Nikon D810 - thermal noise issue

O

Ozei

Messages
3,143
Edit My Images
No
Anyone who owns the D810 might want to read this.

Photography Life have found a thermal noise issue with three D810 bodies. Their article includes instructions on how to reproduce it / find out if your camera is affected.
 
I can see a few on yours (1st link) - there are so many in PL's it looks like dust on my monitor.

I've never seen a single one on my D800E - even at 30 seconds with LENR off.
 
Last edited:
check the second image in the sky, it seems to depend on what the colour is if they show up so much on the blue you can see a lot of them. I've had it in a lot of images and there were loads of comments when the camera first came out.
His image does seem rather bad but he did state he processed it to show it more than it actually is.
 
My D810 has it at 30sec ISO 100.
But, TBH you have to be pretty anal for it to be a major issue. Even pushing shadow recovery 100% and clarity they are not very obtrusive. I imagine they will be even worse at longer exposures of darker scenes.
 
My D800E has far more hot pixels than my D800 had for some reason. It's by far the worst camera I've owned for this? Am I bothered? Nope. If I'm not doing anything that relies heavily on immediately starting the next exposure.. and let's be honest... that pretty much means stacked star trails or time lapse... then I'll just select a user profile where I have set long exposure NR on. If I do need to open the shutter immediately again... then I'll shoot a dark reference frame and use one of the many pieces of software available to remove hot pixels.

It's only an issue with long exposures, and to some degree, al cameras do this. I can assure you my D800E is FAR worse than the D800E example image used in that article.

Why do people obsess and start the spreading stupid rumours that new cameras have "faults" when in reality.... it just suffers from thermal noise.. something pretty much all CMOS sensors will do if left on long enough. They'll be pasting up examples of amp glow next, and calling THAT a fault.

I mean +50 clarity FFS! Those dots would be hardly visible with no clarity, and who the hell uses +50 clarity except the HDR and Flickr crowd anyway? It looks crap.

I absolutely do not believe that D800E example either, especially if there is +50 clarity added. Looks like they left the Long Exp NR on with the D800E to me, as mine's probably worse than the D810 example they posted here.

I'll post up a version of mine later when it gets dark. I bet with +50 clarity it an''t so pretty either.... we'll see later.

Seriously.... scaremongering bull****.
 
Last edited:
Pookeyhead said:
My D800E has far more hot pixels than my D800 had for some reason. It's by far the worst camera I've owned for this? Am I bothered? Nope. If I'm not doing anything that relies heavily on immediately starting the next exposure.. and let's be honest... that pretty much means stacked star trails or time lapse... then I'll just select a user profile where I have set long exposure NR on. If I do need to open the shutter immediately again... then I'll shoot a dark reference frame and use one of the many pieces of software available to remove hot pixels.

It's only an issue with long exposures, and to some degree, al cameras do this. I can assure you my D800E is FAR worse than the D800E example image used in that article.

Why do people obsess and start the spreading stupid rumours that new cameras have "faults" when in reality.... it just suffers from thermal noise.. something pretty much all CMOS sensors will do if left on long enough. They'll be pasting up examples of amp glow next, and calling THAT a fault.

I mean +50 clarity FFS! Those dots would be hardly visible with no clarity, and who the hell uses +50 clarity except the HDR and Flickr crowd anyway? It looks crap.

I absolutely do not believe that D800E example either, especially if there is +50 clarity added. Looks like they left the Long Exp NR on with the D800E to me, as mine's probably worse than the D810 example they posted here.

I'll post up a version of mine later when it gets dark. I bet with +50 clarity it an''t so pretty either.... we'll see later.

Seriously.... scaremongering bull****.
Click to expand...


I said this to him on his page and was told my issue is just hot/stuck pixels and the d810 is different.
Could this be from removing the lowpass filter. If this does end up being worse than a d800E which i found worse than my d800 it could explain it.
 
