D850 vs D810

After using the D850 for over 6 months now, I am noticing significant differences in color accuracy between it and my D810. Which, actually, I noticed immediately after purchasing. I use both cameras on a daily basis in my Real Estate Photography work. I shoot everything in RAW and process in Lightroom. In the beginning, I wasn't happy with the differences as the 810 appeared to have more contrast and much more vibrancy (saturation?) in its colors. However, I have now begun to realize that, in fact the difference is a result of the improved accuracy of the D850, especially in these areas which I really have begun to appreciate!. Specifically, reds and greens. In the past I have owned numerous bodies, going all the way back to the D70. I have used two bodies at all times during the last 8 years and there was never a difference as there appears to be with the D850. Has anyone else noticed these differences? I would suppose that it isn't just a difference between the 810 and 850 but probably the 850 and previous Nikon DSLR bodies. Edited July 22, 2018 by pcassity

Accurate color or appealing color? Accurate (i.e. profiled) color tends to be on the drab side, especially with high dynamic range cameras like the D850. There are usually options to jazz up JPEGs and previews, but RAW is RAW until you process it.

Well, there was a report (if I understood it correctly, and I may not have) that Nikon may have changed the filters between the D800 and D810, giving a different spectral response to "red", "green" and "blue". There's also a small amount of processing that happens on a raw file before it's saved, possibly compensating for the sensitivity of the sensor. And the D850 certainly has very different amplifier and digitising hardware to the D810. So, while the more extreme differences are more likely to be in the jpeg engine, raw is not necessarily raw. And then you're going through a slightly different path in the raw converter, too.

 

I've not noticed a massive difference, but I've not done much shooting side by side of my D850 and D810. I'll have more after a forthcoming trip, so I'll try to report back if I see something. I'm not using Lightroom, though (I usually use DxO), so if there's something different about its conversion, I wouldn't notice.

Using Imaging Resource's 'Camera Comparometer', I'm seeing a slight difference in the rendering of oranges and yellows. Actually more noticeable in pastel tones rather than saturated colours. Nothing that immediately smacks you in the eye though.

 

What's more interesting is that no more detail is visible in the D850 examples than in the D810's.

 

OTOH, the difference in colour rendering between the D850/D810 and Sony's A7R3 is immediately noticeable. As is the greater detail of the Sony.

 

Since Nikon's sensors are made by Sony, I suspect the difference in colour rendering is purely down to processing and not an inherent difference in the Bayer array filters. However, I do find the lack of apparent sharpness in the D850 examples quite surprising.

 

The D850 is now definitely off my shopping list. The A7R3 possibly on it, but we'll see what the near future brings with Nikon's MILC offering(s). Provided Nikon doesn't price themselves out of the game.

Joe: Are you expecting more per-pixel detail, or more detail in general? I wouldn't expect the former (in fact, slightly less, since the demand on the lens has gone up); the latter maybe. Honestly the detail difference between these was always going to be small (1.1x, linear) - the main benefit I see to the resolution increase is to give better 4K/8K video support.

 

Looking at the indoor scene, ISO 64 example, it really doesn't look very good at the pixel level to me on the D850. I don't believe I've seen this level of softness, although I run most things through DxO so some sharpening may be getting applied without my paying attention to it. I do notice they're at f/8 (with a 60mm macro, oddly defined in EXIF as f/3) - the Sony is with a 55mm f/1.7 (yes I know it's supposed to be 1.8, but EXIF says 434/256). f/8 is past the diffraction limit, so it's probably not the best choice for pixel peeping, and the 60mm isn't the sharpest lens out there (especially at longer range) - but it's far from bad, either.

 

Oh, wait. These are out of camera JPEGs? (At least, the ones I was looking at.) Nikon have always followed the tradition of relatively limited sharpening (in the apparent, hopefully these days mistaken, belief that people might want to do post-processing on the image). I've not checked what the settings were, but I'm sure the Sony has had some more in-camera sharpening applied. I'd not read anything into that unless you care about the in-camera JPEGs; Sony apparently have been improving their sharpening engine.

 

Other than my (long) ongoing list of requested improvements, I've had nothing to complain about with the D850. From the reports I've heard and a brief play, I probably would struggle more with an A7R3, but I'm not going to call it a bad camera in any way. But I've not been doing real estate work with the camera, and I always assume I'm going to tweak colours anyway!

With such hi-res sensors as the D810, D850 and A7R3, it's only 'fair' to compare sharpness and color rendering by using the same lens, or at least the same type, ie Sigma 135mm 1.8 on all bodies and then convert the RAWs with the same software settings.

 

If the lens is the limiting factor, then the sensor can only record what's there whatever it's supposed resolution.

