Author Topic: FFT Registration  (Read 1984 times)

Offline Scooot

  • Member
  • *
  • Posts: 64
    • View Profile
FFT Registration
« on: 2016 January 03 02:52:14 »
Hi all, I'm new here & to PI.

I've taken 50 Luna lights and 25 bios, darks &, flats with my canon 450d and thought I'd try and stack them and process in pi. I followed a good tutorial and all went well until I'd debayered all the lights and was ready to align for stacking. However , being a moon image taken at 1/200 there were no stars to align on. So after doing a bit of googling I used the first debayered light as a reference,  added the others to the FFT registration script and ran it. The output box was greyed out but on completion it produced an image which I presumed was the equivalent of a stacked image. Could someone tell me if this is correct?
Also, was I correct to run the script on the debayered lights?
The resulting image was very dark, I couldn't see anything until I stretched it, and it still wasn' as light as I'd expected and the colour changed from the mostly grey of my original RAW images.
Any guidance very welcome  :)
Richard

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4530
    • View Profile
Re: FFT Registration
« Reply #1 on: 2016 January 03 15:09:03 »
correct everywhere. i think FFTRegistration can also write the aligned subs to disk.

it's totally normal for the images to look very dim - PI does not stretch the data on it's own. back of camera displays and various other programs always stretch DSLR data behind your back... but not PI.

rob

Offline Scooot

  • Member
  • *
  • Posts: 64
    • View Profile
Re: FFT Registration
« Reply #2 on: 2016 January 03 16:07:31 »
Many thanks Rob, good to know I'm on track. I didn't realise dslrs stretched the data to generate a viable image. That explains why they vanish when I select linear in canons DPP.

Slightly off topic but the stacked Luna image had a brown hue to it which I quite liked, but is different to most other Luna images I see. I tried colour calibrating it, but as there wasn't anything white to use as a reference the resulting image reverted to mostly grey. Difficult to know what colour the moon should be but as the debayer process on the clean lights used VNG I was expecting it to produce a realistic guess. Anyone any thoughts on this?
Richard

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4530
    • View Profile
Re: FFT Registration
« Reply #3 on: 2016 January 03 20:37:05 »
yep exactly - this is psychovisual in origin. a CCD is a linear device but the human eye/brain has a log response to light. in order to see what the eye sees with a CCD, the image needs to be nonlinearized.

just for fun after color calibrating the image, you can go in and pump the saturation like crazy. you should start to see various colors, which are apparently real - blues and oranges - which indicate differing quantities of different minerals/elements on the moon's surface.

http://cseligman.com/text/moons/mooncolor.htm

rob

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: FFT Registration
« Reply #4 on: 2016 January 03 22:35:23 »
rob

That was an interesting read. I decided to try just what was mentioned about the colored moon and this was what I came up with. This is just a single frame from my Nex-5. I think it agrees nicely with what they showed in the article. I just did a simple color calibration to the image.A simple stretch, a little decon, color saturation and ACDNR to the chroma channel.


Mike