Amp noise will be random, a hot pixel should be constant.
Shouldn't have anything to do with the AA filter.

IMO, it's probably a result of trying to get maximum resolution from the sensor. Along with the modification/removal of the AA filter it wouldn't surprise me if they are doing progressively "less" manipulation in generating the image files.
 
pmac said:
I said this to him on his page and was told my issue is just hot/stuck pixels and the d810 is different.
Could this be from removing the lowpass filter. If this does end up being worse than a d800E which i found worse than my d800 it could explain it.
Click to expand...

A "stuck" pixel will be present all the time, regardless of exposure time or temperature. A hot pixel, or thermal noise are caused by extremely small differences in the sensitivity of individual photosites being exaggerated due to the extremely long exposures and is entirely variable depending on exposure time and ambient temperature. If you could cool your sensor sufficiently, you'd see a massive reduction in this thermal noise. Increase the ISO (the analogue gain) and this made worse still. In very hot conditions it can get very bad for another reason, thermal runaway - caused by the photosites heating excessively. The hotter a semiconductor gets, the harder it works... the harder it works, the hotter it gets.

If you had stuck pixels they'd be present all the time, regardless of conditions or exposure.

The low pass filter has nothing to do with it, as the noise is a product of the sensor, not the optical path. In fact, if exposure times and temperatures were identical, you'd get EXACTLY the same noise pattern even with your lens cap on. That's how long exposure NR works. It activates the sensor for the same amount of time but with the shutter closed, and then compares the frames. Where there is a hot pixel on the image created with the shutter closed (the dark reference frame), the corresponding pixel in your live image is remapped to the value of an adjacent pixel, and thus removes the noise.

Incidentally.. as your camera ages, you'll get more thermal noise. All cameras from new have their hot pixels mapped out at the factory, but sensors degenerate, and the differences between discrete photosite output continues to vary. If you have a camera that has no thermal noise... don't be smug.. wait a year or so... it will start to happen.


It's nothing to worry about.
 
Last edited:
Also.. if those images in the OP's link are at ISO100... then they are an extreme crop of a tiny part of the image, as neither camera would produce such non thermal noise at this ISO at full frame.

Here's my D800E at 30 seconds, ISO100 and 21 degrees ambient temp.

I've done exactly what they did... i.e, +2 stops of EV, and +50 Clarity. I never even noticed they'd added sharpness as well!!! Who adds +50 clarity and +50 sharpness to a D800/D810 image. You don't need to add any!

To remove any non thermal noise from confusing the issue (something Photographylife should have done if their testing methodology was sound) I have left the lens cap on... the ONLY noise in this image is the thermal noise once exaggerated with +50 clarity, +50 Sharpening, and +2 stops EV recovery. I've also cropped into the same amount by comparing non thermal noise size at ISO100. Yes.. they neglected to say that the images are cropped in significantly I notice. Why did they even bother listing the aperture? What's the aperture got to do with thermal noise.

Eh05S1c.jpg



Here's the same raw file more realistically portrayed.. NO exposure compensation (I'm a photographer... why would I need +2 stops of EV compensation in post?).... No clarity... because it sucks, and +25 sharpening, the amount added by default.

oaoIjYE.jpg



Nothing... nada.. zip. Yes.. these are the same raw file.


Ill thought out, poor testing methodology, and blatant scaremongering to get some web hits.

Utterly discredited by common sense and practical demonstration.

So there you go... under the same ridiculous processing conditions my D800E is as bas as their D810

Big fat so what! Photographylife.... stop processing your images like a retard! Your D800E is either exceptionally well mapped (for now... it will throw up thermal noise with time.. don't worry), or some idiot left the long exp NR on. My D800E has only shot 1800 frames from new incidentally.
 
Last edited:
Ozei said:
Anyone who owns the D810 might want to read this.

Photography Life have found a thermal noise issue with three D810 bodies. Their article includes instructions on how to reproduce it / find out if your camera is affected.
Click to expand...

