›

Prominent artifacts with OSC NB images when applying Local Normalization

Michael Lehv
in PixInsight
Hi Adam,

I'm having problems with OSC narrow band integrated images after applying Local Normalization (LN) as shown below. There is always a horizontal blurred band of lightness or colors occupying the top or bottom 20% of the image. The reference frames, whether manually selected or by WBPP (single or averaged) look fine. No errors are reported in the logs. This happens regardless of the exact OSC NB filter used e.g., dual or triband and is unrelated to its specifications.  

The problem occurs whether LN is being applied manually with the PI Process or by WBPP. It does not seem to occur with RGB broadband images. Occasionally, if a NB image is very bright and contrasty (more like a broad band image), the banding won't occur. 

I have tried adjusting numerous parameters in both LN and Image Integration with no success. I haven't seen anyone else reporting this so I have to assume there is something very basic that I am overlooking. 

Thanks for looking at this issue.

Michael


imageimageimage

Comments

  • BlockHeadBlockHead
    I am not certain I understand... why is there white around the borders (especially of the LN reference?). 
    This can't be good...
    -the Blockhead
  • Michael Lehv
    Hi Adam, 

    I thought the white borders were from WBPP's Autocrop feature so I paid them no attention despite them staring me in the face. I certainly can see how the white borders on the LN Reference image might adversely effect the images output from LN. 

    I've gone back and the white borders seem to appear very early in the workflow calibration stage. I've also determined that the particular camera used is not the culprit.  

    I'll get back to you as soon as I isolate the calibration issue.

    Thanks.

    Michael


  • BlockHeadBlockHead
    Thanks for following up.
    -the Blockhead
  • Michael Lehv
    Hi Adam, 

    I believe I've solved the problem. Your comment about the white borders was the big clue. As I mentioned above, I simply was ignoring the white borders on the integrated images and then noted that they were present even on the calibrated (xxx.xisf_c) subs. 

    Apparently my QHY128Pro and QHY410C (the BSI version of the 128) cameras that I have been using for years both leave a black border on the top of all the downloaded lights, biases, darks and flats (see below). So, I went back and using Image Container, quickly cropped all the FITS image subs and the master darks and flats. 

    Then I ran WBPP with LN enabled and the previous blurred light and colored 20% bands on the LN images were gone!

    The next question was why were the white borders appearing? The answer seems to be that in the Calibration Process, the black borders of the FITS image subs were being divided by the near 0 of the black borders of the master flat resulting in a border approaching nearly 1.0 white. Omitting the master flat from the calibration process, leaves the original black borders intact. 

    To test this theory, I cropped off the borders of the master flat and resized it to match the size of the uncropped original image FITS and darks. I then re-ran WBPP with LN enabled. Again, there were no LN artifact bands. However, I wouldn't recommend this approach because resizing the flats really makes a mess of the images. 

    I use The Sky X for capturing images and calibration files so I'll have to see if the black borders are from TSX or the cameras. 

    Thanks again for spotting the white borders as the culprits.

    Michael


    imageimageimageimage
    FITS Image.JPG
    1054 x 765 - 144K
    Master Dark.JPG
    1048 x 754 - 157K
    Master Flat.JPG
    1048 x 759 - 174K
    Calibration without master flat.JPG
    1022 x 753 - 68K
  • Michael Lehv
    Hi Adam,

    As a practical matter, the root of the problem is the black border around the master flat and an (almost) divide by zero in the Calibration Process producing the troublesome white borders. So, by changing the master flat black border to white before calibration by applying "iif(($T < 0.01),1,$T)" the calibrated FITS borders remain black and can be cropped from the master light after WBPP completes. 

    This works well and I'm now able to use Local Normalization without any artifact. 

    Thanks again,

    Michael`
  • BlockHeadBlockHead
    Why is there a border around the (master) flat?

    -the Blockhead
  • Michael Lehv
    Hi Adam, 

    A clear night and I captured a few images with Sharpcap. No borders! So the borders are coming from The Sky X and/or the ASCOM driver. I will figure it out. 

    Since I started imaging, I've always used TSX for image acquisition. So, I've come to expect the borders. It wasn't until I started using Local Normalization that I've run into this problem. And I've spent a few months trying to figure it out.

    Now I know.

    Michael
  • BlockHeadBlockHead
    Got it!
    Thanks for following up!
    -the Blockhead

Sign In or Register to comment.