D850 noise

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

On paper that is a great low light camera with a super dynamic range even at high ISO. Is it possible you are not letting in enough actual light in terms of aperture and shutter speed? ISO is not what causes noise, it just gets associated with low light or low exposure which are some of the true causes. Cropping also can make noise worse by magnifying it. Maybe see if there are situations where you can open the aperture more or lower the shutter speed a little and see if it makes a difference. Of course it's better to get the shot even if it's noisy, but there might be some situations that you don't need as fast a shutter or can open up a touch.
 
I can't get original image uploaded as the smallest file size is over 5MB
Here is the cropped image
d850 noise cropped.jpg
You can only see EXIF info for this image if you are logged in.
550mm f/7.1 1/2500 ISO6400
 
Last edited:
I find I'm getting significant noise at ISO6400 (more than when I use same lens/settings on my D7200). Do other D850 users experience the same?

SIGMA 150-600mm @ f/8 1/2000 - 1/4000
Whether noise is objectionable or not is very subjective and depends on what you are trying to achieve with the image. I have both the D850 and D7200, both of which are the highest rated Nikon sensors in the FX and DX categories respectively and both perform well at higher ISOs. Here's the DxOMark measurements. The green to red bar on the right is an SNR guideline with green showing acceptable SNR ratio and Red not so much. Personally, my threshold is ISO 1600, but of course, I'll go higher if I really want the shot.

SNR D850 vs D7200.JPG
You can only see EXIF info for this image if you are logged in.
 
Thanks Strodav. I'm not a pixel peeper but on editing, to date it seemed like for me the D7200 was performing slightly better than the D850.... guess I'll have to give it more time. When I try to zoom in to the image I uploaded, I am not seeing the same effect as on my original images.
 
Thanks Strodav. I'm not a pixel peeper but on editing, to date it seemed like for me the D7200 was performing slightly better than the D850.... guess I'll have to give it more time. When I try to zoom in to the image I uploaded, I am not seeing the same effect as on my original images.

When you crop, the noise pixels get larger so the image can look noisier ( https://backcountrygallery.com/does-cropping-make-your-photos-noisy/ ). Also, noise can show up in the shadows, but not the highlights of your image. Maybe that's why your D7200 "looks" cleaner than your D850. BTW, I love the D7200; small, light with great IQ. Now, I mainly use it for family events and family vacations. Even though the vast majority of my shots are taken with the D850 and D500, I still can't part with my D7200.
 
I use DeNoise. The image submitted was without noise removal. I was just getting the impression that the 45MP D850 had more noise than the 24MP D7200 under similar conditions

Well, at least it is not a surprise that thay are on approximately the same level compared with light monsters like the D4s. I shot a D7200 until recently and have now a D850.
If we forget about software differences between camera models, the main criteria for producing noise is the area of a single pixel that is able to catch the light. This is what influences the SNR (Signal to Noise Ratio) of a sensor most. And the pixel area size of a D7200 and D850 very similar with the D7200 having the smallest pixel area size of the Nikon DSLRs. The fact that you have the feeling that the D850 creates more noise may be a result of the fact the noise the different cameras produce look somewhat different. This is partly due to different sensor technologies or generations, but also software differences. That said, after having used quite a number of camera models and having started to play with what the software developpers called denoising in times of Viev NX2 nd Capture NX2, one of the conclusions was that as noise that looks worse to us when looking at an unprocessed image, sometimes is easier to remove, because its pattern is easier to recognize by a software algorithm. Looking back I can say that - as someome still shooting a old D4s:
  • I was surprised to see the noise level that a D850 produces considering that it is the much more modern camera, compared to D4s, D7200 and alike
  • I found that the results with up-to-date denoising software are pretty impressive and that the denoising of D850 RAWs shows usually better results than similarly noisy photos from the D7200
ISO 6400 is pretty tough for a 45 MPixel sensor with pixel sizes this tiny. And this is the reaon I keep my old monster. ISO 3200 is the absoute limit I shoot my D850, whereas with todays denoising software I have get - depending on the surrounding conditions - good results at ISO 12800 with occasionally pushing it even to 25600.

Another effect that takes effect in this context is the extended sensitivity for motion blur with high res / small pixel sensors. I can'T recall it, but there were some explanations here and @Steve wrote a comprehensive article about it with example photos taken with a ticking metronome. May be you'll take a look ...
 
Using Auto WB will also generate more noise than using the Kelvin temp setting for a situation. Going above ISO 400 will also make for more noise and no gain with using 14-bit RAW capture. There is also variation in how much noise reduction is done by the camera. The Canon 5D was touted for its IQ at ISO 6400 but if one looked at pictures of people taken at that setting they would see skin that had a plastic look as the camera's processing lost a great deal of the tonality.

Nikon has been inclined to have a neutral image file and let the end user process to their taste. Having used many Nikon cameras with poor high ISO performance I also distinguish between chroma noise that is cannot be adjusted with a NR app and one that has more of a grainy appearance such as one encountered with film. When I saw the ISO 6400 files from cameras like the D7200, D750, D800e, and D850, I have been surprised and delighted by how little noise was present even with dark red clothing or dark skinned subjects and with artificial light sources.