Thanks - but luckily doesn't effect me as I don't do long exposures!
 
That doesn't seem to be the same problem. There appears to be light spots across the frame when shooting in 1.2x (30x20). This thread just shows fairly normal thermal noise that most cameras would, and do display with long exposures and extreme processing.

BTW... your link wasn't working.. I think you meant this one?

https://nikoneurope-en.custhelp.com/app/answers/detail/a_id/61871

If it is the same problem (which is unlikely as they didn't mention it only happening in 1.2X crop mode), then I think it's a case of massive over-reaction akin to the D600 oil issue.

I see nothing on the Photographylife images that almost every camera I've owned wouldn't do under the same conditions. My D800 does it. I think this has to be far more serious that displayed in this thread to warrant a free return to base and repair. If the images posted in the OP were without any processing, then that would be a different matter.

Does seem an odd coincidence though. I'm wondering if the publishers of this article have read about this fault (of which there's been grumblings for a few days now) and were determined to find a fault in theirs.

If I had a D810 that only showed light thermal noise when processing so extremely then I'd certainly not be sending it back, as all my cameras do that.
 
Last edited:
Checked on my serial no today after an email from Nikon UK, and I've got one that needs to go back to Nikon.
They've sent anpost paid label just have to drop it into a post office.
But they're saying two to four weeks for repair!
Two questions, is that timescale acceptable for a camera that I've had for only three weeks? And is it worth sending in to repair a function that I'm probably not going to use?
 
Ozei said:
No, I meant the link I used in post 14 - to the serial checker. Select English in the dropdown & click Submit, then type in the D810 serial at the next screen & hit Check - the answer that comes up will tell you if the camera has the 'issue'.
Click to expand...

Sorry.. thought you were trying to link to the article itself.

Cowleystjames said:
Checked on my serial no today after an email from Nikon UK, and I've got one that needs to go back to Nikon.
They've sent anpost paid label just have to drop it into a post office.
But they're saying two to four weeks for repair!
Two questions, is that timescale acceptable for a camera that I've had for only three weeks? And is it worth sending in to repair a function that I'm probably not going to use?
Click to expand...

Have you actually checked it has a fault, or just assuming it has because of it's serial number? What Photographylife's images are showing in this thread is actually pretty normal. All my cameras do this. Would be a shame to send your camera way for up to four weeks for nothing.
 
Last edited:
Pookeyhead said:
Sorry.. thought you were trying to link to the article itself.



Have you actually checked it has a fault, or just assuming it has because of it's serial number? What Photographylife's images are showing in this thread is actually pretty normal. All my cameras do this. Would be a shame to send your camera way for up to four weeks for nothing.
Click to expand...
The work I do with it, landscape, micro, portrait, shows no fault even blowing up to A2 print.
Low light is almost as good as my D4s and still very acceptable.
I don't push this camera to the same extent as the testing done,simply because I don't do that. It's not my remit.
Most extreme, really low light sporting etc, is on the D4s.
Thanks Pookeyhead, you've made my mind up. I'll keep hold of it, it puts bread on the table.
 
Cowleystjames said:
I don't push this camera to the same extent as the testing done,simply because I don't do that. It's not my remit.
Click to expand...


If you see thermal noise on a 30 second exposed raw file with no processing, then sure... that's not normal but if you have to do what the article in this thread did - +2 stops exposure recovery, then +50 clarity, then +50 sharpening.... then all my cameras here show the same amount of thermal noise. That's just normal.

Looks like another D600 oil fault style panic over nothing. I have a D600. Yes it had oil on the sensor. I cleaned the sensor. LOL
 
Does this sound a bit like panic stations at Nikon? After the D600 debacle, it's understandable.
When I get home I'll check for the black dot but if I ever do a long exposure then I'll turn Lenr on, after all, that what it's there for.
 
