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

Nikon Coloured Concentric Rings

  • Please log in to reply
217 replies to this topic

#1 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 19 December 2020 - 06:37 PM

There's anecdotal evidence that the Nikon D5300 can suffer from coloured concentric rings in stacked astro-images.  They become apparent after calibration with flats.  So I thought I'd investigate further.  I have a Nikon D5300 which I bought as a backup camera but it's been sitting around gathering dust because my testing uncovered the green stars issue.  Out of the cupboard it came!

 

Here's a series of flats taken at ISO 100 in 14-bit mode with a legacy 50mm f/1.8 Nikon lens, which has no electronic chip and is unrecognised by the camera firmware:

 

NikonD5300_ISO100.jpg

 

Going left to right, top to bottom each exposure is half the previous one.  The exposure in the middle of the 3x3 montage has its back-of-camera histogram peak 1/4 from the left which is a typical recommendation for shooting light frames.  The above shots are the out-of-camera JPGs but for the rest of the analysis I use the raws.

 

Each flat is divided by the bright one at the top left, using PixInsight as follows:

  • Debayer using SuperPixel mode to preserve the original raw pixel values
  • Subtract the bias level of 600
  • Perform 8x8 binning (to reduce noise) by using IntegerResample using the "average" downsample mode
  • Divide by the 8x8 binned, bias subtracted, superpixel debayered top left flat
  • Apply an unlinked screen transfer function

Here's the result:

 

NikonD5300_ISO100Concentric.jpg

 

The coloured concentric rings are pretty clear.  It's the kind of thing you might see appearing in ISO 100 calibrated astro-images.  Sometimes it will be the light frames that contain the rings and sometimes it will be the flat frames.  Sometimes it will be both.

 

Anecdotally images taken at higher ISOs don't show any problem, so I thought I'd test this.  Here's an equivalent 3x3 montage generated from ISO 800 flats.  Again each flat differs by one stop and the central image had its back-of-camera histogram peak 1/4 from the left:

 

NikonD5300_ISO800Concentric.jpg

 

So the rings are slightly different and generally much fainter but they don't disappear.

 

At the present time I have no idea what is causing these rings except that it must be caused by some kind of internal camera processing of the raw data. Also I've no idea how many other Nikon models might show the same artefacts.

 

My test procedure is quite clear (I think) so try it on your own cameras and feel free to add your results to this thread.

 

Mark


Edited by sharkmelley, 19 December 2020 - 06:46 PM.

  • Ivo Jager, whwang, endless-sky and 1 other like this

#2 limeyx

limeyx

    Apollo

  • -----
  • Posts: 1,228
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 19 December 2020 - 07:03 PM

There's anecdotal evidence that the Nikon D5300 can suffer from coloured concentric rings in stacked astro-images.  They become apparent after calibration with flats.  So I thought I'd investigate further.  I have a Nikon D5300 which I bought as a backup camera but it's been sitting around gathering dust because my testing uncovered the green stars issue.  Out of the cupboard it came!

 

Here's a series of flats taken at ISO 100 in 14-bit mode with a legacy 50mm f/1.8 Nikon lens, which has no electronic chip and is unrecognised by the camera firmware:

 

attachicon.gifNikonD5300_ISO100.jpg

 

Going left to right, top to bottom each exposure is half the previous one.  The exposure in the middle of the 3x3 montage has its back-of-camera histogram peak 1/4 from the left which is a typical recommendation for shooting light frames.  The above shots are the out-of-camera JPGs but for the rest of the analysis I use the raws.

 

Each flat is divided by the bright one at the top left, using PixInsight as follows:

  • Debayer using SuperPixel mode to preserve the original raw pixel values
  • Subtract the bias level of 600
  • Perform 8x8 binning (to reduce noise) by using IntegerResample using the "average" downsample mode
  • Divide by the 8x8 binned, bias subtracted, superpixel debayered top left flat
  • Apply an unlinked screen transfer function

Here's the result:

 

attachicon.gifNikonD5300_ISO100Concentric.jpg

 

The coloured concentric rings are pretty clear.  It's the kind of thing you might see appearing in ISO 100 calibrated astro-images.  Sometimes it will be the light frames that contain the rings and sometimes it will be the flat frames.  Sometimes it will be both.

