Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

How to remove Newton Rings in post processing?

  • Please log in to reply
19 replies to this topic

#1 Stargazer3236

Stargazer3236

    Aurora

  • ****-
  • topic starter
  • Posts: 4,920
  • Joined: 07 Aug 2010
  • Loc: Waltham, MA

Posted 30 July 2020 - 11:14 AM

Any ideas?


  • philmor56 likes this

#2 Tapio

Tapio

    Aurora

  • -----
  • Posts: 4,823
  • Joined: 24 Sep 2006
  • Loc: Tampere, Finland

Posted 30 July 2020 - 11:42 AM

Flats might help.

But using tilt adapter might be better solution.


  • Paul Hyndman likes this

#3 sg6

sg6

    Voyager 1

  • *****
  • Posts: 10,184
  • Joined: 14 Feb 2010
  • Loc: Norfolk, UK.

Posted 30 July 2020 - 11:48 AM

Separate the 2 surfaces causing the Newtons rings. They are caused by interference between a lens and a flat surface, usually has to have the lens resting on the flat surface. Also only really met them with monchromatic light.

 

Next question is: Are we talking Airy rings not Newton rings?



#4 chemman

chemman

    Apollo

  • -----
  • Posts: 1,027
  • Joined: 21 Jun 2011
  • Loc: Colorado

Posted 30 July 2020 - 01:32 PM

Separate the 2 surfaces causing the Newtons rings. They are caused by interference between a lens and a flat surface, usually has to have the lens resting on the flat surface. Also only really met them with monchromatic light.

 

Next question is: Are we talking Airy rings not Newton rings?

Don't know if Stargazer has this but here is one that I noticed "Newton Rings" on.  I shot this one with the PGR.

Attached Thumbnails

  • newton rings-.jpg

Edited by chemman, 30 July 2020 - 01:58 PM.


#5 MalVeauX

MalVeauX

    Cosmos

  • *****
  • Posts: 9,589
  • Joined: 25 Feb 2016
  • Loc: Florida

Posted 30 July 2020 - 04:11 PM

Heya,

 

Flat calibration, either a flat made during acquisition, or a pseudo flat done in post processing. Ideally remove it in acquisition (this is covered in my tutorials) either via real time flat overly in FireCapture via defocus or via a diffuser. Ideally just stop them from happening with a tilt adapter all together (especially for a full disc).

 

If you can attach the file you want to see the newton rings processed out of, I can attempt it purely in post and if successful share a step by step how to via Photoshop CS5.1 (old version so should work in anything and may even work in Gimp with similar naming).

 

Very best,


Edited by MalVeauX, 30 July 2020 - 04:18 PM.

  • ilan_shapira likes this

#6 philmor56

philmor56

    Messenger

  • -----
  • Posts: 452
  • Joined: 17 Feb 2008
  • Loc: Bolton, ON

Posted 01 August 2020 - 08:00 AM

Hi there:

I'm having the same frustrations. I've tried using flats (as per Martin's and other's tutorials but to no success (long story)), and I'm waiting on my tilter.

 

This is a recent capture with my 2x barlow. I'd love to find a way to get rid of the rings in post. I've tried but my understanding of P/S is still limited.

Any advice, as always, is appreciated!

 

prt-imp-ps-zm-mon-01.jpg

 

Thanks in advance.


  • Stargazer3236 likes this

#7 MalVeauX

MalVeauX

    Cosmos

  • *****
  • Posts: 9,589
  • Joined: 25 Feb 2016
  • Loc: Florida

Posted 01 August 2020 - 08:31 AM

Hi there:

I'm having the same frustrations. I've tried using flats (as per Martin's and other's tutorials but to no success (long story)), and I'm waiting on my tilter.

 

This is a recent capture with my 2x barlow. I'd love to find a way to get rid of the rings in post. I've tried but my understanding of P/S is still limited.

Any advice, as always, is appreciated!

 

Thanks in advance.

Did you do a flat calibration during acquisition? Ie, either defocus or with a diffuser of some kind? This is key if you do not have a tilt adapter (and still important even with a tilt adapter due to gradients).