@Cowleystjames

Just grab the body.. leave the body cap on. Set to ISO 100, and 30 seconds with the camera on manual. Ensure that long exposure noise reduction is OFF. Make a blank 30 second exposure. Load up the raw in lightroom or whatever, and look at it at 100%. Are there any WHITE dots? If not... relax.. nothing wrong with your camera regardless of whether it has the black dot in the tripod bush or not.

If you pump up the exposure, clarity and sharpness... you'll probably see some white dots of thermal noise... Any camera will do that though. My D610 does it, my D800E does it and my D600 does it.
 
Last edited:
Just "black frame" tested mine and I get ZERO noise at a 1min exposure ISO 100 (and I mean NONE).
I also did another 30sec ISO 100 "scene test" and I found zero issues. But this test differed from the first "scene test" (where I found some spots) in that I had to use f/45 for a proper exposure. In the first test it was a much darker situation with the aperture wide open and the resulting image still somewhat underexposed.

I also tested in Dx mode (not the 1.2x) with the same results.

This leads me to the conclusion that the issue is really photon shot noise... That's not really surprising; it's why ISO tests in good light don't really tell much about low light performance.

This also leads me to think that I won't be sending mine in... even if some pixels are more affected by photon shot noise, the only way they can "fix" that is to map those pixels out. And in the end, that's going to be a compromise on IQ in every other situation (although probably minuscule).

Hmm, where's my black paint pen.....
 
Last edited:
Cowleystjames said:
Checked on my serial no today after an email from Nikon UK, and I've got one that needs to go back to Nikon.
They've sent anpost paid label just have to drop it into a post office.
But they're saying two to four weeks for repair!
Two questions, is that timescale acceptable for a camera that I've had for only three weeks? And is it worth sending in to repair a function that I'm probably not going to use?
Click to expand...

The only thing is if you at any time want to move this camera on,everybody going to ask "has it had the fix"
 
Ozei said:
What does the serial number checker say about yours? (link in post 14)
Click to expand...
That it's affected.
I've uploaded the 1min BlackFrame. Raw capture w/ noise reduction off. All settings in LR neutral/off except shadows +100 and sharpening +100. I would have expected much worse with those settings but it's very clean. If I then push clarity to ~ 50% I start to get some noticeable noise.


D810 BlackFrame Exposure by skersting66, on Flickr

Full size can be viewed on Flick'r
 
Last edited:
Another bloody panic over nothing this. Steve... I bet if you did to your RAW what Photographylife did (and I did in my test) you'd get noise. It's normal.


I bet the only ones complaining are the ones over processing their stupid light painting images or something.

It's the "Oh no I have to clean my sensor, give me a refund" debacle all over again.
 
Pookeyhead said:
Steve... I bet if you did to your RAW what Photographylife did (and I did in my test) you'd get noise. It's normal.
Click to expand...
I did most of what they suggested, just left out the clarity adjustment. And even pushing clarity to 100% it isn't bad at all. But push the exposure up a couple stops as well and the image looks like a confetti storm...
 
sk66 said:
I did most of what they suggested, just left out the clarity adjustment. And even pushing clarity to 100% it isn't bad at all. But push the exposure up a couple stops as well and the image looks like a confetti storm...
Click to expand...

Every camera I own would exhibit quite apparent thermal noise if you do that. While mine wasn't exactly a confetti storm, the fact that you also exposed for 1 minute instead of the 30 seconds Photographylife did would probably explain that.
 
I'm not sure that's really thermal noise you see when you bump the exposure on a black frame. I'm thinking that's more just a program trying to boost "color information" when it's got nothing to go off of.
 
Last edited:
The longer the exposure, the greater it is, which would indicate it's a thermal condition of some sort, and not merely the effects of amplification.
 
http://www.dpreview.com/forums/post/54230332

Interesting.

What are Nikon playing at by laying down and rolling over so the camera nerds can tickle their tum tum... There's sod all wrong with these cameras. Nikon should just issue a statement saying "Stop processing the sh1t out of your images".
 