 

Anecdotally images taken at higher ISOs don't show any problem, so I thought I'd test this.  Here's an equivalent 3x3 montage generated from ISO 800 flats.  Again each flat differs by one stop and the central image had its back-of-camera histogram peak 1/4 from the left:

 

attachicon.gifNikonD5300_ISO800Concentric.jpg

 

So the rings are slightly different and generally much fainter but they don't disappear.

 

At the present time I have no idea what is causing these rings except that it must be caused by some kind of internal camera processing of the raw data. Also I've no idea how many other Nikon models might show the same artefacts.

 

My test procedure is quite clear (I think) so try it on your own cameras and feel free to add your results to this thread.

 

Mark

Great post Mark. Very frustrating as I know a lot of people use this camera successfully and that's why I bought this one. I don't really want to have to get a new camera ...

 

I am currently imaging in a slightly darker site and with almost no moon and the rings are less pronounced but under extreme stretch definitely still there :(

 

Very disappointing

 

I usually use ISO200. Maybe I need to consider 400 or even 800



#3 gezak22

gezak22

    Gemini

  • *****
  • Posts: 3,385
  • Joined: 15 Aug 2004
  • Loc: On far side of moon. Send help.

Posted 19 December 2020 - 07:24 PM

I had this on my Sony A7R3 with a specific lens, even when all camera-internal lens corrections were turned off. The only way I was able to resolve it was by covering up the electrical contacts between lens and camera. Unfortunately you seem to be getting these rings even without any electrical connection between camera and lens. :(



#4 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 19 December 2020 - 07:27 PM

Another useful method of analysis is to take a single flat and divide one channel by another (after superpixel debayering, bias subtraction and binning).

 

Here is the result for the middle image of the top row in the ISO 100 analysis (DSC_0034.NEF):

 

NikonD5300_ISO100ChannelDivision.png

 

It appears to suggest that for this individual flat most of the problem is in the green channel.

 

Mark


Edited by sharkmelley, 20 December 2020 - 05:21 AM.

  • limeyx likes this

#5 vidrazor

vidrazor

    Viking 1

  • *****
  • Posts: 918
  • Joined: 31 Oct 2017

Posted 19 December 2020 - 07:47 PM

At the present time I have no idea what is causing these rings except that it must be caused by some kind of internal camera processing of the raw data.

If it may be on the firmware side of things, there is a firmware hack that is used on the D5100 (I forget the name offhand), I'm not sure if that firmware will work on other bodies, but if so it might be worth to try the process through the third-party firmware and see what results that may yield.
 


  • sharkmelley likes this

#6 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 20 December 2020 - 05:21 AM

It appears to suggest that for this individual flat most of the problem is in the green channel.

Take a look at the RawDigger histogram for the green channel of the same flat DSC_0034.NEF:

 

NikonD5300GreenHistogram.png

 

My guess is that the rings are caused by the peaks and valleys in the histogram which I believe result from Nikon's lossy data compression algorithm.  However, without knowing the details of this algorithm it will not be easy to reproduce.  Unfortunately data compression cannot be switched off on the D5300 so it's impossible to perform a direct comparison test.  It's something that could be tried with other Nikons.

 

Mark


Edited by sharkmelley, 20 December 2020 - 05:36 AM.

  • SandyHouTex likes this

#7 endless-sky

endless-sky

    Apollo

  • -----
  • Posts: 1,013
  • Joined: 24 May 2020
  • Loc: Padova, Italy

Posted 20 December 2020 - 05:56 AM

Thanks for the useful comparison/information, Mark!

 

I typically shoot at 200 ISO, so it would be worth for me to do the test as well. Maybe I'll set aside some time, later this afternoon.



#8 wsbrown

wsbrown

    Vostok 1

  • *****
  • Posts: 132
  • Joined: 04 Jan 2019
  • Loc: Baxley, GA

Posted 20 December 2020 - 09:40 AM

I had rings like this in an image I captured a few nights ago. I had recently switched from ISO 400 to 200 just to try it out. I looked at my flats after reading your post and though they are not as prominent as the rings in your ISO 100 per channel examples, they are there, very faint. I will be going back to ISO 400 as I never had any issues there. My camera is a D5300 as well.



