process in PI is used by different tools, under different names and with varying degrees of user control for different reasons. So dont worry if you play around and think this tool does the same job as this other one. And you can probably do everything with PixelMath alone if you have that level of understanding :)