 

Very best,



#8 philmor56

philmor56

    Messenger

  • -----
  • Posts: 452
  • Joined: 17 Feb 2008
  • Loc: Bolton, ON

Posted 01 August 2020 - 08:53 AM

Did you do a flat calibration during acquisition? Ie, either defocus or with a diffuser of some kind? This is key if you do not have a tilt adapter (and still important even with a tilt adapter due to gradients).

 

Very best,

Thanks for the reply Marty.

Yeah, I've tried the defocus flat multiple times. Every time I apply the flat, as per your video, the flat "explodes" around the capture in a kind of corona effect, making processing impossible... 

Maybe I'm missing something here.

 

(ZWO is apparently modifying their adaptor and the back-orders are piling up, so I've been waiting on that. My dealer is hoping his next shipment may have them)

Thanks again.


  • MalVeauX likes this

#9 PDB

PDB

    Vostok 1

  • -----
  • Posts: 130
  • Joined: 18 Nov 2016

Posted 01 August 2020 - 10:12 AM

Hi,

 

to come back on the question about post-processing.

 

Yes there are tricks produce good results in most of the cases, if NR's are present on the images.

 

You can look at these info on CN: https://www.cloudyni...es-from-images/

 

en another one in Dutch at https://www.astrofor...em-en-oplossing (maybe google can try to translate)

 

Paul


  • philmor56 likes this

#10 Paul Hyndman

Paul Hyndman

    Apollo

  • -----
  • Posts: 1,384
  • Joined: 13 Jul 2004
  • Loc: CT Shoreline

Posted 01 August 2020 - 08:52 PM

It's been many, many years since my solar imaging days but... most of my post-processing efforts to deal with NR were lackluster at best (flat fielding actual or pseudo flat fielding) when sufficiently applied to remove traces of the rings, also removed detail in hi-res (SBIG STL11K w/90/90/30 Coronado setup). The best solution for me, as mentioned above by Tapio, had been to use a tilt adapter. IMO, it is easier to start with the best raw images rather than deal with uh-ohs later on.

 

I no longer maintain a website, but FWIW some of my notes on NR can be found here.

 

https://web.archive....tons-rings.html


  • philmor56 likes this

#11 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 27 April 2021 - 05:05 PM

Hi there:

I'm having the same frustrations. I've tried using flats (as per Martin's and other's tutorials but to no success (long story)), and I'm waiting on my tilter.

 

This is a recent capture with my 2x barlow. I'd love to find a way to get rid of the rings in post. I've tried but my understanding of P/S is still limited.

Any advice, as always, is appreciated!

 

attachicon.gifprt-imp-ps-zm-mon-01.jpg

 

Thanks in advance.

This took me 25 seconds in ImageJ using FFT blob removal techniques. Not kidding. Literally less than 30 seconds. This technique is BEAUTIFUL!

 

NR FFT Removal.jpg

 

NR removed.jpg

JT


Edited by jt_3232, 27 April 2021 - 05:11 PM.

  • philmor56 and torsinadoc like this

#12 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 27 April 2021 - 05:09 PM

Don't know if Stargazer has this but here is one that I noticed "Newton Rings" on.  I shot this one with the PGR.

I played with your image as well. Here are the results that took me less than 1 minute to do.

 

NR 2 Removed.jpg

 

Jaxon



#13 philmor56

philmor56

    Messenger

  • -----
  • Posts: 452
  • Joined: 17 Feb 2008
  • Loc: Bolton, ON

Posted 22 May 2021 - 08:45 AM

This took me 25 seconds in ImageJ using FFT blob removal techniques. Not kidding. Literally less than 30 seconds. This technique is BEAUTIFUL!

 

attachicon.gifNR FFT Removal.jpg

 

attachicon.gifNR removed.jpg

JT

 

I played with your image as well. Here are the results that took me less than 1 minute to do.

 

attachicon.gifNR 2 Removed.jpg

 

Jaxon

 

 

Hi Jaxon

 

Wow, those are really good results!

So, after seeing your reply I eventually found and downloaded ImageJ.

Um... really above my level of knowledge at this stage and age ! shocked.gif gramps.gif

I dropped an image into the program and played around a bit and I found the FTT function and that "flat" window popped up but that's about where I lost it. I spent a good half hour trying to do something and, well something's not clicking.

 

How about a quick workflow for that 25 second fix. Maybe then I can figure out what is being done.