#9 Kevin_A

Kevin_A

    Viking 1

  • *****
  • Posts: 636
  • Joined: 22 Aug 2018
  • Loc: Belmont, Ontario Canada

Posted 20 December 2020 - 09:41 AM

Mark,

I really appreciate your time and effort exploring, testing and uncovering the strengths and pitfalls of the Nikon platform. All your hard work with the Z6 and the D5300 helps us fellow Nikonians steer our efforts in the right direction to produce much better images, while avoiding a lot of wasted clear sky time shooting at subpar camera/exposure settings. I am one who is truly thankful and appreciative for your vital contributions over the years...

 

Thanks a whole bunch Mark!!!!


  • SandyHouTex and Dagryl like this

#10 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,766
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 20 December 2020 - 02:02 PM

There's anecdotal evidence that the Nikon D5300 can suffer from coloured concentric rings in stacked astro-images.  They become apparent after calibration with flats.  So I thought I'd investigate further.  I have a Nikon D5300 which I bought as a backup camera but it's been sitting around gathering dust because my testing uncovered the green stars issue.  Out of the cupboard it came!

 

Here's a series of flats taken at ISO 100 in 14-bit mode with a legacy 50mm f/1.8 Nikon lens, which has no electronic chip and is unrecognised by the camera firmware:

 

attachicon.gifNikonD5300_ISO100.jpg

 

Going left to right, top to bottom each exposure is half the previous one.  The exposure in the middle of the 3x3 montage has its back-of-camera histogram peak 1/4 from the left which is a typical recommendation for shooting light frames.  The above shots are the out-of-camera JPGs but for the rest of the analysis I use the raws.

 

Each flat is divided by the bright one at the top left, using PixInsight as follows:

  • Debayer using SuperPixel mode to preserve the original raw pixel values
  • Subtract the bias level of 600
  • Perform 8x8 binning (to reduce noise) by using IntegerResample using the "average" downsample mode
  • Divide by the 8x8 binned, bias subtracted, superpixel debayered top left flat
  • Apply an unlinked screen transfer function

Here's the result:

 

attachicon.gifNikonD5300_ISO100Concentric.jpg

 

The coloured concentric rings are pretty clear.  It's the kind of thing you might see appearing in ISO 100 calibrated astro-images.  Sometimes it will be the light frames that contain the rings and sometimes it will be the flat frames.  Sometimes it will be both.

 

Anecdotally images taken at higher ISOs don't show any problem, so I thought I'd test this.  Here's an equivalent 3x3 montage generated from ISO 800 flats.  Again each flat differs by one stop and the central image had its back-of-camera histogram peak 1/4 from the left:

 

attachicon.gifNikonD5300_ISO800Concentric.jpg

 

So the rings are slightly different and generally much fainter but they don't disappear.

 

At the present time I have no idea what is causing these rings except that it must be caused by some kind of internal camera processing of the raw data. Also I've no idea how many other Nikon models might show the same artefacts.

 

My test procedure is quite clear (I think) so try it on your own cameras and feel free to add your results to this thread.

 

Mark

Holy crap, but nice job Mark.



#11 limeyx

limeyx

    Apollo

  • -----
  • Posts: 1,228
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 20 December 2020 - 04:58 PM

I've been playing with this during imaging nights - admittedly far less controlled. I've increased ISO, extended my lens shroud etc etc

 

I shot three nights here in Bortle 5 with moon at 0-13% and when stacked with PI didnt see any appreciable circles (I think one night at ISO400 and two at 200)

When I stacked with DSS I did see more of the effect under extreme stretch (My flats also may have been a bit under because I dont have my regular setup)

 

Last night at ISO400 with a larger moon % (although it was behind a hill the entire night and not actually visible) and I have the rings again ...

 

Too hard for a smoking gun of the moon but I'll hopefully get some clear nights at the next new moon and can see if this repeats



#12 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,206
  • Joined: 27 Oct 2014

Posted 20 December 2020 - 06:13 PM

Nikon D5500.

 

I avoid ISO 100, it's known to have glitches.

 

Using ISO 200 I've only had the problem once, and could mostly remove it in processing.

 

So, just as a practical deal, isn't a concern for me.  Which explains why there's so many strong positive opinions for the D5300/5500/5600.



#13 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,206
  • Joined: 27 Oct 2014

Posted 20 December 2020 - 06:16 PM

If it may be on the firmware side of things, there is a firmware hack that is used on the D5100 (I forget the name offhand), I'm not sure if that firmware will work on other bodies, but if so it might be worth to try the process through the third-party firmware and see what results that may yield.
 

