Light Artifact in all my subs

Discussion in 'STX and STXL Series Cameras' started by Mark McComiskey, May 13, 2021.

  1. Mark McComiskey

    Mark McComiskey Standard User

    Joined:
    May 16, 2019
    Messages:
    54
    Thanks for the quick response.

    The camera does have a fair few hot columns, but they calibrate out without an issue when I run the standard calibration process in Pixinsight (including Cosmetic Correction, which I assume does more or less what Remove Bad Pixels does).

    The camera is quite new, and lightly used. I bought it Jan 2020 and started imaging Nov 2020, always from my backyard. Serial number is: x19110001.

    As to the flats, I use an Alnitak flipflat, and take the flats long before sunrise (see photo of set up above), and there is no light source near the scope. I uploaded some raw flats to share a couple of posts back, if that helps, but when I look at them, I can't see anything that would cause the issue. They seem about as expected.

    I recognize that you don't weigh in on processing issues, but might you give me some guidance on what to look for to verify if this is a flat issue or not? I would be delighted to find out that I simply needed to adjust my flat technique, but the ripple is so consistent across so many different targets taken over a several month period, that I worry it might be something else. I am happy to do the leg work and report back, it is just that I have exhausted the list of things that i know to look for.
     
  2. Doug

    Doug Staff Member

    Joined:
    Sep 25, 2014
    Messages:
    8,526
    It has to be the flat. What can the camera do that would cause that? If there was a fixed "glow" in every frame, it would be present in the darks and subtract out. If there was a sensitivity variation across the chip, it would be consistent between the lights and flats and would flat out. If it's so very consistent as you say then calibration would remove it - it has to.
     
  3. Mark McComiskey

    Mark McComiskey Standard User

    Joined:
    May 16, 2019
    Messages:
    54
    I'm not arguing. I am new enough to this hobby that I know I still have a lot to learn. I do want to run things down definitively, though. I actually have all the raws for the flats. What should I look for in the flats to confirm that the issue is with the flats?
     
  4. William B

    William B Cyanogen Customer

    Joined:
    Jan 8, 2015
    Messages:
    402
    Location:
    Christchurch, Dorset UK
    Mark.

    In addition to the comments and help from Doug.

    If not already set, use a higher dither setting to better compensate for the permanent hot and cold columns on the left side of the sensor, use a random dither of at least two pixels between subs.

    As you are using PixInsight, attached is the result of a basic calibration and integration workflow in PixInsight using your supplied reduced data set.

    Calibration and Integration workflow followed exactly that described by Warren A. Keller in his book "Inside PixInsight" (Springer, 1st edition, ISBN 978-3-319-25680-1).

    Cosmetic Correction was used to deal with the hot and cold columns on the sensor left side using a reference dark frame plus auto detect.

    During integration of the lights, pixel rejection (2) settings for Winsorised Sigma Clipping were adjusted to 1.200 for Sigma High and 4.000 for Sigma Low to deal with the variable warm column in the image centre.
    The highest SNR frame as determined by the SubFrameSelector module was used for the reference frame during integration.

    Note that it would have been better to have created a pixel defect list for the central variable warm column (and the left side hot and cold columns) in the Cosmetic Correction module and run with closer to default settings for Sigma High and Sigma Low in the Pixel Rejection (2) selection of the Image Integration module but this was just a run through to see if anything obvious appeared in your data and I did not have the time to create the defect list for Cosmetic Correction.
    Had a defect list been used in Cosmetic Correction and closer to default settings for Sigma High and Sigma Low in Pixel Rejection (2) with Winsorised Clipping in Image Integration then the noise level would have been lower in the final image allowing a higher stretch.

    The final step was to apply a quick rough-and-ready DBE instance to remove the residual linear gradient across the image.

    I could not see any issues with your flats, darks or bias frames and as you can see from the result of processing in PixInsight with this reduced data set the resulting image has calibrated perfectly.
    (There was some brightening in the bottom right corner in some of the subs that is not present in the darks and with nothing to indicate to the contrary this may be from an extraneous light source since it is not present in all the frames).

    With a larger data set and following the same workflow steps in PixInsight using the recommended module settings during each step as suggested in Warren's book you should not have any problems successfully calibrating your camera data.

    HTH.

    William.

    Basic_plus_DBE.jpg
     

    Attached Files:

    Last edited: Jun 8, 2021
    Tim Povlick likes this.
  5. Mark McComiskey

    Mark McComiskey Standard User

    Joined:
    May 16, 2019
    Messages:
    54
    Okay - thanks much for the continued engagement on this. I just ran WBPP on the same data set and then DBE, and still have the artifact. I will run the calibration process entirely manually, using the settings in Inside PI. If that cures the issue, I will try to identify what WBPP is doing that differs from the manual process. I will report back either way.
     
  6. William B

    William B Cyanogen Customer

    Joined:
    Jan 8, 2015
    Messages:
    402
    Location:
    Christchurch, Dorset UK
    Mark.

    Give WBPP a miss, it is a blunt instrument that provides virtually no feedback when you have to deal with problems like this.

    Until you know how the various settings in WBPP change the resulting image it's best to stick to the individual modules, manually process and inspect the results of each module as you work the data through the sequence, only then will you know how to configure the settings of WBPP, and even then , I still think that the result of manual pre-processing in PI beats WBPP every time as no two data sets are alike and each data set requires slightly different settings to achieve the very best results.

    This same issue constantly surfaces on the PixInsight forum where new users struggle with WBPP and the advice given by the software authors is always the same, WPBB was not meant to replace manual pre-processing, only to give a preliminary image with which to judge what steps are required to achieve a satisfactory final image.

    William.
     
    Last edited: Jun 8, 2021
    Tim Povlick likes this.

Share This Page