Arcsinh Stretch

[smr]

I've just come across this methodology of processing, to preserve star colour etc.

http://www.markshelley.co.uk/Astronomy/Processing/Colour_Preserving_Stretch/colour_preserving_stretch.html

I process everything in Photoshop, where in the processing work flow should I implement this? At the moment I open the image in PS, and do some initial linear stretches to reveal some stars, and then separate the stars from the nebulosity. 

I'm not sure at which point I should apply the arcsinh stretching methods. 

[matt_baker]

I'd say when you've done your linear stage and transferring into the non-linear. Don't do a harsh stretch, just enough to keep the stars in good shape then mask them out to just focus on the nebulosity from that point onwards

[scotty38]

1 hour ago, smr said:

 At the moment I open the image in PS, and do some initial linear stretches to reveal some stars, and then separate the stars from the nebulosity. 

I'm not sure at which point I should apply the arcsinh stretching methods. 

As Matt says but just to reiterate I would not do any other stretching before using this method.

[barbulo]

Following this thread with interest since I am starting my AP journey and have a lot to improve.. 

I've heard before about this stretching algorithm and would like to give it a try some day.

1 hour ago, matt_baker said:

I'd say when you've done your linear stage

I believe I know the difference between linear and non-linear data but when you talk about linear stage, (here the newbie question) what actions can be done to maintain the linearity? I can imagine some like binning, cropping, flipping... But I am sure there are hundreds more and some of them very useful. Any clue?

[AstroMuni]

1 hour ago, smr said:

I'm not sure at which point I should apply the arcsinh stretching methods. 

This is tutorial for SIRIL a free software. I know you are trying to use PS but this gives you an idea of sequence of operations. https://siril.org/tutorials/tuto-scripts/

Several folk here use SIRIL to do the initial stretching etc and then do the final processing in PS.

 

Edited November 18, 2021 by AstroMuni

[The Lazy Astronomer]

1 hour ago, barbulo said:

I believe I know the difference between linear and non-linear data but when you talk about linear stage, (here the newbie question) what actions can be done to maintain the linearity? I can imagine some like binning, cropping, flipping... But I am sure there are hundreds more and some of them very useful. Any clue?

You can basically do anything that doesn't involving stretching the histogram.

[ONIKKINEN]

Another vote for SIRIL. Easy to learn (for an astronomy software ayway) and works well as a linear stage processing tool.

There is an Asinh stretch function in siril that preserves colour well. There is also the "regular" histogram transformation tool that i feel like works much better than fiddling with curves and levels in PS or gimp when it comes to colour retention. Also includes a photometric colour calibration tool which works pretty much every time. 

[900SL]

Just recently discovered Siril and I'm a fan already.

[Andy R]

On 18/11/2021 at 15:36, smr said:

I've just come across this methodology of processing, to preserve star colour etc.

http://www.markshelley.co.uk/Astronomy/Processing/Colour_Preserving_Stretch/colour_preserving_stretch.html

I process everything in Photoshop, where in the processing work flow should I implement this? At the moment I open the image in PS, and do some initial linear stretches to reveal some stars, and then separate the stars from the nebulosity. 

I'm not sure at which point I should apply the arcsinh stretching methods. 

I thoroughly recommend this tutorial for the initial arcsine stretch, there’s a couple of more tutorials on his channel worth looking at too  

hth

Andy

 

Edited November 20, 2021 by Andy R

[Gunshy]

If you can browse the Pixinisght documentation, you will find the equations / methodology behind the arcsinh stretch.   There is nothing particularly magic about the arcsinh stretch itself - just that it is a gentle, logarithmic stretch, that for modest stretches, distributes contrast evenly enough that artifacts don't arise in colors when it is employed.   What the arcsinh stretch process iin Pixinsight does, is to stretch the luminance channel, and normalize the stretch factor (by dividing the actual stretched luminance by the unstretched luminance).  Then this normalized stretch is applied across the individual colour channels to increase the standard deviation (saturation) of the individual channels  to enhance the colors.    If you do this with something harsher than logarithmic stretch, you will likely encounter colour artifacts, which is why the arcsinh function itself is used.

If you keep repeating a very soft stretch, such as a logarithmic stretch or an arcsinh stretch on an image  - you will get star bloat because while it appears gentler on the stars, repeated use or a very large stretch factor is way harder on the stars than even a true hyperbolic stretch, performed 1-3 times to get your  viewable non-linear image.  This is because the arcsinh stretch isn't particularly focused on your linear histogram peak.  The arcsinh stretch process is indeed wonderful at enhancing colors, when not overdone, but you have to know its limitations.  

If you  are a Pixiinsight  users,, you can tailor make your stretch to best preserves your colors and avoid star bloat, with a new script shown here https://stargazerslounge.com/topic/387548-pixinsight-improved-image-stretching/.    You can use this script to either alone (it is really good at preserving colors) or with a couple of shots of the arcsinh stretch process to further enhance them.   Even if you don't use Pixinsight, you might find the documentation helpful.

 

Clear Skies!