I'm not sure the firmware did anything much other the raise the black point from zero, which is where Nikon set it on the 5100, and most (all?) cameras before the 5300.



#14 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 20 December 2020 - 06:49 PM

I'm not sure the firmware did anything much other the raise the black point from zero, which is where Nikon set it on the 5100, and most (all?) cameras before the 5300.

On the D5100 the Nikon hack also removed the raw data filtering (a.k.a. star eater) and switched off the lossy compression of NEF files.  It's the lossy compression that I think may be responsible for the concentric coloured banding.

 

It's not clear to me if these options are available on the D5300 version of the Nikon hack.

 

Mark


Edited by sharkmelley, 20 December 2020 - 06:52 PM.


#15 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,766
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 20 December 2020 - 08:26 PM

Nikon D5500.

 

I avoid ISO 100, it's known to have glitches.

 

Using ISO 200 I've only had the problem once, and could mostly remove it in processing.

 

So, just as a practical deal, isn't a concern for me.  Which explains why there's so many strong positive opinions for the D5300/5500/5600.

Since those cameras all use the same sensor with green stars and concentric rings at low ISO, I think I’ll pass.


Edited by SandyHouTex, 20 December 2020 - 08:27 PM.


#16 17.5Dob

17.5Dob

    Fly Me to the Moon

  • ****-
  • Posts: 7,084
  • Joined: 21 Mar 2013
  • Loc: Colorado,USA

Posted 20 December 2020 - 09:29 PM

With 3 years of stacking, over multiple targets, using either ISO 200 or 400...I've seen seen any 'rings' with my D5300


Edited by 17.5Dob, 20 December 2020 - 09:29 PM.


#17 vidrazor

vidrazor

    Viking 1

  • *****
  • Posts: 918
  • Joined: 31 Oct 2017

Posted 20 December 2020 - 11:56 PM

On the D5100 the Nikon hack also removed the raw data filtering (a.k.a. star eater) and switched off the lossy compression of NEF files.  It's the lossy compression that I think may be responsible for the concentric coloured banding.

 

It's not clear to me if these options are available on the D5300 version of the Nikon hack.

 

Mark

It's probably worth checking out. It would be interesting to see if it helps. I forget where to get the firmware tho. I was gonna get it for my D5100, but then started shooting with a MFT body instead.



#18 Mike in Rancho

Mike in Rancho

    Viking 1

  • -----
  • Posts: 677
  • Joined: 15 Oct 2020
  • Loc: Alta Loma, CA

Posted 21 December 2020 - 05:05 AM

Interesting research! waytogo.gif

 

I have been fighting these "rainbow rings" at times, we've even had a couple recent threads going on in beginner DSO.  They make post-processing a bear, if not nearly impossible.

 

The only "fix" I've really tried is to stop down the aperture on my camera lens - which as a zoom I was suspecting to be the culprit.  I think I've seen it on both the Tamron and Nikkor.  I changed from f/5.6 (wide open) to f/10, which took me from a full screen rainbow to just a remnant blue ringing I was able to kill off in processing.  Of course, stopping down means upping the ISO, going from 200 to 400.  I also imagine I was underexposed on both of those, if that might mean anything.

 

I have not noticed the rainbow rings, at least not yet knock on wood, with the same camera on an ED telescope.  Those have been shot at ISO400 of late, typical 1/4 to 1/3 histogram.

 

I have been meaning to do more experimenting, time permitting, to try to flush it out in case I plan on using the camera lenses again.

 

Are the procedures you went through (over my head!) akin to what happens in stacking?  I had discarded the thought of flats being the problem, on the assumption that the stacking programs create a master flat which is grayscale.  No?  Or could ringing luminance in the grayscale master sort of "reverse engineer" the colors back in when applied to the lights?



#19 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 21 December 2020 - 06:05 AM

Are the procedures you went through (over my head!) akin to what happens in stacking?  I had discarded the thought of flats being the problem, on the assumption that the stacking programs create a master flat which is grayscale.  No?  Or could ringing luminance in the grayscale master sort of "reverse engineer" the colors back in when applied to the lights?

You raise an important point.  Yes my procedure is akin to what happens when calibrating an exposure with a flat frame.  In my example I am calibrating one flat frame with another but the same thing will happen when calibrating a light with a flat.

