873 replies to this topic

[jasondain]

On one of these forums, I saw a chart that showed the best ISO and exposure setting (histogram level) to avoid rings based on experimentation done with a Nikon camera.  I think if I recall, the sweet spot was ISO3200 with 50 percent exposure level for the lights and almost 80+ percent for the flats.  Has anybody got that chart handy?

 

Thanks

[sharkmelley]

On one of these forums, I saw a chart that showed the best ISO and exposure setting (histogram level) to avoid rings based on experimentation done with a Nikon camera.  I think if I recall, the sweet spot was ISO3200 with 50 percent exposure level for the lights and almost 80+ percent for the flats.  Has anybody got that chart handy?

It very much depends on the camera model because Nikon does different things with different cameras.

[jasondain]

OK.  Do you have any info on the 850 in terms of best exposure and iso to avoid the rings?  If not, would i just follow your methodology to determine this using various settings?

 

Thanks

[sharkmelley]

For the Nikon D850 the best advice is to shoot at a high ISO (e.g. ISO 1600 or higher) and use a combination of ISO and exposure length that puts the peak of the back-of-camera histogram central.  Shoot flats so the back-of-camera histogram is well over to the right but without saturation.

[Infomastr]

For the Nikon D850 the best advice is to shoot at a high ISO (e.g. ISO 1600 or higher) and use a combination of ISO and exposure length that puts the peak of the back-of-camera histogram central.  Shoot flats so the back-of-camera histogram is well over to the right but without saturation.

Would you recommend the same for the Z6ii? 

[sharkmelley]

Would you recommend the same for the Z6ii? 

Yes, the same recommendation applies to Z6II because the main Z6II problem is the hardcoded correction (the lossy data compression is a non-issue because it can be switched off).  So shoot at a high ISO (e.g. ISO 1600 or higher) and use a combination of ISO and exposure length that places the peak of the back-of-camera histogram roughly central.  For flats, follow the usual DSLR advice of shooting them so the back-of-camera histogram is well over to the right but without saturation. 

[Infomastr]

Yes, the same recommendation applies to Z6II because the main Z6II problem is the hardcoded correction (the lossy data compression is a non-issue because it can be switched off).  So shoot at a high ISO (e.g. ISO 1600 or higher) and use a combination of ISO and exposure length that places the peak of the back-of-camera histogram roughly central.  For flats, follow the usual DSLR advice of shooting them so the back-of-camera histogram is well over to the right but without saturation. 

Thank you.

[jasondain]

For the Nikon D850 the best advice is to shoot at a high ISO (e.g. ISO 1600 or higher) and use a combination of ISO and exposure length that puts the peak of the back-of-camera histogram central.  Shoot flats so the back-of-camera histogram is well over to the right but without saturation.

In my tests, even ISO 1600 gives rings especially with the lens recognized by the camera.  I can shoot lower ISO with no issues on telescopes or taped over lens contacts on manual aperture lenses.  I have found that mid histogram lights at ISO3200 with 80 percent flats works well reliably with Nikon or other lenses with electronic aperture control where you can't tape the contacts.

[primeshooter]

In my tests, even ISO 1600 gives rings especially with the lens recognized by the camera.  I can shoot lower ISO with no issues on telescopes or taped over lens contacts on manual aperture lenses.  I have found that mid histogram lights at ISO3200 with 80 percent flats works well reliably with Nikon or other lenses with electronic aperture control where you can't tape the contacts.

I never got rings at 800, let alone 1600 as long as the histogram was at least 1/3 over, but closer to half. How far over did you shoot lights?

[jasondain]

I never got rings at 800, let alone 1600 as long as the histogram was at least 1/3 over, but closer to half. How far over did you shoot lights?

I was shooting lights at about 50 percent histogram peak.  What lens were you shooting with?

[primeshooter]

I was shooting lights at about 50 percent histogram peak.  What lens were you shooting with?

70-200 2.8S. I also use a 50 a lot. Are you sure they are the rings we are talking about, let's see an example if you can?

[james7ca]

I think it also depends upon the darkness of your skies and if your base exposure and integration have lots of signal you may never even see this issue. Thus, I think light pollution and gradients make the problem even worse and the more you stretch your images the more likely you will see this problem. I also believe vignetting affects the results and thus with vignetting you are more likely to see the rings.

[primeshooter]

70-200 2.8S. I also use a 50 a lot. Are you sure they are the rings we are talking about, let's see an example if you can?

I meant E of course, on f mount. I have the 70-200/2.8S now for the Z8.

[primeshooter]

I think it also depends upon the darkness of your skies and if your base exposure and integration have lots of signal you may never even see this issue. Thus, I think light pollution and gradients make the problem even worse and the more you stretch your images the more likely you will see this problem. I also believe vignetting affects the results and thus with vignetting you are more likely to see the rings.

Maybe. Mark has done so much investigation into this problem I just listen to his advice and it has helped immensely.

[sharkmelley]

Thanks to a set of test files from CN member vidrazor I can confirm that the Nikon D600 has the hardcoded ring problem when used with a lens recognized by the camera firmware:

 

 

Above is my usual plot where I divide the (bias-subtracted) red channel by the (bias-subtracted) green channel in the top row and the (bias-subtracted) blue channel by the (bias-subtracted) green channel.  The progression of rings from frame to frame can be clearly seen.