That said... LensRentals did a piece on the effect of sensor stack thickness on resolution (since it's part of the optical path) and therefore the impact on adapted lenses. I'm not sure how much it would affect a longer lens, and I suspect that Sigma don't do much correction for it (since they offer a mount conversion service), but it doesn't feel quite as clear cut as it might be at first glance. My first response, though, was to see how much softer the 60mm f/2.8 is than the 85mm Sigma on tests (answer: at least at f/8, "a bit"). It's not that unfair to compare the lenses actually available for each system as part of the consideration of a camera, though. Having said that, there's a well-known 55mm lens that would definitely not be the limiting factor for sharpness. :-)

 

But now I've realised what I'm looking at, I absolutely blame different JPEG sharpening settings for the Imaging Resource comparator. Unless there's a way to configure it to look at optimised raws?

If you mean the Nikon 2.8 macro, the Sigma 85mm and 135mm remove it forcefully from the ball-park at all apertures. I've got 3 copies (don't ask!) and none of them come close.

 

I'm getting a D600 to IR convert soon and have been thinking about mono too. If the bayer has been removed I guess the RAW still has 'colour' info, even though it's not there?

Using Imaging Resource's 'Camera Comparometer', I'm seeing a slight difference in the rendering of oranges and yellows. Actually more noticeable in pastel tones rather than saturated colours. Nothing that immediately smacks you in the eye though.

 

The white balance neutrality is better preserved across lighting conditions in the new cameras (D5, D850) in my experience (haven't used D500 but it is reported to be similar in this respect). I found the D810 images tended to be more warm in some circumstances. Of course, colors are set in the final raw conversion and editing unless in-camera JPGs are used as the final result.

 

What's more interesting is that no more detail is visible in the D850 examples than in the D810's.

 

That isn't the purpose of those images. First, 45MP is only a very slightly higher resolution image than 36MP, and you probably need to use the optimum aperture of the very finest lenses to see that difference. This is also seen with 24MP DX cameras such as D7100 (or D7200). However, once appropriate technical precautions are taken (use tripod, EFCS, stop down the lens to optimum aperture), I am very satisfied with the per pixel results of the D850 and it is more or less what I expected. As the pixel density gets higher, it becomes harder and harder to see significant benefits. In prints of up to A2 (40 x 60cm), I've been perfectly happy with 24MP. But, the D850 does have much better AF than the D3X or D810 and this is especially evident when shooting full body portrait orientation shots with the face at the top, the cross-type sensors make a big difference to the consistency of focus in such shoots (such as in concert photography).

 

One think I've noticed is that I could get aliasing / moire in hair with the D810 using lenses such as 105/1.4 wide open, but I've never managed to see moire with the D850. The 45MP sensor seems to be high resolution enough to avoid it in most circumstances. This is a very good state of affairs IMO.

 

OTOH, the difference in colour rendering between the D850/D810 and Sony's A7R3 is immediately noticeable. As is the greater detail of the Sony.

 

They're jpgs from the camera, and subject to processing differences, I wouldn't make conclusions of detail retention of images shot to illustrate in-camera jpg color differences. Appropriate precautions may not have been taken to ensure correct focus etc. That said, of course the lens used also makes an impact. I couldn't find what lenses that site used.

 

Since Nikon's sensors are made by Sony

 

Sony fabricates the sensors according to Nikon designs. Nikon design the sensors (photodiode geometry, sensor filter and microlens array, wiring, data transfer, testing of optical artifacts with numerous different geometries to minimize unwanted flare effects with current and future NIkon lenses etc.) of their key cameras themselves. Sony technology may be involved obviously but the design is Nikon's and they seem particularly keen to optimize the optical components for the best results with their own lenses. Sony colours differ from Nikon's and it would be a strange coincidence if they used the same CFAs. As far as I know, Sony chose a thick filter stack for their mirrorless sensors to be able to sharpen up the CFA for better color separation, whereas Nikon are using a thinner stack for maximum compatibility with older lenses.

 

Pixels for Geeks: A peek inside Nikon’s super-secret sensor design lab

Edited July 23, 2018 by ilkka_nissila

The EXIF of the image I looked at said 60mm f/3.0 (whereas the Sony says 55mm f/1.7). Maybe Nikon were adjusting max aperture for focal distance; I'm not sure about Sony. I can't tell whether it's the AF-S or AF-D from the EXIF; I guess the former. I suppose it's possible that the 60mm lenses are sharper at macro range - apparently this is true of the 105mm, and the Sigma 150mm seems a little soft at infinity. At f/8 all of these lenses should be in the same ball park (at f/2.8 not so much), but that ball park is being diffraction softened. Still, the same should apply to the Sony, which is why I'm blaming the JPEG processing.