 

There are many variables.  For instance, if the optics do not suffer from vignetting then the rings are much less likely to be a problem.  That's why stopping down the lens might help but it's not a good solution because it means you need a much longer total integration time to obtain the same image quality.

 

From the evidence I have so far, here are my recommendations when using low ISOs on the Nikon D5300:

  • Shoot your lights so the peak of the back-of-camera histogram is slightly below the 1/4 level.  
  • Shoot your flats so the peak of the back-of-camera histogram is as far to the right as possible (without clipping) or alternatively slightly below the 1/4 level.

It's likely that these precautions are necessary only if the telescope or lens suffers from obvious vignetting.  But further experiments are needed to verify this.

 

Mark


  • limeyx and Mike in Rancho like this

#20 endless-sky

endless-sky

    Apollo

  • -----
  • Posts: 1,013
  • Joined: 24 May 2020
  • Loc: Padova, Italy

Posted 21 December 2020 - 06:40 AM

Interesting research! waytogo.gif

 

I have been fighting these "rainbow rings" at times, we've even had a couple recent threads going on in beginner DSO.  They make post-processing a bear, if not nearly impossible.

 

The only "fix" I've really tried is to stop down the aperture on my camera lens - which as a zoom I was suspecting to be the culprit.  I think I've seen it on both the Tamron and Nikkor.  I changed from f/5.6 (wide open) to f/10, which took me from a full screen rainbow to just a remnant blue ringing I was able to kill off in processing.  Of course, stopping down means upping the ISO, going from 200 to 400.  I also imagine I was underexposed on both of those, if that might mean anything.

 

I have not noticed the rainbow rings, at least not yet knock on wood, with the same camera on an ED telescope.  Those have been shot at ISO400 of late, typical 1/4 to 1/3 histogram.

 

I have been meaning to do more experimenting, time permitting, to try to flush it out in case I plan on using the camera lenses again.

 

Are the procedures you went through (over my head!) akin to what happens in stacking?  I had discarded the thought of flats being the problem, on the assumption that the stacking programs create a master flat which is grayscale.  No?  Or could ringing luminance in the grayscale master sort of "reverse engineer" the colors back in when applied to the lights?

 

You raise an important point.  Yes my procedure is akin to what happens when calibrating an exposure with a flat frame.  In my example I am calibrating one flat frame with another but the same thing will happen when calibrating a light with a flat.

 

There are many variables.  For instance, if the optics do not suffer from vignetting then the rings are much less likely to be a problem.  That's why stopping down the lens might help but it's not a good solution because it means you need a much longer total integration time to obtain the same image quality.

 

From the evidence I have so far, here are my recommendations when using low ISOs on the Nikon D5300:

  • Shoot your lights so the peak of the back-of-camera histogram is slightly below the 1/4 level.  
  • Shoot your flats so the peak of the back-of-camera histogram is as far to the right as possible (without clipping) or alternatively slightly below the 1/4 level.

It's likely that these precautions are necessary only if the telescope or lens suffers from obvious vignetting.  But further experiments are needed to verify this.

 

Mark

Maybe worth mentioning:

 

I have been using my D5300 since April, mostly with a kit zoom lens, Nikkor 70-300mm. I have seen the dreaded rings in a couple of occasions, while using it with that lens. But back then I didn't know how to take flats properly (I was exposing them to 1/2 back of camera histogram, instead of 50% full well - which corresponds to, as Mark stated, as far as to the right as possible, without clipping). I also fully blamed the lens, because of its tilting, aberrations, being a zoom, having internal reflections, etc.

 

In no moment I thought it was the camera, as I was unaware of these inherent problems that Mark has brought to the surface with his posts/tests.

 

I have only been using my telescope setup since October, and only finished post-processing two images (while waiting for clear skies to finish acquiring two more, that are almost done). But I haven't seen the rings anymore, with the new setup. Not even with aggressive stretching.

 

All the frames I have been taking (since the new setup) have ADU >> 10*RN^2 (900-1100 mean), 200 ISO. And calibrated with bias and flats (50% full-well, according to PixInsight Statistics). The logic behind the higher ADUs was because of some posts Mark made that seemed to indicate that for DSLRs, staying too close to the 10*RN^2 value could cause artefacts/rings and it seemed to be better to go higher, in order to counteract this.

