›

NSG script issues. Many, many images with a low weight!

MAURIZIO MOLLICA
in PixInsight
Hi Adam!
I am using NSG with 6888 luminance data downloaded from your site. I have already used NSG with my own data without having any particular problems.
In this case, however, if I try to use it on luminance, by choosing a better frame (n ° 310), many images have a weight of less than 0.50. I also decided to choose a "medium" frame but the result does not change. I detected some passages of clouds in the dataset and some changes in the focus, but nothing so important, in my opinion. Maybe I'm doing something wrong? Has anyone tried? 
Obviously the frames are calibrated and registered.
Thanks in advance!
Maurizio Mollica
6888_nsg_result.png
515 x 905 - 61K

Comments

  • BlockHeadBlockHead
    Correct. Normalization is a relative process. Changing the reference frame will, in general, not make a big difference. Without looking at your data, I cannot say if the answer is correct. It really should be though. 

    Make certain you are using version 1.3 like I am. (You should be if you are up to date with Pixinsight updates.)

    If could very well be you are seeing, perhaps for the first time, what the actual quality range of your data is! 
    The weights as calculated in ImageIntegration (default) have been wrong for years. So... is a frame in your set that says "0.2 weight" ... can you give a reason you do not believe it is true?

    -the Blockhead
  • MAURIZIO MOLLICA
    Thank you, Adam, for fast reply!

    First of all let me say that the work you are doing is simply fantastic, always clear and innovative!

    However, the data we are talking about is yours, downloaded from your site (ngc 6888 from skylive), so, in case you have time and desire, you could try nsg on that luminance data ...  

    In any case, I fixed by using "local normalization" instead of NSG on luminance and NSG (without any issues) on RGB data.

    My perplexity, however, remains, because with local normalization I could use all the frames that NSG would have discarded me instead. In fact, they do not seem to be so bad, in my opinion!

    I also finished the processing workflow, following your precious tips...
    Here is the result!

    Thanks for watching!

    - Maurizio Mollica
    Lha_rha_GB_finalcut.jpg
    2272 x 1687 - 3M
  • BlockHeadBlockHead
    A few screenshots demonstrating what you did would be helpful.
    Sometimes people make the mistake of mixing filters during NSG for example. 
    -the Blockhead
  • BlockHeadBlockHead
    Ok.. You are correct. 
    I looked at the Luminance data and there is a difference in quality. And guess what... for the first #%#!! time in PixInsight history you are getting the REAL truth about the quality of the frames. I will make a video on this- but NormalizeScaleGradient did it's job perfectly here. Absolutely sensational! It is weighting the images for what they are actually worth- something that PixInsight has not done properly in the past. Astrophotography is hard!!

    -the Blockhead
  • MAURIZIO MOLLICA
    Ok, Thank you Adam, once again!

    Then this is the harsh reality!

    I look forward to the new video then!

    - Maurizio Mollica
  • BlockHeadBlockHead
    I just finished it.
    Will publish today or tomorrow.
    -the Blockhead

  • BlockHeadBlockHead
    Please see the video here:

  • MAURIZIO MOLLICA
    Fantastic! 
    Thank you!

  • Roger Smith
    My experience with the NSG1.3 is that the output NWEIGHTS vary a lot more than I expected. My reference best gradient image was 1.0. The highest was 3.5, and the lowest was .36. That is almost 10x range. I had 100 light polluted images over 5 nights of varying conditions. As Adam said we are now seeing the real quality of our subs.

    The accept/reject slider is just that and John Murphy said it is up to each user to decide how much data to keep. On my above data it said over 50% of my images would not be used in integration. So I integrated with this default, and did it again with all 100 images. The 100 images result was lower in background noise, but only very slightly better in the galaxy (M101). Now my slider will be set to 1%.

    When my skill in imaging gets to be so bad that I should throw away 50% of my data, then I will hang it up and do something else!

    Now I am going to watch Adam's new video!

    Roger
  • BlockHeadBlockHead
    Yes...I hope my video clarifies the reasoning behind the threshold and how to approach the weights you get. Remember, the weighting is a relative measure within a set of data. Imagine all of the data could be acceptable...but the best is still twice as good as the worst... certainly in this case, you include most of the data.

    The above rarely happens though.... 

    -the Blockhead
Sign In or Register to comment.