Last edited:
Well my new 810 arrived today and it has the problem according to its serial number.

I wonder if the UK service centre will deal with cameras bought through Panamoz?

Great :(
 
Last edited:
Dave Semmens said:
Well my new 810 arrived today and it has the problem according to its serial number.

I wonder if the UK service centre will deal with cameras bought through Panamoz?

Great :(
Click to expand...


Why so upset. Why not test if it actually has anything wrong with it before you send it anywhere? Why send it back because of a number?

Not only that, you'd be doing others a massive favour too, as so far, the only pictures I've found of this "fault" are Photographylife's, and all they've demonstrated is completely normal behaviour for a digital camera. If yours behaves as expected, then that's two D810s in this thread that according to serial numbers, have a fault, but probably don't.

Body cap on. Manual. ISO100. Long Exposure NR OFF. Take a shot. See anything? No? Camera fine. In fact... do as Steve did and do a 1 minute exposure. Still OK? Definitely fine.
 
simonblue said:
The only thing is if you at any time want to move this camera on,everybody going to ask "has it had the fix"
Click to expand...

This is still a valid concern

Pookeyhead said:
Why so upset. Why not test if it actually has anything wrong with it before you send it anywhere? Why send it back because of a number?

Not only that, you'd be doing others a massive favour too, as so far, the only pictures I've found of this "fault" are Photographylife's, and all they've demonstrated is completely normal behaviour for a digital camera. If yours behaves as expected, then that's two D810s in this thread that according to serial numbers, have a fault, but probably don't.

Body cap on. Manual. ISO100. Long Exposure NR OFF. Take a shot. See anything? No? Camera fine. In fact... do as Steve did and do a 1 minute exposure. Still OK? Definitely fine.
Click to expand...

Right - tried it @ 30 seconds with NR off and iso100- until my one and only lens arrives later this week it is about the only type of photo I could take :)

The only thing I did different was to close the view finder flap as I have read that is a good idea on long exposures.

Blacker than a really black thing. I cannot see anything even if I ramp up clarity, sharpening etc.

So no problem here :)

Dave.

Edit - converted to jpg and uploaded to flickr

DSC_0001 by davepsemmens, on Flickr
 
Last edited:
Pookeyhead said:
The longer the exposure, the greater it is, which would indicate it's a thermal condition of some sort, and not merely the effects of amplification.
Click to expand...
Yes and no I think... I've heard that ACR (LR/PS) has automatic hot pixel mapping so I used Capture NX-D this time. These are 100% screen shots of the D810 30 sec and 60 sec black frame exposures. Both have had their exposures pushed 4 stops. BTW, these shots were taken back-to-back so the 60sec would be more representative of a 90sec exposure for thermal noise.

While some pixels do show as "hotter" in the 60sec exposure, others have just gone away (I highlighted a couple spots). And a lot of the color noise has kind of "morphed" a little. (BTW, these are 60% qual jpegs in order to post them)
 
Last edited:
Gary Coyle said:
With due respect to those who assume theres no problem, then why are Nikon offering a fix to a non existant issue??????
Click to expand...
By offering "a fix" right away; maybe they are hoping to prevent it being another issue like the D600 oil spot? They also include this disclaimer:
"As image resolution and sharpness have been increased with the D810, even after cameras are serviced to resolve this issue, bright spots may be more noticeable in exposures longer than 30 s than they were in images captured by previous cameras with the same exposure time."

Some cameras may have an issue I'm not seeing... And maybe some of my other cameras are worse. In the DpReview link someone posted a comparison to a Df image which looks a lot worse. That's pretty hard to believe IMO, but whatever...
 
So we all accept there is an issue but some see it as more of an issue than others, would it bother me, probably never as i wouldnt go looking for it and nor do i shoot under the required conditions which makes it show, cest la vie
 
You must log in or register to reply here.
Back
Top