Author Topic: Differences between Frame Adaption and Linear Fit  (Read 242 times)

Offline smccully

  • Newcomer
  • Posts: 20
    • View Profile
Differences between Frame Adaption and Linear Fit
« on: 2019 October 18 15:18:16 »
I was looking around trying to figure out what the difference between a Linear Fit and Frame Adaption actually was and was reading that they are supposed to be the same.
The problem is I have some images that have been pre-processed and i am trying to register them with FA turned on. But I end up with a large number of images  no signal or full of noise after registration. Doing a stand alone linear-fit on the per-registered image and reference image is fine. So there appears to be some difference between a FA and LF. Really i would like to understand what i really need to do to have Frame Adaption work.

Offline smccully

  • Newcomer
  • Posts: 20
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #1 on: 2019 October 18 15:21:16 »
I should have added that prior to the registration process, i have batched LinearFit the images against the same reference image used in StarAlignment and verified that all the LinearFit images are ok. Then when doing StarAlignment with FrameAdaption turned on, i still end up with a large number of images that fail frame adaption (i.e. no signal, or full of noise).

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4528
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #2 on: 2019 October 18 19:32:49 »
i think for anyone to help you'd have to post some example frames somewhere...

rob

Offline smccully

  • Newcomer
  • Posts: 20
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #3 on: 2019 October 18 20:52:31 »
Sure, I was just hoping somebody could explain the difference between Frame Adaption and Linear Fit.

Reference
https://www.dropbox.com/s/xpo8yovm20k4mon/Blue-1x-120s-26_c_a_f_a.xisf?dl=0
Linear Fitted to Reference
https://www.dropbox.com/s/5dwx1bgj41exdx7/Blue-1x-120s-0_c_a_f_a.xisf?dl=0
Registered with Frame Adaption
https://www.dropbox.com/s/chw06u848wjbn0t/Blue-1x-120s-0_c_a_f_a_r.xisf?dl=0

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4528
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #4 on: 2019 October 18 21:51:08 »
well... there is a lot going on here.

first is when you did linear fit, are the two images already registered to one another? linear fit can only be performed on identical (registered) images because it does a pixel-by-pixel comparison (and thus the pixels should have the same sky data in each.)

second is the calibration of 26_c_a_f_a seems bad - there is inverse vignetting and bright dust spots.

i guess third is there's not very much signal in any of these images, and the tracking/guiding does not look so hot in 0_c_a_f_a. it's probably not a good reference, generally speaking.

i also get weird results with frame adaptation and i do wonder if it has to do with the bad calibration in 26_c_a_f_a. the tooltip for frame adaptation does state that it is assumed that the illumination profile of both images is flat and that's definitely not the case here.

rob

Offline smccully

  • Newcomer
  • Posts: 20
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #5 on: 2019 October 19 10:28:49 »
Ignoring the Issues with the actual images, dust spots, eccentricity on the stars. It appears that I should only be doing LinearFit after image registration.

On a side note, i've redone the calibration on some of my images that was causing the reverse vignette. So thanks for pointing that out.

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4528
    • View Profile
Re: Differences between Frame Adaption and Linear Fit
« Reply #6 on: 2019 October 19 10:35:46 »
ok great... it would be interesting to see if there are any differences between LF and FA when the images have been calibrated well.

rob