Nikon service advisory on D810 "white spots

[Paul2660]

I noticed today, on Nikon Rumors, that Nikon has released a service advisory for the D810 for white spots on longer exposures.  Per Nikon certain serial numbers will need to be sent back to Nikon.  Mine is one of them and have noticed the white spots.  I have read on different blogs, some people have been told by Nikon, that the fix involves firmware.  I don't understand if it's firmware why the body has to go back to Nikon.   Here is the link to Nikon's site with the advisory:

http://www.nikonusa.com/en/Service-And-Support/Service-Advisories/hyvanded/Technical-Service-Advisory-for-Users-of-the-Nikon-D810.html

What's even more interesting to me is that the D800e I own has also always done this on longer exposures with the long noise reduction off.  This needs to be off if you are stacking for night exposures as the long exposure noise reduction does not run in a buffer mode in the background instead locking you out of the camera until it's run.  

I wrote about this issue back in May or so of 2013 and opened calls with Nikon on this issue, but never go anywhere.  However it seems D810 owners have complained enough to Nikon to offer a fix?  I would love to see a fix on the D800e, as the spots are exactly the same.  They appear as a faint grain and will effect the overall quality of a long exposure.  

http://photosofarkansas.com/2013/08/02/d800-reticulation-issues-during-night-photography-white-dot-problems/

It's good that Nikon is "fixing" this issue, but I would like to learn more about it as to what the actual fix is as if it's only firmware, I would rather not send the body off.  

Paul

[BernardLanguillier]

Paul,

I would try Iridient Developper if I were you (and if you are on OSX).

I am not in front of my Mac now and cannot remember the exact name of the noise reduction type that works on this, but it used to do wonder on the many white pixels I experienced with my Mamiya ZD files with no visible side effects. It is in the sharpness/noise reduction tab.

As far as this issue goes, I have not seen it with my D810, but I haven't really done any long exposures with it. Mine also appears to be in the defective batch though.

I personally like the way they are dealing with this issue, it seems they have learned from past mistakes with handling this kind of situations. We can of course say that no defect in the first place would be preferable, but the reality is that fully validating the usage envelope of such complex DSLRs where most components are brand new... is nearly impossible at affordable price points.

Cheers,
Bernard

[Manoli]

This is another link via Nikon Rumours to a distributors memo that lists (a) firmware and (b) NEW Pixel Defect Compensation - by revised service adjustment software (?)

Apparently a 7-10 day turnaround in the USA.

[fdisilvestro]

The white spots are a rather normal thing in image sensors during long exposures. This might be (just speculation) that Nikon's "Hot-pixel suppression algorithm" (or star-killer algorithm) was not active, which could be actually a good thing. This should be easily fixed with a dark frame subtraction.
Ideally, this should be a configuration option (if you want the hot pixel suppression or not), but I doubt they will ever do it.

the long exposure noise reduction does not run in a buffer mode in the background instead locking you out of the camera until it's run.  

Paul

The Long exposure noise reduction based on dark frame subtraction cannot run in a buffer mode in the background. It needs to capture a new frame with the same exposure time as the original image.

[Paul2660]

Actually I would disagree with the commonality of white spots.   I have used Canon:

6D, 7D, 5D MKII2, 5D MKIII, and 1D MKiv, Canon will give a few large white spots, much larger than a single pixel, but nothing as bad as the Nikon's do.   Canon is an easy fix when you get the random large white spots.  With Nikon the spots are in the thousands, and not easily handled. 

Nikon D800, D800e and now D810 all show this same problem, on long exposures with "long exposure noise reduction" off.  For stacking it has to be off.  I have been curious if the D4s shows it, but the entry point for that camera is too high for me right now.

I am also glad to see they are fixing it, and Manoli, thanks for the info, as that does help.  It seems that there is some form of suppression that has to be coded in.  It's strange that some bodies made it out with the fix. 

Bernard, I found that Capture One, Pro with the "single Pixel noise slider' will remove most of the problematic white spots, without effecting the details of the shot.  LR takes a more pronounced noise reduction across the whole file, thus you have to run 2 sets of shots, sky, and one for the rest.

Again, I realize I am in a very small minority, I am working at night, for hours, shooting stacks, at 1' 40" to 4'.  Believe me it's a thermal issue for sure as in the winter you don't see it, but once the summer gets here, and the ambient heat builds up, the D800 and D800e will show this very harshly.   

Paul

[fdisilvestro]

Actually I would disagree with the commonality of white spots.   I have used Canon:

6D, 7D, 5D MKII2, 5D MKIII, and 1D MKiv, Canon will give a few large white spots, much larger than a single pixel, but nothing as bad as the Nikon's do.   Canon is an easy fix when you get the random large white spots.  With Nikon the spots are in the thousands, and not easily handled. 

It might be, but remember that no camera manufacturer give access to a true sensor raw dump. The various "Raw" formats (NEF, CR2) etc are processed files, some more heavily than others. Long exposures with no noise and no hot pixels? We're not there yet, I'm afraid

[PhotoEcosse]

I explored this issue further with Nikon UK (having received their e-mail with the Service Advisory) and established three things:

1, The "cure" consists of altering some of the sensor settings and installing adjusted firmware.