 

If this works as good as it seems you are indeed a life saver!

Thanks again. waytogo.gif

 

Cheers



#14 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 22 May 2021 - 11:35 PM

Hi Jaxon

 

Wow, those are really good results!

So, after seeing your reply I eventually found and downloaded ImageJ.

Um... really above my level of knowledge at this stage and age ! shocked.gif gramps.gif

I dropped an image into the program and played around a bit and I found the FTT function and that "flat" window popped up but that's about where I lost it. I spent a good half hour trying to do something and, well something's not clicking.

 

How about a quick workflow for that 25 second fix. Maybe then I can figure out what is being done.

 

If this works as good as it seems you are indeed a life saver!

Thanks again. waytogo.gif

 

Cheers

Josh (jbalsam) did a great job here https://www.cloudyni...es-from-images/ which was posted above. However, I think it can be simpler than what he does.

 

1.) Open ImageJ (by clicking the Windows search bar and typing "imageJ" and clicking the result)

 

2.) Open the photo using File->Open and navigating to the image.

 

3.) Calculate the FFT of the image by going Process->FFT->FFT

 

2.jpg

 

4.) The centered log-magnitude stretched FFT of the original image is shown, zoom in towards the center using Ctrl+scroll wheel

 

3.jpg

 

5.) Look for the odd disparities. Useful Aside: The FFT gives you back the image transformed into the frequency domain. If you invert this back, you get the exact same image back. The frequency domain shows how the image is composed of different frequencies. In the centered, log magnitude FFT, the very center is frequency 0 or the "DC" (like direct current vs AC), while the further away from the center, the higher the frequency. The 0 frequency is the average value of the entire image, which is usually a very very large number, so it is very bright. The rest of the FFT tends to be relatively smooth, depending on the image content. If you had a large picture of a striped shirt, you might think you'd see a specific frequency of how often those stripes occur and you would be right! However, in the FFT, the frequency is always rotated 90 degrees relative to the actual image's frequency. So, if his shirt had very consistent stripes going up and down, you would see a large spike in the FFT close to the center (but not on the center) that is on the left and right of the center. (note, up and down stripes results in spikes on the FFT on the left and right side). In our image, we see these big frequency stripes going left to right. So, we should expect a spike to occur above or below the center. Indeed, we do see that spike. Note: the stripes don't necessarily always go up and down or left and right, they could be at angle, and the FFT would show the corresponding spikes oriented 90 degrees from the image's stripes.

 

6.) Drag and select a box around the spikes and hit the delete key to remove the spike. (Removing the spike in the FFT will remove that specific frequency component in the actual image, which is what we want).

 

4.jpg

 

7.) Run the Inverse FFT by going Process->FFT->Inverse FFT.

 

5.jpg

 

8.) Inspect the resulting image. Satisfied? Yes- Save it and show it off!  No- Try making bigger (or smaller?) boxes around the frequency spikes.

 

8.) Voila!

 

Let me know if this was helpful!

 

Take care,

 

Jaxon


  • R Botero, chemman, MalVeauX and 1 other like this

#15 philmor56

philmor56

    Messenger

  • -----
  • Posts: 452
  • Joined: 17 Feb 2008
  • Loc: Bolton, ON

Posted 23 May 2021 - 07:32 AM

Josh (jbalsam) did a great job here https://www.cloudyni...es-from-images/ which was posted above. However, I think it can be simpler than what he does.

 

1.) Open ImageJ (by clicking the Windows search bar and typing "imageJ" and clicking the result)

 

2.) Open the photo using File->Open and navigating to the image.

 

3.) Calculate the FFT of the image by going Process->FFT->FFT

 

attachicon.gif2.jpg

 

4.) The centered log-magnitude stretched FFT of the original image is shown, zoom in towards the center using Ctrl+scroll wheel

 

attachicon.gif3.jpg

 