 

From his last findings, instead, it would seem to be better to stay even below 1/4.


  • Mike in Rancho likes this

#21 Mike in Rancho

Mike in Rancho

    Viking 1

  • -----
  • Posts: 677
  • Joined: 15 Oct 2020
  • Loc: Alta Loma, CA

Posted 21 December 2020 - 03:46 PM

You raise an important point.  Yes my procedure is akin to what happens when calibrating an exposure with a flat frame.  In my example I am calibrating one flat frame with another but the same thing will happen when calibrating a light with a flat.

 

There are many variables.  For instance, if the optics do not suffer from vignetting then the rings are much less likely to be a problem.  That's why stopping down the lens might help but it's not a good solution because it means you need a much longer total integration time to obtain the same image quality.

 

From the evidence I have so far, here are my recommendations when using low ISOs on the Nikon D5300:

  • Shoot your lights so the peak of the back-of-camera histogram is slightly below the 1/4 level.  
  • Shoot your flats so the peak of the back-of-camera histogram is as far to the right as possible (without clipping) or alternatively slightly below the 1/4 level.

It's likely that these precautions are necessary only if the telescope or lens suffers from obvious vignetting.  But further experiments are needed to verify this.

 

Mark

Thanks, Mark, good stuff.  I saved your current recommendations and will see how things go, and add confirmation or any other findings here.

 

I for sure have vignetting with the lenses.  I'll have to chart that out, also comparing them to my scope, with and without a focal reducer.  As for histograms, well, I've been getting a sneaking suspicion they are not all created equal!

 

Initially I was using only the BOC screen, and still do so for my flats.  But lights acquisition I now take with BackyardNikon, setting up my exposure based on the histogram that it displays.  I wonder if I need Google Translate in order to compare the two?

 

I see "ADU" being mentioned and will have to read up on that.  Is that a firm standard that will be the same regardless of whatever software is being used to generate it?



#22 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,206
  • Joined: 27 Oct 2014

Posted 21 December 2020 - 06:22 PM

Thanks, Mark, good stuff.  I saved your current recommendations and will see how things go, and add confirmation or any other findings here.

 

I for sure have vignetting with the lenses.  I'll have to chart that out, also comparing them to my scope, with and without a focal reducer.  As for histograms, well, I've been getting a sneaking suspicion they are not all created equal!

 

Initially I was using only the BOC screen, and still do so for my flats.  But lights acquisition I now take with BackyardNikon, setting up my exposure based on the histogram that it displays.  I wonder if I need Google Translate in order to compare the two?

 

I see "ADU" being mentioned and will have to read up on that.  Is that a firm standard that will be the same regardless of whatever software is being used to generate it?

It's firm, but.

 

You have to be careful about how many bits are being used.

 

Images need to be linear, not stretched.  Stretching changes things in ways that are not easily dealt with.


  • Mike in Rancho likes this

#23 sharkmelley

sharkmelley

    Aurora

  • *****
  • topic starter
  • Posts: 4,621
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 21 December 2020 - 06:30 PM

 As for histograms, well, I've been getting a sneaking suspicion they are not all created equal!

 

Initially I was using only the BOC screen, and still do so for my flats.  But lights acquisition I now take with BackyardNikon, setting up my exposure based on the histogram that it displays.  I wonder if I need Google Translate in order to compare the two?

 

I see "ADU" being mentioned and will have to read up on that.  Is that a firm standard that will be the same regardless of whatever software is being used to generate it?

The histogram displayed on the back of camera is the histogram of the in-camera JPG.  In other words it is the histogram of data that has been non-linearly transformed into a colour space such as sRGB or AdobeRGB.  Histograms that you see in other applications might be histograms of the original linear data.  It's always important to know which you are looking at because they are very different.  I'm not sure what Backyard Nikon displays.

 

ADU means "analogue-digital-unit" and means exactly the same as DN which is "digital number".  They are the pixel values in the raw data file.  They are referred to as ADU or DN to distinguish them from electrons which is the alternative unit for measuring the brightness of a pixel.

 

Mark


Edited by sharkmelley, 21 December 2020 - 06:33 PM.