2. There will be no default "cure" in any regular future firmware update.

3. The "cure" will not be total. It will merely reduce the problem.

Now I have to decide whether the problem is likely to bother me enough to warrant sending the camera away for adjustment. Possibly not for my own use but perhaps to maintain the re-sale value. At least Nikon UK have supplied a post-paid label to use.

[Bart_van_der_Wolf]

It might be, but remember that no camera manufacturer give access to a true sensor raw dump. The various "Raw" formats (NEF, CR2) etc are processed files, some more heavily than others. Long exposures with no noise and no hot pixels? We're not there yet, I'm afraid

I assume it is related to the Blackpoint offset clipping on Nikon Raw files, whereas e.g. Canon uses no blackpoint clipping. Apparently that Blackpoint clipping level is set deeper inside the Nikon parameter settings than the normal Firmware can address.

Cheers,
Bart

[trevarthan]

Actually I would disagree with the commonality of white spots.   I have used Canon:

6D, 7D, 5D MKII2, 5D MKIII, and 1D MKiv, Canon will give a few large white spots, much larger than a single pixel, but nothing as bad as the Nikon's do.   Canon is an easy fix when you get the random large white spots.  With Nikon the spots are in the thousands, and not easily handled. 

Nikon D800, D800e and now D810 all show this same problem, on long exposures with "long exposure noise reduction" off.  For stacking it has to be off.  I have been curious if the D4s shows it, but the entry point for that camera is too high for me right now.

Paul

Why does it have to be off for stacking? I've been using LENR ever since day two with my D810 because I noticed the noise right after my first night shoot. It's annoying, for sure, because LENR adds time to the exposure (a lot of time, sometimes, which is weird - not sure why the time varies), but it works fine for one off exposures. I haven't tried stacking yet because the SDK isn't out for the D810 yet so we don't have remote control.

[Paul2660]

If you stack, and use Long exposure noise reduction, you will get a gap.  Example, for every 2 minute exposure you take, then the camera takes a 2 minute dark frame, so your night shot will be gaped.  You can use software like "Startracer" to close these gaps, however this is where the white dots become very problematic.  Since the dots are always in the the same place, being that they are stuck pixels. when you run a gap closing software, the dots will then move and you get a series of dotted lines.  Thousands of them actually. 

Shooting in stacked series, even if you don't ever stop the camera creates small gaps anyway, so the trails when printed large or viewed at 100% will show a distinct start and stop point.  "Startracer"  works great to give you a much smooth and continuous trail, more pleasing to the eye and in a print.  Even leaving the camera open for one long exposure, you get a series of gaps.

As I shoot with the moon, for illumination, you have to stop the camera at times to see if the moon has started to create flare in the shots, as with the 14-24, this is a very common problem.  You also need to stop the camera every hour or so and shoot one series at around 1/500 of second as this helps to clear out stuck pixels, of the standard nature, red green blue, etc.  It won't effect the white one.  Also as the moon moves across the sky, you need to possibly adjust your exposure time, less or greater.  All of these things create gaps, that "startracer" fixes.  To be honest, if the "Startracer" code was not available, then this style of shooting would not work IMO. 

It's pretty apparent that Nikon fixed this issue in later models of the D800e and D800, with a similar fix, as I have seen examples of D800e night shots taken with cameras less than 6 months old, that don't have this issue.  However Nikon did not offer a fix to the D800e family.  Nikon's service advisory also implies this problem only happens in 1:2 mode, which is not true, as I only use the full frame FX mode when shooting at night.  No doubt this was a fix applied to later D800e cameras that was a "silent" fix.  Very common in the entire industry.  Thankfully more folks are working with night exposures and someone that "mattered" called this out to Nikon.  I complained about this same issue on the D800e for months, with Nikon but never got anywhere.   

Paul

[trevarthan]

If you stack, and use Long exposure noise reduction, you will get a gap.  Example, for every 2 minute exposure you take, then the camera takes a 2 minute dark frame, so your night shot will be gaped.  You can use software like "Startracer" to close these gaps, however this is where the white dots become very problematic.  Since the dots are always in the the same place, being that they are stuck pixels. when you run a gap closing software, the dots will then move and you get a series of dotted lines.  Thousands of them actually. 

Shooting in stacked series, even if you don't ever stop the camera creates small gaps anyway, so the trails when printed large or viewed at 100% will show a distinct start and stop point.  "Startracer"  works great to give you a much smooth and continuous trail, more pleasing to the eye and in a print.  Even leaving the camera open for one long exposure, you get a series of gaps.

As I shoot with the moon, for illumination, you have to stop the camera at times to see if the moon has started to create flare in the shots, as with the 14-24, this is a very common problem.  You also need to stop the camera every hour or so and shoot one series at around 1/500 of second as this helps to clear out stuck pixels, of the standard nature, red green blue, etc.  It won't effect the white one.  Also as the moon moves across the sky, you need to possibly adjust your exposure time, less or greater.  All of these things create gaps, that "startracer" fixes.  To be honest, if the "Startracer" code was not available, then this style of shooting would not work IMO. 

