›

Issues with artifact for stars with ArcSinhStretch for both RGB and Nb images

Christopher Foster
in PixInsight
Adam,

I have now gone through all your videos and many several times. THey have helped a lot. I am still have my most trouble in two areas: Stretching RGB images to non-linear (either NB or LRGB filters) and blending Nb data into RGB data.

I have found that the ArcSinhStrech is very prone, with my data, to create rather ugly stars. I dont know if its my data or the way I am using the tool. I have experimented with it extensively. If I use it to fully stretch the image, my stars look terrible with ugly halos and rings. If I use the STF and HT method, I dont see the same issue, but Ioose the benefit of the color saturation with ArcSinhStretch. I am forced to do a star replacement at some point to repair the star artifacts. I dont know what is causing this issue, but it is consistent across multiple data sets and targets. I have also had feedback on cloudy nights with other uses with the same issue.  Any help in this area would be greatly appriciated. 

I am still working on NB blending, but dont really want to dive into that yet accept to say that we really could use a video on that topic including the NBLRGBCombination script or other methods of combining Nb data into the RGB or Lum.

thanks,

Chris Foster

Comments

  • BlockHeadBlockHead
    Chris,

    I generally do not use ArcSS to do all of the work. I usually get as close as I can with a permanent stretch I choose after manipulating the ScreenTransfer process (and then HT). *Then* I use ArcSS for the remainder of the stretch. Basically you need to stretch the stars a bit- or yes, they will be pretty ugly. I use ArcSS to color the faintest things... NOT to protect the bright things (per se). I am using it in a way that is similar to Shadows/Highlights of PS. Same kind of methodology.

    -the Blockhead
  • BlockHeadBlockHead
    By the way, there is no reason why you could not blend the two methods you know together.... 
    -the Blockhead
  • Christopher Foster
    Adam,

    Thanks. That is really good advice. I have been doing the reverse - using ArcSS first and then HT. I will try to do it the other way and see how it goes. I also have been blending things regularly for many applications such as NB addition to RGB, Star replacement, HDRMLT, and USM.

    I posted some data on CN:

  • jeffrey merrilljeffrey merrill
    I've been struggling with ArcSS and star cores as well..I tried every combination ..just ArcSS for RGB in small increments, HT then ArcSS, and ArcSS then HT..the last of which got something "acceptable" I guess..so I love the color of the stars from a distance and my galaxy color, but with the final LRGB the core of the stars/ star artifacts are still issues

    According to Mark Shelley's pixinsight tool info ArcSS should be used with linear data, no previous stretch.  so not sure mathematically what might be the issue with using HT first...

    I'm wondering what other alternatives with something like maskedstretch/curves saturation etc would be less damaging to stars?

    jeff
  • Christopher Foster
    I have basically not been having issues with this any more. Here is what I do:
    0) dont use any chromatic noise reduction in LRGBCombination in the linear state - this kills the star cores
    1) use STF for an initial weak stretch so that I just barely see the DSO
    2) use ArcSS for a the next brightening or use mask stretch for the same purpose to bring out the color
    3) use histogramTransformation to reach the final brightness, but this is not really bright and I also set the blackpoint.
    4) After stretch, I run repaired HSV script to help the bright cores. 

    THis works well
Sign In or Register to comment.