  • SandyHouTex and Mike in Rancho like this

#24 Mike in Rancho

Mike in Rancho

    Viking 1

  • -----
  • Posts: 677
  • Joined: 15 Oct 2020
  • Loc: Alta Loma, CA

Posted 23 December 2020 - 02:43 AM

So with one of my recently-taken flats, I went histogram hunting...

 

The back of the camera displayed at the half point, as intended.  Backyard Nikon, Darktable, ViewNX-i, Capture NX-D, and Gimp also all showed the data at half the histogram.  However, if I changed the Gimp setting to "linear space," it then dropped to around 1/5 of the histogram.  ASTAP showed something between 1500-3000 ADU, also about 1/5 the histogram after resetting the scale, but was a bit clunky to use.  IRIS auto-scaled to 16383 and showed two  peaks, at about 1250 and 2250.

 

I took test shots and found I could get the ADU histogram around 8K (half the 16484 for 14-bit) at just past 7/8 of the BOC histogram.  If I pushed it as far as I could (but still not edged 100% over into clipping), it would go up in the range of 13K ADU, so back at 7/8 is where I settled.  This was with my 300mm lens, and I will have to see if that rule of thumb holds with different optics.

 

Flats BOC.jpg

 

I took new flats matched up to the same data I had used before as an example of the "rainbow rings" in another thread, then restacked everything.  I set focus using a nearby mountaintop, so while maybe not perfect should have been pretty close to the proper infinity setting.  Next, I stretched both using matching steps in Gimp, to push it to reveal what was there.  Still some issues, but I'd say a large improvement.

 

NGC1499 Flat Comparo.jpg

 

That original data was so bad that I could barely make anything out of it that didn't still have remnants of those colored rings in them, whether in StarTools, or Gimp.  In the end of course I ditched it all and acquired new data stopped down to something like f/10.

 

Now though, with that old data at f/5.6, but with the new flats I just took, and run through StarTools, I was able to produce reasonable results without heavy cropping and wiping away half the nebula.

 

I'm now thinking some of my old data might be salvageable with newly-taken flats.  And for sure I probably need to take new flats for the data I currently have waiting to process.

 

Thanks again Mark. waytogo.gif  



#25 limeyx

limeyx

    Apollo

  • -----
  • Posts: 1,228
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 23 December 2020 - 07:53 PM

So with one of my recently-taken flats, I went histogram hunting...

 

The back of the camera displayed at the half point, as intended.  Backyard Nikon, Darktable, ViewNX-i, Capture NX-D, and Gimp also all showed the data at half the histogram.  However, if I changed the Gimp setting to "linear space," it then dropped to around 1/5 of the histogram.  ASTAP showed something between 1500-3000 ADU, also about 1/5 the histogram after resetting the scale, but was a bit clunky to use.  IRIS auto-scaled to 16383 and showed two  peaks, at about 1250 and 2250.

 

I took test shots and found I could get the ADU histogram around 8K (half the 16484 for 14-bit) at just past 7/8 of the BOC histogram.  If I pushed it as far as I could (but still not edged 100% over into clipping), it would go up in the range of 13K ADU, so back at 7/8 is where I settled.  This was with my 300mm lens, and I will have to see if that rule of thumb holds with different optics.

 

attachicon.gifFlats BOC.jpg

 

I took new flats matched up to the same data I had used before as an example of the "rainbow rings" in another thread, then restacked everything.  I set focus using a nearby mountaintop, so while maybe not perfect should have been pretty close to the proper infinity setting.  Next, I stretched both using matching steps in Gimp, to push it to reveal what was there.  Still some issues, but I'd say a large improvement.

 

attachicon.gifNGC1499 Flat Comparo.jpg

 

That original data was so bad that I could barely make anything out of it that didn't still have remnants of those colored rings in them, whether in StarTools, or Gimp.  In the end of course I ditched it all and acquired new data stopped down to something like f/10.

 

Now though, with that old data at f/5.6, but with the new flats I just took, and run through StarTools, I was able to produce reasonable results without heavy cropping and wiping away half the nebula.

 

I'm now thinking some of my old data might be salvageable with newly-taken flats.  And for sure I probably need to take new flats for the data I currently have waiting to process.

 

Thanks again Mark. waytogo.gif  

Interesting. I have noticed that if I dim my lightpad to where I can get 3-4 sec exposures on flats with the D5300, they correct dust etc far better than <1 sec

Not sure about the rings yet -- I will check that out




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