 

We currently have no information on what happens if the same camera is used on a telescope.

 

Mark

[vidrazor]

Thanks to a set of test files from CN member vidrazor I can confirm that the Nikon D600 has the hardcoded ring problem when used with a lens recognized by the camera firmware:

Above is my usual plot where I divide the (bias-subtracted) red channel by the (bias-subtracted) green channel in the top row and the (bias-subtracted) blue channel by the (bias-subtracted) green channel.  The progression of rings from frame to frame can be clearly seen.

We currently have no information on what happens if the same camera is used on a telescope.

Mark

Thanks Mark.

 

So what would be a way to mitigate it? I know for the D5300 you've suggested to shoot at base ISO, would that apply here, or would shooting a high ISO and allowing the noise levels to diffuse it be a better approach?

[sharkmelley]

So what would be a way to mitigate it?

The good news from the D600 files you sent me is that the hardcoded correction being applied is not a very strong correction.  If a histogram of the raw values is plotted and the peak of the histogram is below a raw value of approx 300 then no rings should appear in the data.  So choose a combination of (low) ISO and exposure length that achieves this.  It means that the back-of-camera histogram will be well over to the left.

[sharkmelley]

Thanks to a Reddit contributor who sent me a set of test files, we can now confirm that the Nikon Z5 suffers from the hardcoded correction that causes concentric coloured rings.  The test files were taken with a Nikon AI lens that was not recognised by the firmware.  Here's the result:

 

 

The top row shows the (bias-subtracted) red channel divided by the (bias-subtracted) green channel and the bottom row is the (bias-subtracted) blue divided by the (bias-subtracted) green. From left to right the exposure is reducing, so the rings become more widely spaced. 

 

 

[vidrazor]

Thanks to a Reddit contributor who sent me a set of test files, we can now confirm that the Nikon Z5 suffers from the hardcoded correction that causes concentric coloured rings.  The test files were taken with a Nikon AI lens that was not recognised by the firmware.  Here's the result:

So in these instances, it's best to shoot low ISO with the flats around the 3/4 mark?
 

[sharkmelley]

So in these instances, it's best to shoot low ISO with the flats around the 3/4 mark?
 

My usual advice is that the back-of-camera histogram peak should be approximately halfway across for the lights and much further to the right (without saturating) for the flats. Use a combination of exposure length and ISO that achieves this - preferably a high ISO.

[vidrazor]

My usual advice is that the back-of-camera histogram peak should be approximately halfway across for the lights and much further to the right (without saturating) for the flats. Use a combination of exposure length and ISO that achieves this - preferably a high ISO.

So high ISO for a Z5? I know for the D600 and D5300 you've suggested a low ISO. What would require one over the other?

[sharkmelley]

Thanks to Cloudy Nights member vidrazor here are the results of the Nikon Z50 with an adapted non-CPU lens:

 

 

As usual, the top row shows the (bias-subtracted) red channel divided by the (bias-subtracted) green channel and the bottom row is the (bias-subtracted) blue divided by the (bias-subtracted) green. From left to right the exposure is reducing, so the rings become more widely spaced.

 

Since the hardcoded correction exists for a lens that the firmware does not recognise, we can be sure that it will be applied for all lenses and telescopes.

[sharkmelley]

So high ISO for a Z5? I know for the D600 and D5300 you've suggested a low ISO. What would require one over the other?

That's an interesting question.  The Reddit contributor who sent me the the Z5 test files also sent me a exposure of M45 (The Pleiades) without rings.  It was an ISO 100 exposure where peak of the back-of-camera histogram was approx. 1/10 from the left.  This worked for the Z5 but wouldn't work for the Z50.

 

In general, as the exposure reduces, the Nikon rings become more and more widely spaced until they disappear altogether.  But for the Z50 the rings don't disappear until the (bias-subtracted) values in the red and blue channels are down near 40 digital units which would be a crazily low exposure.

[vidrazor]

That's an interesting question.  The Reddit contributor who sent me the the Z5 test files also sent me a exposure of M45 (The Pleiades) without rings.  It was an ISO 100 exposure where peak of the back-of-camera histogram was approx. 1/10 from the left.  This worked for the Z5 but wouldn't work for the Z50.

In general, as the exposure reduces, the Nikon rings become more and more widely spaced until they disappear altogether.  But for the Z50 the rings don't disappear until the (bias-subtracted) values in the red and blue channels are down near 40 digital units which would be a crazily low exposure.

So high ISO for Z50 then?
 

[primeshooter]

I've been putting this in front of Nikon for a while now. He who dares wins. I got a few stupid replies about it just being the vignette slider in LR, but I started emailing them Mark's website pages on the issue and this thread. I'm hoping it catches their attention, leading to an eventual fix. It once took Thom Hogan to physically visit Nikon to get them to realise they were doing something destructive to the old DSLRs that caused another similar image quality issue that was eventually fixed. I can't even remember what it was but it was way back in the day...

Attached Thumbnails

• 
• 

Edited by primeshooter, 09 December 2024 - 11:20 AM.