Author Topic: Curved mesh artifacts  (Read 353 times)

Offline Denis

  • Newcomer
  • Posts: 2
    • View Profile
Curved mesh artifacts
« on: 2019 August 11 09:43:23 »
Sorry if the topic has already been addressed, and I will raise it again, but I have not yet found a solution.
Through the batch preprocessing, I calibrate with darks and biases, and integrate 5 frames, for 900 seconds. When I stretch the histogram, dips appear in the form of a curved grid.
As I understand it, these are alignment artifacts.
How can I get rid of them?

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4550
    • View Profile
Re: Curved mesh artifacts
« Reply #1 on: 2019 August 11 11:36:36 »
they are probably calibration artifacts; I assume this is a narrowband image and as such the background is probably not much different than a dark. what happens is the master dark is a more precise measurement of the dark current in the sensor, and when you subtract that master dark from a subframe, there are pixels in the subframe where the dark current was lower than the average and as such the calibrated pixel would have a negative value... which gets clamped to 0 by imagecalibration.

so you then have these patterns of 0s in your calibrated frames, and when they are registered, tend to give you a moire pattern like you see in your image.

to see if this is your problem, you can open one of the calibrated subframes and apply the pixelmath expression iif($T==0,1,0) and you'll get an image where every pixel in the original image which is 0 is marked with a bright pixel. in a well-calibrated image you should get an all-black image after applying that expression... if you see a bunch of pixels lit up then you probably have the problem I am describing.

to get around this, you can add an output pedestal when calibrating the lights, try 100 DN or 200 DN or so and see what happens.

rob

Offline Denis

  • Newcomer
  • Posts: 2
    • View Profile
Re: Curved mesh artifacts
« Reply #2 on: 2019 August 11 22:06:19 »
Thanks a lot Rob!  :)
I tried to calibrate with a different pedestal (from 0 to 300), and indeed, with its value of 100, I got the minimum number of bright points in the PixelMath. After that, I added the images and the grid disappeared!
Thanks again!

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4550
    • View Profile
Re: Curved mesh artifacts
« Reply #3 on: 2019 August 12 12:06:06 »
great - glad that worked. I've had this same problem with narrowband images taken from an extremely dark site...

rob