Lessened overall quality of NSG images

Comments

• BlockHead

  November 2022

  Hi Michael,

  
  

  @johnmurphy is probably best to answer these kinds of questions. I was able to download the first image. The second image (link) is not public (accessible to all that have the link). 

  
  

  I would not expect any kind of change in color regarding using NSG or not... 

  So, I guess I would need to see the images. 

  In the version that is not using NSG- what were the image integration settings?

  
  

  -the Blockhead
• John Murphy

  November 2022

  Hi Michael,

  When used correctly, NSG should never produce a poorer quality image.

  
  

  It does not directly change the color balance or the color saturation, but since we are normalizing the target images to a reference image, the final result will be close to the color balance and saturation of the chosen reference image. Choose the reference image wisely!

  
  

  A surprisingly common mistake is to accidentally use NSG with the unregistered images. This is catastrophic, and tends to reduce the image contrast and color saturation. The final image quality would then be very poor.

  
  

  However, it is difficult to diagnose the problem without the data. One of the images is not public. The original reference image would also be useful.

  
  

  Regards, John
• Michael Lehv

  November 2022

  Adam and John,

  
  

  I apologize for the restricted link. It should now be working.

  
  

  The reference images used for Registration, within NSG and in the non-NSG integration were all identical. 
  

  
  

  The only processing difference between the two was in the integrations. The NSG integration used the settings populated by NSG (Local Normalization, PSF Scale SNR and GESD). The non-NSG image used Additive with Scaling, PSF SNR, Winsorized Sigma Clipping and Scale + Zero Offset. 

  
  

  I don't know if these Image Integration differences could affect the saturation differences between the Image Integration masters. 

  
  

  Thank you, 

  Michael 

  
  

  
  

  
  

  
  
• John Murphy

  November 2022

  Hi Michael,

  
  

  You also need to provide a link to the NSG reference image. Since all target images are normalized to this image, it is vital to know what the reference image looked like in order to be able to judge if the normalization worked correctly.

  
  

  Note that you should not default the NSG reference image to the one that was used for registration. The NSG reference image must be chosen with care. This is important because all the target images will be normalized to this NSG reference. I would recommend using the NSG Blink section to choose the NSG reference image.

  
  

  Regards, John
• John Murphy

  November 2022 edited November 2022

  Hi Michael,

  I have just loaded the two images and had a look at them.

  I personally think that the NSG image is the better image.

  
  

  I think that you have applied an unlinked auto STF to both images. The NSG image has a flatter background, and the nebula contains more contrast, so the auto STF has affected the image differently. 

  
  

  When stretching an image, the brightest parts of the image end up with reduced saturation. Since the STF applied to the two images differ, this is one reason why the color saturation looks different.

  
  

  For example, most of the Orion Nebula's real color should be a salmon pink due to the dominant Ha and Hb emissions. However, the central part of the nebula is so bright compared to the rest of the nebula, the aggressive stretch reduces the color saturation dramatically, so it often appears almost white. If the image was processed to only display the central region (and hence a much less aggressive stretch) the image would then display the real salmon pink color.

  
  

  Color in astrophotography is a tricky and complex area. You need to adjust the color balance and saturation in post processing.

  
  

  Regards, John

  [Edited]
• Michael Lehv

  November 2022

  John, thanks for your critique and explanation.  

  
  

  I understand what you are saying about the error in using the default STF unlinked auto-stretch images to make a judgment about NSG's results. I guess the real problem is my incorrect assumption that STF's unlinked auto-stretch will always display the optimally saturated image without further adjustment.. 

  
  

  I'm glad to understand that NSG works as intended and really appreciate your labors in making it available to all. 

  
  

  Thanks again,

  Michael