›

WBPP : Don't understand what's wrong with calibration

2»

Comments

  • BlockHeadBlockHead
    It does not clarify. You need to show your WBPP configuration that produced the masterLight you sent me. Use the Diagnostics button to generate and show all WBPP panels. This is a configuration issue. 
    -the Blockhead
  • BlockHeadBlockHead
    Also. You need to do something. You need to create your masterdark again manually.
    Then you need to use WBPP and create the masterdark only. Just look the dark frames you used manually into WBPP. Do nothing else. Then run WBPP and create this masterdark. They will be the same. They have to be the same. If they are different there is a problem.

    -the Blockhead
  • Jean-François HERVIER
    Here is the WBPP configuration use with my own Masterdark :

    00_Bias.jpg
    2880 x 1620 - 305K
    01_Darks.jpg
    2880 x 1620 - 496K
    02_Flats.jpg
    2880 x 1620 - 475K
    03_Lights.jpg
    2880 x 1620 - 842K
    04_Calibration.jpg
    2880 x 1620 - 351K
    05_Post-Calibration.jpg
    2880 x 1620 - 297K
    06_Pipeline.jpg
    2880 x 1620 - 338K
  • Jean-François HERVIER
    Here is the WBPP configuration used with WBPP Masterdark :
    00_Bias.jpg
    2880 x 1620 - 305K
    01_Darks.jpg
    2880 x 1620 - 541K
    02_Flats.jpg
    2880 x 1620 - 469K
    03_Lights.jpg
    2880 x 1620 - 842K
    04_Calibration.jpg
    2880 x 1620 - 385K
    05_Post-Calibration.jpg
    2880 x 1620 - 300K
    06_Pipeline.jpg
    2880 x 1620 - 370K
  • Jean-François HERVIER
    And here are the 2 Masterdark :
    - One build manually with ImageIntegration
    - One build with WBPP

    So, I have a problem because they're not the same.

    Jean-Francois
    MasterDark_with_WBPP.jpg
    2121 x 1464 - 1M
    MasterDark_with_ImageIntegration.jpg
    2124 x 1467 - 657K
    ImageIntegration.jpg
    741 x 1622 - 186K
    01_Darks.jpg
    2880 x 1620 - 436K
    04_Calibration.jpg
    2880 x 1620 - 349K
  • BlockHeadBlockHead
    edited January 2024
    Well... we are certainly narrowing things down.
    So... you need to provide me with the raw dark frames. 
    (you may have given them to me... I just need to make certain we are using the same ones) 
    I will do exactly as you are doing (or what I think you are doing).
    Then we will find the answer.

    Thanks... we are getting close.
    -the Blockhead
  • Jean-François HERVIER
    Yes, I sent you my Dark raw files yesterday.
    They are in a zip file which also include the Masterdark I built with Imageinegration.
  • BlockHeadBlockHead
    At the moment I just cannot seem to catch up.
    So I have a suggestion. You should open your WBPP process container from the log folder.
    Then you can look at the ImageIntegration instance (setup and configuration) for the darks that are integrated.

    You should be able to directly compare that setup with your manual ImageIntegration. 

    WBPP and you are using the exact same process. The only difference can be in the configuration you used.

    Review this video on the WBPP process container.
    -the Blockhead
  • Jean-François HERVIER
    The only difference is the Rejection algorithm.
    WBPP used "Winsorized Sigma Clipping" when I used "Linear Fit Clipping" (see attached images).

    My understanding was that above 25 files, "Linear Fit Clipping" was a better choice.


    Imageintegration_WBPP(left)_vs_Myself(right).jpg
    1482 x 3000 - 1M
    Masterdark_WBPP(above)_vs_Myself(below).jpg
    2117 x 2950 - 2M
  • BlockHeadBlockHead
    edited January 2024
    I do not think we are done yet.

    You need to manually stack the darks. 
    One time using WSC and the other using LFC. 
    Make certain you can reproduce the difference you see. 

    Look at the % of rejection. I am guessing that LFC is doing more aggressive rejection. Then when you display with AutoSTF you will have a difference appearance because LFC is rejecting more at the low end. 

    I predict that if you adjust the rejection strength of WSC you will achieve similar (near identical) result of LFC. The default WSC values are not too aggressive. 

    In part, I want to make clear ... the reason the images look so different is more to do with the display than anything else. You are getting different amounts of rejection..and this affects the display. But you can likely do a statistical analysis of the values and find the images are similar..and display them in similar ways using the STF.

    -the Blockhead. 
  • Jean-François HERVIER
    I manually stacked with WSC and LFC and, yes, the rejection of LFC is more aggressive.
    But I'm not enough comfortable with the parameters of pixel rejection to have the same result with the two algorithms.

    So, if I understand correctly, the conclusion is, except for the rejection of WSC which is less aggressive than FSC (with the default parameters), the two Masterdarks are identical and the visual difference is only the consequence of a different stretching.
    Is it correct ?
  • BlockHeadBlockHead
    Well..they are not identical...since one has greater rejection than the other. But given the same settings..you will get the same results. The greater rejection of one made the image look different than the other (because you have different histograms) when displaying with AutoSTF.

    Yes, I think we have a conclusion. 
    I am  perhaps surprised the darks are so noisy that the rejection method mattered that much.
    But that is another rabbit hole...
    -the Blockhead
Sign In or Register to comment.