If noise is a concern then stay at ISO 400 or less and take longer exposures or make use of "available light" or as W. Eugene Smith stated "Available light is any damn light that is available" and so if flash was an option and improved a shot then he used it without a second thought. Flash is often not used at all and in key situtions it is needed to get the best possible image. With flash I can underexpose the background by up to two f-stops and have the subject properly exposed and this allows me to use a lower ISO setting and a slower shutter speed and to improve separation of the subject from the background (instead of relying 100% on shallow DOF.
 
D850, 500mm, f8, 1/3200, ISO 5000 / 8000

Yes, agree ! With todays denoising technology things like this are possible, but as I said, it depends on the surrounding conditions. The brighter the scene, especially the blurred background the more you can push ISO. The darker it gets the more work the software has to do and the earlier it reaches the limit.
 
I am also wondering if what the OP could be seeing is unfamiliarity of looking at a 45MP image at 100% vs. 24MP at 100%. I know that when I started shooting with higher resolution bodies and looking at the images at 100%, I thought they looked a bit noisier than lower MP camera bodies that I was more familiar with. Then again, I realized that I was seeing a much smaller portion of the image filling my screen. Just a thought.

--Ken
 
Thank you everyone, and yes Ken, that was my perception.
Glad you did not take offense at my comment. I mentioned it because it my my first reaction when I started shooting with a D750. And when I saw prints from the images I shot, I realized that I had no worries. And it did not hurt that a few folks reminded me about viewing images at 100%. I guess I am a creature of habit and get used to what I know.

--Ken
 
I am also wondering if what the OP could be seeing is unfamiliarity of looking at a 45MP image at 100% vs. 24MP at 100%. I know that when I started shooting with higher resolution bodies and looking at the images at 100%, I thought they looked a bit noisier than lower MP camera bodies that I was more familiar with. Then again, I realized that I was seeing a much smaller portion of the image filling my screen. Just a thought.

Yup, fully agree. I shot D7200 and D750 myself before and this is what happened to me when processing the first images of my D850, not mentioning that by now I am shooting a 16.2 MPixel oldie side by side with the D850 :). And your observation also goes in sync with @Steve 's recommendation documented in his video about cropping requiring to drop the max. ISO you shoot at. If you forget about software for while the SNR (signal to noise ratio) of a sensor is more or less directly following the single pixel area size. Basically smaller pixel area size means more noise compared for shooting at the same ISO. On the other hand downsampling helps to eliminate noise or - better said reduce it's visibility.

This is the main reason, why I always recommend to compare sensor performance in terms of SNR for "screen" rather than "print" in the DXO terminology. The "screen" setting is looking at active displaying like with monitors, beamers etc. and focuses on what you experienced with viewing pictures at 100%. The "print" setting is comparing images that are re-sampled to a reference print size ans quality for comparison. The latter usually makes the differences between sensors smaller than they really are.

The diagram shown above to compare the D850 and the D7200 is showing the print comparison and due to the resolution advantage of almost factor 2 the D850 gives you a significant better headroom in terms of noise being visible in prints, because downsampling helps to eliminate noise. If you look at the the same comparison for "screen" you'll see that the D7200 and the D850 are absolutely on par, which is no surprise because the pixel area size of the D7200 DX sensor with 24 MPixel and the D850 FX sensor with 45 MPixel is almost identical.

If, on the other hand you compare a D850 with a D4s, the pixel area size of the latter is almost three times as big, and as a consequence in terms of noise level the oldie gives you about 2 full stops advantage in noise level translated in ISO setting - which is quite a lot.
But this comes at the price that the 16.2 MPixel don't give even anything close to the cropping headroom you have with a D850.

Another aspect is that the te noise created by different sensors may actually look pretty different. Due to the effect described above my D850 (45 Mpixel) creates significantly more noise compared to my D4s (16,2 MPixel) at the same ISO level when comparing a full image of the D4s against a 16,2 MPixel crop of the D850. But at the same time the D4s has more white noise (image looking "mosaicy" with variations of brightness on pixel level) whereas the D850 creates more color noise (image showing RGB sprinkles).
The latter type of noise appears to more distracting to our eyes.

As always, if you hve to gain on one side, you have to pay a price on another and which way to go always depends on what you are trying to do :).
 
I use DeNoise. The image submitted was without noise removal. I was just getting the impression that the 45MP D850 had more noise than the 24MP D7200 under similar conditions
Do you have any NR turned on in camera? I actually like the NR that Nikon uses, I have an older version of Capture that allows me to set the amounts (more specific than the in camera) and I find I rarely need anything else. As said previously, the more you crop the more noise you'll see.
 
Last edited:
I find I'm getting significant noise at ISO6400 (more than when I use same lens/settings on my D7200). Do other D850 users experience the same?

SIGMA 150-600mm @ f/8 1/2000 - 1/4000
How are you evaluating and comparing the noise in the two images? If you're zooming in to 100% zoom level (1:1 pixel view) with both cameras then the noise will look much higher in a D850 as you're zooming in much further to fill the monitor with the D850's high resolution than you would with a lower resolution camera like the D7200.