 

Of course the Sony should be able to get some extra sharpness in sensor-shift mode, but I assume we're not talking about that here. I envy it, but I so rarely bring a tripod and shoot somewhere with no wind that it would in reality not be that much use to me.

 

With my limited experience of exactly one IR-converted camera, I quite like having a slightly different spectral response so I can distinguish skies (orange in the custom WB provided to me) from foliage (pure white). I'm not sure I'd care enough to get a 590nm camera for full colour, though - it's too convenient to have moderate IR without a filter. Mirrorless seems like a more obvious solution for different filters. But yes, the default conversions all do some interpolation on the assumption that there's a Bayer grid there; I believe there are some raw converters available that just read the pixels directly and assume mono. Whether there's some in-camera difference in colour channels going on that mangles the raw file I couldn't say.

That said, of course the lens used also makes an impact. I couldn't find what lenses they used.

 

(See my comment on the EXIFs.)

 

Nikon fabricates the sensors according to Nikon designs. Nikon design the sensors (photodiode geometry, sensor filter and microlens array, wiring, data transfer, testing of optical artifacts with numerous different geometries to minimize unwanted flare effects with their lenses etc.) of their key cameras themselves.

 

Well, a bit. The D850 is allegedly full custom (despite Sony's attempts to get Nikon to use the A7RIII sensor). The D800 (and presumably D810) sensor is, I think, pretty much a standard Sony part to which Nikon had exclusive access with the D800, before it later appeared in the original A7R and the Pentax K-1. Presumably the Expeed was still Nikon's doing, though, so anything from that point in the pipeline on will be proprietary. I don't have internal knowledge to confirm, though. The D3, D4 and D5 series are definitely Nikon custom designs, though.

The D810 was also Nikon-designed with ISO 64. Increasing the number of photoelectrons that could be stored per photosite from 48818 to 78083 (sensorgen estimates for D800 and D810) while maintaining the pixel pitch was a feature which has not been available in other manufacturers who use Sony-fabricated 36MP sensors. I doubt the 36MP sensor was purely Sony-designed either, it has the high QE which first appeared in the D3s and Nikon's exclusive use of the 36MP sensor for two years (?) would suggest they provided something to the design which led to exclusive use for a time. Gapless microlenses, perhaps.

 

You seem to have copied my typo ("Nikon fabricates" which should have been "Sony ..." which I noticed only on my second read).:) In any case Nikon cameras have used sensors fabricated by Sony, Toshiba, Renesas, Aptina etc. it doesn't seem that they're particularly tied to Sony - even though Sony have been buying fabs left and right since, are they trying to be a monopoly?

Edited July 23, 2018 by ilkka_nissila

The EXIF of the image I looked at said 60mm f/3.0 (whereas the Sony says 55mm f/1.7).

I don't understand why you are seeing these weird EXIF information - for the Nikon D810 and D850, the lens is identified as "AF-S Micro Nikkor 60mm f/2.8G ED" and for the A7RIII it is the "Sony FE 55mm F1.8 ZA"' aperture values aren't displayed in fractions either.

 

I see very little value in comparing JPEGs from different cameras (with likely different parameter settings) and shot with different lenses - just look at the A7RIII resolution image to see all kinds of weird reflections indicating that not exactly great care was used in placing the lights and setting up the shot. On the rare occasions that I used that site to compare images, I am usually discouraged by obvious differences in exposure and white balance of the images that to me make any evaluation of color rendition or ISO performance differences moot.

I certainly don't have the technical knowledge that many of you have expressed. But I thank all for the input. Its been very educational. I thought I would do a quick comparison test to show the differences that I am seeing. Both photos were shot with the the same lens (24-70mm) at 24mm. Using the exact same settings on both cameras. (For those that are examining the EXIF, you will notice the time difference is substantial. They are actually just minutes apart. I need to reset the time on the 810. What I am noticing may just be a difference in white balance. The 810 does, in fact appear to be a bit warmer. I processed both in LR with all settings zeroed out, exported to jpeg.

By the way, which white balance setting do you use? The D850 supports natural light auto, auto-0, auto-1, and auto-2. auto-0 is more neutral than auto-1 where some of the light source color variations are allowed to stay in the image and auto-2 permits even more variation (i.e. under skylight the images appear bluish, in warm incandescent lights, orange, under auto-1 and 2, but less so in auto-0).

The D810 has two options for the AutoWB setting (nromal and keep warm lighting colors), the D850 has three (reduce warm, normal, keep warm). I compared the EXIF data of the bottom set - for the D810 the color temperature is 5100K and the tint is +3; for the D850 the values are 4550K and +13, respectively. Are both cameras set to the same option for AutoWB?

