›

NormalizeScaleGradient

John Gill
in PixInsight
Hey Adam,

I watched your tutorials on NormalizeScaleGradient, excellent as usual! So, I used my standard workflow to ImageIntegration and also ran a workflow incorporating the NormalizeScaleGradient. Visually the images without the NormalizeScaleGradient were better. The data was of the Prawn Nebula (IC 4628) in Ha, Sii and Oiii. The images were not great to start with, mostly focus issues and some bad seeing. If you want to look at the .XISF or .JPG files, I can upload to my GoogleDrive.

Regards
John

Comments

  • John Gill
    ... a further update...
    So last night I imaged the same object, but fixed all the focus and other issues. Anyway, I reprocessed with standard ImageIntegration and with the NormalizedScaleGradient. My results were
    Ha - not much difference between the images
    Oiii - all detail lost and a "mist" covers the entire image
    Sii - as above

    I don't think it makes a difference, but I used a Ha image as the reference for all three filters.

    John
  • BlockHeadBlockHead
    Hi John,

    Perhaps I didn't make that clear enough in my tutorial. This is done on a per filter basis. You certainly do not mix filter data. That makes a very big difference. I know I mentioned this..but I do not think I repeated this many times. 

    Regarding the Ha (since this one filter set was done correctly)- you will *not* see a big difference if the data is similar.  The weights should be correctly calculated- and you can look at the numbers to see potential differences if there are differences in the quality of data from frame to frame (as I show).

    -the Blockhead

  • John Gill
    Hi Adam,

    I do process each filter separately. The results for the Ha were very similar to the standard ImageIntegration, but the Oiii and the Sii turned out badly after using the NSG script separately.

    John


  • BlockHeadBlockHead
    edited June 2021
    You need to chose a reference in each filter (an OIII reference for the OII data... an SII reference for the SII data). Above you said you used the Ha as a reference for all filters. 
    That is the reason why it did not work.
    I hoping for an "OOooooh" response from you. 
    :)

    By the way, it isn't the appearance alone that matters. It is the calculation of the weights (the numbers). If the images are similar- indeed there will not be much of a difference. But when the data is variable, and this is the norm for most people, then NSG will calculate proper weights. 

    -the Blockhead
  • John Gill
    Ok, here is the BIG  "OOooooh" ... multiple issues, mostly not paying enuff attention in class, assuming too much and jumping the gun....

    Anyway, I redid the entire process, remembering to use the filter reference and not the StarAlignment reference, and then making sure "No Normalization" was selected in both places during ImageIntegration.  Overall the 3 resulting images were a little better than the standard ImageIntegration.

    Many thanks for your help and guideance.
    John
Sign In or Register to comment.