It's pretty apparent that Nikon fixed this issue in later models of the D800e and D800, with a similar fix, as I have seen examples of D800e night shots taken with cameras less than 6 months old, that don't have this issue.  However Nikon did not offer a fix to the D800e family.  Nikon's service advisory also implies this problem only happens in 1:2 mode, which is not true, as I only use the full frame FX mode when shooting at night.  No doubt this was a fix applied to later D800e cameras that was a "silent" fix.  Very common in the entire industry.  Thankfully more folks are working with night exposures and someone that "mattered" called this out to Nikon.  I complained about this same issue on the D800e for months, with Nikon but never got anywhere.   

Paul

Is this only a problem with star trails? Would a stack at night of a cityscape pose a problem? Either way, it sounds worth my time to send my D810 in for service, but I'm curious.

[Paul2660]

That is a good question, the actual dots themselves, can be removed in LR or Capture One, I have not tried Nikon NX-D yet.  Capture One to me with the single pixel noise reduction slider does by the best job, as it will remove 98% of them without really effecting the overall image.

The dots actually appear as a reticulation look, i.e. over process film.  They give the image an overall grainy look. 

I have sent my camera in as of this morning to the LA Nikon service center, so I will report back, when I get mine back.

Overall, I would tend to agree that this issue will only show up in a very limited application of long exposures, and if you turn on Long exposure noise reduction, it seems to help.  However you have to remember, that as your exposures approach longer times, 5 minutes, or so, you are really starting to waste battery, as the corresponding dark frame will also be 5 minutes and you are totally locked out the camera until this is done. 

Canon has been able to figure out a way to run this in the background, so the camera is usually able to fire again.  You will eventually buffer out with Canon also, however you should be able to get a long series of 5 minute exposures without ever maxing out the buffer.

I am positive that this has been address above board with the service fix, unlike the apparent fix for the D800e.

Paul

[Ligament]

I explored this issue further with Nikon UK (having received their e-mail with the Service Advisory) and established three things:

1, The "cure" consists of altering some of the sensor settings and installing adjusted firmware.

2. There will be no default "cure" in any regular future firmware update.

3. The "cure" will not be total. It will merely reduce the problem.

Now I have to decide whether the problem is likely to bother me enough to warrant sending the camera away for adjustment. Possibly not for my own use but perhaps to maintain the re-sale value. At least Nikon UK have supplied a post-paid label to use.

I'm confused by all the above. Why can't we install the firmware ourselves?

More concerning, however, is pint #3. If the cure is not total, why not fully replace the camera? These things are only a month old now. We should expect a fully functional camera, or fixes that make it fully functional. Do they expect us to sell our 1 month old cameras to buy a new one?

[Bart_van_der_Wolf]

I'm confused by all the above. Why can't we install the firmware ourselves?

Because of the "altering some of the sensor settings"?

More concerning, however, is pint #3. If the cure is not total, why not fully replace the camera?

Maybe because it's impossible to fully cure, just reducing is maybe the best one can expect?

Cheers,
Bart

[aragdog]

I did not jump to the 810.  It is sort of like when I review a new restaurant.  They need a few months to sort out the kinks.  So I am no longer jumping to the newest thing.  I will wait until all is ok and then probably jump in.  I am selling my D800E and will just hang on for a couple more months.

[BernardLanguillier]

Because of the "altering some of the sensor settings"?

Maybe because it's impossible to fully cure, just reducing is maybe the best one can expect?

I think that they need to tune the fix to actual behaviour of each and every sensor part, some sort of pixel mapping.

I hadn't seen any issues with mine but I had the fixed applied on Saturday, 30 minutes, done.

Cheers,
Bernard

[Paul2660]

Just wanted to add an update. I received my D810 back yesterday and spent some time with it last night. The fix is most impressive. I took a series a 2 minute exposures at ISO 160 and 200. Very clean with 10 15 problem spots instead of thousands.

The fix works.

Paul

[photodan]

Just wanted to add an update. I received my D810 back yesterday and spent some time with it last night. The fix is most impressive. I took a series a 2 minute exposures at ISO 160 and 200. Very clean with 10 15 problem spots instead of thousands.

The fix works.

Paul

Ditto for my experience.

I got my D810 back from Nikon amazingly less than one week after UPS'ing it to them (I live near Los Angeles where one of their service centers is located). My before and after test shots indicated the exact same results as Paul.

Additionally I noticed that the AF-fine tuning for my two lenses (Nikon 70-200/4 and Sigma 35/1.4) needed adjusting. They now are closer to the central '0' value than before, and the test shots indicate results that are as sharp and possibly sharper than before, outside the very central area. The service notes indicated 'chkd lens mount', not 'adjusted lens mount', but I do think something was done to tweak the camera. In any case it was definitely worth me sending in the camera!