Both cameras are set to Auto1Normal

Belatedly...

 

Ilkka: I assumed the ISO64 was a function of the amplifiers, which I assumed was part of Expeed. I am, of course, making assumptions. :) I could quite believe that there was some cross-licensing agreement going on. It's true that the DX cameras, especially, seem to have jumped between manufacturers. I suspect Nikon would quite like to keep their options open.

 

Dieter: Exiftool does give me more information, including the lens ID, you're right. I was using the ImageMagick "identify" tool, which directly reports some (but not all) of the EXIF data - and several fields in EXIF are defined as rationals. Therefore if you see an aperture or a focal length, for example, it's actually stored in the EXIF file as a pair of values, giving a fraction. I imagine you're reading it using a tool that converts (and possibly rounds) the result for you. In this case, identify reports the focal length as "600/10", the MaxApertureValue as "30/10" and FNumber as 80/10. I'm a little hazier on how "30/10" got correctly turned into 2.8; there may be another field not reported by identify.

 

For your images: white balance and possibly exposure. Thom Hogan reported that the D850 seems to look like it's over-exposing its JPEGs when the raw files still have headroom. Hence one of my requests for Nikon is a separate exposure compensation (by post-processing) for JPEG and raw.

Well, then the nominally "same" setting does not produce the same outcome - the D850 is some 550K cooler for the bottom set of images. Would be interesting to see the outcome if both cameras are set to the same value (like 5000) and also if there are difference when settings like "daylight" or "shade" are used.

 

Exiftool

The EXIF data are readily available by clicking on the "EXIF data" button below each image on the comparometer website. Which incidentally uses ExifTool.

I assumed the ISO64 was a function of the amplifiers, which I assumed was part of Expeed.

 

The ISO 64 setting depends on the photosite being able to store more electrons before saturating. It's a physical difference in the photosite. The maximum SNR and the number of tones that can be distinguished is dictated by the number of photoelectrons that can be held.

The EXIF data are readily available by clicking on the "EXIF data" button below each image on the comparometer website. Which incidentally uses ExifTool.

 

So they are. Well, observation skills were never my strength. :-) Thank you for also failing to spot it on first effort, or that would have been really embarrassing!

 

I assumed the ISO64 was a function of the amplifiers, which I assumed was part of Expeed.

 

The ISO 64 setting depends on the photosite being able to store more electrons before saturating. It's a physical difference in the photosite. The maximum SNR and the number of tones that can be distinguished is dictated by the number of photoelectrons that can be held.

 

Fair enough (and it explains a difference in sensor between the D800 and D810).

Thank you for also failing to spot it on first effort, or that would have been really embarrassing!

Where did I say I didn't find it on the first attempt? Ilkka mentioned he couldn't find the lens info.

It appears to me that the 'bluegrass' image from the OP's D850 simply has the wrong colour temperature set, for whatever reason. A simple adjustment of CT towards warmer, or an adjustment of the blue curve, made it match the D810 image near perfectly.

 

I'd suggest checking that the B filter setting on the D850 hasn't been accidentally activated.

 

WRT the sharpness difference apparent on the Imaging Resource site. In the past Dave Etchells has taken great pains to ensure (where possible) the lens quality is equivalent across cameras. For example, I know at one time he used a Sigma Macro that was available in various mounts. So I'd be very surprised if a different lens or aperture was used between testing the D810 and D850. -What remains is that the D850 images just appear softer, which may be down to lower in-camera sharpening, agreed.

Well, paint me surprised! On checking the EXIF data, Dave has indeed abandoned the 70 mm Sigma macro on the D850 for a 60mm Nikkor. So if lens difference is responsible for the softer result, then the 60mm Nikkor macro is also crossed off my shopping list.

Where did I say I didn't find it on the first attempt? Ilkka mentioned he couldn't find the lens info.

 

Oops. Sorry, I tried searching for who said it, and clearly managed to conflate the two of you. In which case, I'm glad that Ilkka shared my embarrassment. :-)

 

I'm very strongly of the opinion that the JPEG difference is due to default sharpening. Nikon (and Canon) have traditionally applied very little sharpening to their out-of-camera images by default, as has been commented on by generations of people coming from compact cameras and expecting a highly-processed result. I'm not aware of the 60mm macro being a particularly bad lens (if not quite at the rarefied level of the Art lenses for sharpness), and it certainly shouldn't be at f/8, other than diffraction effects; no moderate prime lens has any business being soft at f/8. Since I'd expect the Sony to have the same issue, good out-of-camera sharpness from them suggests some different processing.