To compare noise in an apples to apples way the images should be resized to the same pixel dimensions and then viewed at the same zoom level. IOW either downsample the D850 image to D7200 pixel dimensions or interpolate (upscale) the D7200's image to the D850's pixel dimensions. Then compare the two images at the same zoom level in LR or PS or similar tools.

Maybe you're already normalizing the image sizes but a lot of folks stumble on this when they first get a very high resolution camera. They compare noise at the 1:1 pixel view between cameras at very different resolutions which isn't a direct comparison of similar output images.
 
Yup, fully agree. I shot D7200 and D750 myself before and this is what happened to me when processing the first images of my D850, not mentioning that by now I am shooting a 16.2 MPixel oldie side by side with the D850 :). And your observation also goes in sync with @Steve 's recommendation documented in his video about cropping requiring to drop the max. ISO you shoot at. If you forget about software for while the SNR (signal to noise ratio) of a sensor is more or less directly following the single pixel area size. Basically smaller pixel area size means more noise compared for shooting at the same ISO. On the other hand downsampling helps to eliminate noise or - better said reduce it's visibility.

This is the main reason, why I always recommend to compare sensor performance in terms of SNR for "screen" rather than "print" in the DXO terminology. The "screen" setting is looking at active displaying like with monitors, beamers etc. and focuses on what you experienced with viewing pictures at 100%. The "print" setting is comparing images that are re-sampled to a reference print size ans quality for comparison. The latter usually makes the differences between sensors smaller than they really are.

The diagram shown above to compare the D850 and the D7200 is showing the print comparison and due to the resolution advantage of almost factor 2 the D850 gives you a significant better headroom in terms of noise being visible in prints, because downsampling helps to eliminate noise. If you look at the the same comparison for "screen" you'll see that the D7200 and the D850 are absolutely on par, which is no surprise because the pixel area size of the D7200 DX sensor with 24 MPixel and the D850 FX sensor with 45 MPixel is almost identical.

If, on the other hand you compare a D850 with a D4s, the pixel area size of the latter is almost three times as big, and as a consequence in terms of noise level the oldie gives you about 2 full stops advantage in noise level translated in ISO setting - which is quite a lot.
But this comes at the price that the 16.2 MPixel don't give even anything close to the cropping headroom you have with a D850.

Another aspect is that the te noise created by different sensors may actually look pretty different. Due to the effect described above my D850 (45 Mpixel) creates significantly more noise compared to my D4s (16,2 MPixel) at the same ISO level when comparing a full image of the D4s against a 16,2 MPixel crop of the D850. But at the same time the D4s has more white noise (image looking "mosaicy" with variations of brightness on pixel level) whereas the D850 creates more color noise (image showing RGB sprinkles).
The latter type of noise appears to more distracting to our eyes.

As always, if you hve to gain on one side, you have to pay a price on another and which way to go always depends on what you are trying to do :).

To me the only fair real world comparison is comparing the same output resolution at the same size and viewing distance, so I like the 'print' dxo comparison. It's similar to diffraction in a way. You set a diffraction calculator to pixel for pixel and the larger pixels of the low megapixel cameras will win, but set it to print, which adjusts for viewing size at a given resolution and then the diffraction is not so bad on the higher megapixel cameras even with smaller pixels.
 
Personally I have not had an issue in the past when I shot D850's, I will say that any higher ISO and cropped does make it look worse. Steve did a good video about this. I highly suggest Topaz DeNoise and Sharpen AI. They will completely blow you away with how they can improve an already great cameras file.
 
I use DeNoise. The image submitted was without noise removal. I was just getting the impression that the 45MP D850 had more noise than the 24MP D7200 under similar conditions

Having owned both cameras (and still using the D850) I am surprised by this. In my experience the D850 is substantially better at high ISO than the D7200, as would be expected of a full frame compared to a crop sensor. I've never seen that much noise on the D850 at ISO 6400.
 
It looks I have about the same amout of noise at 6400 iso on my D850.
This is a not denoised cropped image, is it ?
Even if I had some very good result at 5000 iso in some cases, I generally try not to go up to 3200 iso because I can get rid of noise but loose too much details to my taste.
Using topaz denoise, or better (I find) DxO stand alone.
 
It looks I have about the same amout of noise at 6400 iso on my D850.
This is a not denoised cropped image, is it ?
Even if I had some very good result at 5000 iso in some cases, I generally try not to go up to 3200 iso because I can get rid of noise but loose too much details to my taste.
Using topaz denoise, or better (I find) DxO stand alone.

I haven't seen that much noise at 6400 on my D850, although I do usually switch to my D750 if I need to use high ISO because it is better than the D850.
 
You experience seems inexplicable to me. I've used the D7200 and D850 (and D7500 and D500) and the D850 is light years better than the DX bodies at high ISOs. I've been surprised time after time at usable ISO 3200 and 6400 images that would be trash on the DX bodies. I can't explain your results at all.
 
Back
Top