5.) Look for the odd disparities. Useful Aside: The FFT gives you back the image transformed into the frequency domain. If you invert this back, you get the exact same image back. The frequency domain shows how the image is composed of different frequencies. In the centered, log magnitude FFT, the very center is frequency 0 or the "DC" (like direct current vs AC), while the further away from the center, the higher the frequency. The 0 frequency is the average value of the entire image, which is usually a very very large number, so it is very bright. The rest of the FFT tends to be relatively smooth, depending on the image content. If you had a large picture of a striped shirt, you might think you'd see a specific frequency of how often those stripes occur and you would be right! However, in the FFT, the frequency is always rotated 90 degrees relative to the actual image's frequency. So, if his shirt had very consistent stripes going up and down, you would see a large spike in the FFT close to the center (but not on the center) that is on the left and right of the center. (note, up and down stripes results in spikes on the FFT on the left and right side). In our image, we see these big frequency stripes going left to right. So, we should expect a spike to occur above or below the center. Indeed, we do see that spike. Note: the stripes don't necessarily always go up and down or left and right, they could be at angle, and the FFT would show the corresponding spikes oriented 90 degrees from the image's stripes.

 

6.) Drag and select a box around the spikes and hit the delete key to remove the spike. (Removing the spike in the FFT will remove that specific frequency component in the actual image, which is what we want).

 

attachicon.gif4.jpg

 

7.) Run the Inverse FFT by going Process->FFT->Inverse FFT.

 

attachicon.gif5.jpg

 

8.) Inspect the resulting image. Satisfied? Yes- Save it and show it off!  No- Try making bigger (or smaller?) boxes around the frequency spikes.

 

8.) Voila!

 

Let me know if this was helpful!

 

Take care,

 

Jaxon

 

Wow... !

 

Voila indeed! waytogo.gif

Got it, thanks for clicking on the light bulb! 

Even with my tilter I've noticed some ringing here and there and this is a wonderful fixer.

Thanks again for teaching this old dog a new trick! bow.gif

 

Cheers



#16 chemman

chemman

    Apollo

  • -----
  • Posts: 1,027
  • Joined: 21 Jun 2011
  • Loc: Colorado

Posted 23 May 2021 - 07:51 AM

Wow Jaxon that is a great explanation.   I will definitly be giving that work in ImageJ a try.  I shot 200 frames yesterday complete with Newtonian rings. The flat was near useless but had to have the event that was unfolding at the time. 

 

Your process Jaxon looks to be specific for each image so even if I can get the process down to 1 minute my shot will take more than 3 hours of processing.  Is there a Batch switch on ImageJ?  

 

Thanks for the tutoring.

Chuck



#17 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 23 May 2021 - 09:57 AM

Wow Jaxon that is a great explanation.   I will definitly be giving that work in ImageJ a try.  I shot 200 frames yesterday complete with Newtonian rings. The flat was near useless but had to have the event that was unfolding at the time. 

 

Your process Jaxon looks to be specific for each image so even if I can get the process down to 1 minute my shot will take more than 3 hours of processing.  Is there a Batch switch on ImageJ?  

 

Thanks for the tutoring.

Chuck

Hi Chuck,

 

I actually just did my semester Master's DIP course project looking at the automatic removal of Newton's rings. This turned out to be very difficult, and the only good way I could make this work live was using an ideal band stop filter. Needless to say, for best results, it is best to do it manually. 

 

For a video, I do not know if there are scripts on ImageJ or how to write one, but it shouldn't be hard to make one do that.

 

For your video, you might consider making a finalized image and then apply NR removal. This would be the simplest way and should still work fine. In fact, it might be easier to see the NR spikes on the FFT if you wavelet sharpen the image first. However, I can only speculate as I only worked with single processed images.

 

Please ask any other questions you may have. 

 

Jaxon



#18 chemman

chemman

    Apollo

  • -----
  • Posts: 1,027
  • Joined: 21 Jun 2011
  • Loc: Colorado

Posted 23 May 2021 - 08:40 PM

Heya,

 

Flat calibration, either a flat made during acquisition, or a pseudo flat done in post processing. Ideally remove it in acquisition (this is covered in my tutorials) either via real time flat overly in FireCapture via defocus or via a diffuser. Ideally just stop them from happening with a tilt adapter all together (especially for a full disc).

 

If you can attach the file you want to see the newton rings processed out of, I can attempt it purely in post and if successful share a step by step how to via Photoshop CS5.1 (old version so should work in anything and may even work in Gimp with similar naming).

 

Very best,

Marty here is one I am working with today.  I tried a defocused and a trashbag flat but the little rings remained.  Sometimes I can use up to 100 flats and get it to work for a while but it seems like after a while they come back.  I have used a tilt adapter but not spent enough time adjusting it to get satisfactory results. That will be coming soon.  I am going to work with the 2" barlow I got  from Harry Siebert.  It has a T2 connection on the top for the camera.  I have a Baader clicklock 1.25" adapted to T2 for visual.  These are Harry's Focus adapter series and have 110mm distance a T2 extension tube needs to occupy.  I am going to adapt a JMI electric focuser in that 110mm path to move the camera closer and further and hopefully get a some +- 2x adjustment of amplification from the telecentrics. Marty, Should the tilt adapter be directly attached to the camera or attached to the Barlow T2 100mm up the light path?  

 

 

Side question for you Marty.  The scope I am using LS100MT is 714mm focal length but with the two etalons it is extended by about 100mm to 814mm, is that the new focal length or should my calculations still use 714mm?

 

Jaxon, I was able to do a pretty good job of removing from a flat I shot with distinct Newtonian rings.  But when I tried to find the "Spikes" you mention in step 6 of post #15 I obviously can't locate them properly.  I ended up with a big blurry image when I did an Inverse FFT.  Got some suggestions?   The image I am working with is the same one for Marty.

 

Chuck

Here:

nr1qy3d.jpg


Edited by chemman, 23 May 2021 - 08:45 PM.


#19 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 25 May 2021 - 10:45 PM

Marty here is one I am working with today.  I tried a defocused and a trashbag flat but the little rings remained.  Sometimes I can use up to 100 flats and get it to work for a while but it seems like after a while they come back.  I have used a tilt adapter but not spent enough time adjusting it to get satisfactory results. That will be coming soon.  I am going to work with the 2" barlow I got  from Harry Siebert.  It has a T2 connection on the top for the camera.  I have a Baader clicklock 1.25" adapted to T2 for visual.  These are Harry's Focus adapter series and have 110mm distance a T2 extension tube needs to occupy.  I am going to adapt a JMI electric focuser in that 110mm path to move the camera closer and further and hopefully get a some +- 2x adjustment of amplification from the telecentrics. Marty, Should the tilt adapter be directly attached to the camera or attached to the Barlow T2 100mm up the light path?  

 

 

Side question for you Marty.  The scope I am using LS100MT is 714mm focal length but with the two etalons it is extended by about 100mm to 814mm, is that the new focal length or should my calculations still use 714mm?

 

Jaxon, I was able to do a pretty good job of removing from a flat I shot with distinct Newtonian rings.  But when I tried to find the "Spikes" you mention in step 6 of post #15 I obviously can't locate them properly.  I ended up with a big blurry image when I did an Inverse FFT.  Got some suggestions?   The image I am working with is the same one for Marty.

 

Chuck

Here:

 

Sorry for the delayed response, I've been moving.

 

So, I'm having a hard time figuring out what exactly the Newton rings are here. Is it the 3-4 almost vertical lines right underneath the solar spot/flare? If so, that is a real challenge for a few reasons. One, the rings are curved, meaning the FFT won't be a spike, but more like a hazy part that is just slightly brighter than the surrounding areas. If you zoom in on the FFT you will see sort of a "cone" shape. This is because the frequencies are changing direction (curved lines) thus it is not a single spike. So you have to delete a large portion. The second issue is that doing it for this specific frequency causes issues on the edge of the Sun in which the curve is the same orientation.

 

I learned more from this example.

Takeaways:

1.) images of the Sun which include parts of the edge with the same orientation as the Newton's rings are hard to work with.

2.) Not all Newton's rings look the same.

3.) More curved Newton Rings are even more difficult to work with.

 

Thanks for letting me try your image, it is a really nice pic. Here is my best result (note the distortion on the edge):

 

Inverse FFT of test.jpg

 

Jaxon



#20 jt_3232

jt_3232

    Mariner 2

  • *****
  • Posts: 273
  • Joined: 27 Aug 2017

Posted 25 May 2021 - 10:51 PM

Or, at this point, you can cheat and use some really quick photoshop work (I use photoshop elements, the buy once for a forever license stripped down version of photoshop):

 

test_photoshop.jpg




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics