For the past few years, we have seen many manufacturers push for a lot more resolution. With sensor ISO performance pretty much leveling across the field on most cameras, manufacturers are now pushing to increase the resolution to differentiate new cameras from their predecessors and the competition. Except in some cases, as it turns out, increasing camera resolution is pretty much pointless if lenses do not have enough resolving power to be able to provide enough detail.
While testing lenses, I have been able to compare a number of DSLR lenses from Nikon and other third party manufacturers on both 36 MP and 45 MP sensors. As I expected, I found most Nikon F lenses unable to resolve over 36 MP, sometimes even over 24 MP (for many zoom lenses). That’s nothing new, considering that Nikon has previously issued different lists of recommended lenses for high-resolution cameras like D810 and D850. However, even many of the recommended lenses for the Nikon D850 turned out to be not good enough for 45 MP sensors, as I demonstrate below.
Table of Contents
Prime Lenses
Below is Nikon’s list of recommended prime lenses for the D850, along with my remarks on each lens’ performance on both D810 and D850 camera bodies:
- PC NIKKOR 19mm f/4E ED (exceeds 36 MP at f/5.6)
- AF-S NIKKOR 20mm f/1.8G ED (exceeds 36 MP at f/2.8)
AF-S NIKKOR 24mm f/1.4G ED(resolving power less than 36 MP)AF-S NIKKOR 24mm f/1.8G ED(resolving power less than 36 MP)PC-E NIKKOR 24mm f/3.5D ED(resolving power less than 36 MP)- AF-S NIKKOR 28mm f/1.4E ED (sharpest Nikon F tested, exceeds 36 MP at f/2)
AF-S NIKKOR 28mm f/1.8G(unable to resolve past 36 MP)AF-S NIKKOR 35mm f/1.4G(unable to resolve past 36 MP)- AF-S NIKKOR 35mm f/1.8G ED (exceeds 36 MP at f/4)
PC-E Micro NIKKOR 45mm f/2.8D ED(unable to resolve past 36 MP)- AF-S NIKKOR 58mm f/1.4G (barely over 36 MP at f/4)
- AF-S Micro NIKKOR 60mm f/2.8G ED (not lab tested)
AF-S NIKKOR 85mm f/1.4G(resolving power less than 36 MP)AF-S NIKKOR 85mm f/1.8G(resolving power less than 36 MP)- PC-E Micro NIKKOR 85mm f/2.8D (exceeds 36 MP at f/5.6)
- AF-S NIKKOR 105mm f/1.4E ED (exceeds 36 MP at f/2)
AF-S VR Micro-Nikkor 105mm f/2.8G IF-ED(unable to resolve past 36 MP)AI AF DC-Nikkor 105mm f/2D(resolving power less than 36 MP)- AI AF DC-Nikkor 135mm f/2D (not lab tested)
- AF-S NIKKOR 200mm f/2G ED VR II (not lab tested)
- AF-S NIKKOR 300mm f/2.8G ED VR II (exceeds 36 MP at f/4)
- AF-S NIKKOR 300mm f/4E PF ED VR (exceeds 36 MP at f/4)
- AF-S NIKKOR 400mm f/2.8E FL ED VR (exceeds 36 MP at f/4)
- AF-S NIKKOR 500mm f/4G ED VR (exceeds 36 MP at f/5.6)
- AF-S NIKKOR 500mm f/4E FL ED VR (exceeds 36 MP at f/4)
- AF-S NIKKOR 600mm f/4E FL ED VR (exceeds 36 MP at f/4)
- AF-S NIKKOR 800mm f/5.6E FL ED VR (exceeds 36 MP at f/5.6)
As you can see, even Nikon’s list of recommended lenses for the D850 isn’t ideal. Many of the lenses weren’t able to resolve more than 36 MP, with some lenses like the Nikon 24 MP f/1.4G ED, PC-E 24 MP f/3.5D and DC 105mm f/2D being quite poor in resolution, even when stopped down. Other lenses like the Nikon 58mm f/1.4G barely made it, with only one sample out of three demonstrating solid sharpness at f/4, but at the huge expense of the corners due to heavy field curvature. The exotic super telephotos obviously shine, but surprisingly, some of the older lenses like the Nikon 400mm f/2.8G VR, or the Nikon 600mm f/4G VR were already not good enough for 45 MP sensors.
Take a look at the below MTF charts for the Nikon 24mm f/1.8G, measured both on Nikon D810 and D850:
Note: For the above results, I provided “the best” potential results from 3 different samples when testing them on both D810 and D850 bodies. That’s why the corner results, as well as results at some apertures like f/2.8 are different.
As you can see, while the D850 has noticeably more resolution, the MTF numbers for both cameras hover around the 3300 range at f/4 and they never actually increase. This shows that the Nikon 24mm f/1.8G is simply incapable of resolving past 36 MP. So if you were to mount this lens on the D850, you would never be able to reach the resolution potential of the camera, and the camera would always out-resolve the lens.
Let’s see another example, this time from the Nikon 85mm f/1.8G lens. Here are the two test charts from Imatest, using the same copy of the lens:
Once again, we see a very similar situation – the Nikon 85mm f/1.8G does not improve drastically on the D850, which shows that the lens does not have enough resolving power to take advantage of the 45 MP sensor. In fact, compared to the 24mm f/1.8G, it shows inferior center performance at similar apertures, so it is not even able to take full advantage of the 36 MP sensor.
Now compare the above charts to those of the newer Nikon 28mm f/1.4E ED lens:
It is very clear that the Nikon 28mm f/1.4E is an absolute beast when it comes to resolving power. While its performance on the D810 was impressive, it was able to get much higher MTF numbers with the D850, showing that it can take advantage of a 45 MP sensor. This is the type of performance difference you want to see between 36 MP and 45 MP cameras.
Unfortunately, this is not the case for most Nikon F and even third party lenses. There are exceptions for third party lenses, of course. The exotic Zeiss Otus lenses were able to achieve exceptionally high resolution numbers, while some of the recent releases from Sigma, Tamron and Voigtlander also did quite well in tests. The Sigma Art-series primes were excellent overall when compared to other brands.
What about Nikon’s classics? I tested a few Ai-S lenses, such as the 28mm f/2.8 Ai-S, 35mm f/2.8 Ai-S, 50mm f/1.2 Ai-S, 50mm f/1.4 Ai-S and NOCT-Nikkor 58mm f/1.2 Ai-S, as well as some very old Nikkors. Most of them did not do well. The exception were the 50mm f/1.2 Ai-S, 50mm f/1.4 Ai-S and NOCT-Nikkor 58mm f/1.2 Ai-S lenses that resolved past 36 MP, but would probably struggle with anything over 45 MP. This means that some of the very finest classics should be able to provide enough resolving power to exceed 36 MP, maybe even 45 MP. However, most of them were designed for film cameras and look quite bad in the corners.
Zoom Lenses
Next, we will take a look at zoom lenses. Here is the list of recommended zoom lenses from Nikon for the D850:
- AF-S Fisheye NIKKOR 8-15mm f/3.5-4.5E ED (not lab tested)
- AF-S NIKKOR 14-24mm f/2.8G ED (exceeds 36 MP at f/4 @ 14mm)
AF-S NIKKOR 16-35mm f/4G ED VR(resolving power less than 36 MP)- AF-S NIKKOR 24-70mm f/2.8G ED (exceeds 36 MP at f/2.8 @ 24mm)
- AF-S NIKKOR 24-70mm f/2.8E ED VR (barely exceeds 36 MP at f/4 @ 24mm)
AF-S NIKKOR 24-120mm f/4G ED VR(resolving power less than 36 MP)AF-S VR Zoom-Nikkor 70-200mm f/2.8G IF-ED(resolving power less than 36 MP)AF-S NIKKOR 70-200mm f/2.8G ED VR II(unable to resolve past 36 MP)- AF-S NIKKOR 70-200mm f/2.8E FL ED VR (exceeds 36 MP at f/4 @ 135mm)
AF-S NIKKOR 70-200mm f/4G ED VR(resolving power less than 36 MP)AF-S NIKKOR 80-400 f/4.5-5.6G ED VR(resolving power less than 36 MP)AF-S NIKKOR 200-400mm f/4G ED VR II(resolving power less than 36 MP)- AF-S NIKKOR 200-500mm f/5.6E ED VR (barely exceeds 36 MP at f/5.6 @ 300mm)
As you can see, the list of zoom lenses is even shorter, and many of the recommended lenses did not make the cut. Not sure what Nikon was thinking while including the 16-35mm f/4G or the 24-120mm f/4G – neither of these lenses can resolve even 36 MP, let alone 45 MP.
Nikon included the older 200-400mm f/4G ED VR II in this list, which couldn’t resolve past 36 MP, whereas the newer 180-400mm f/4 VR did much better, clearly providing more detail than its predecessor, even wide open at 180mm. Nikon should update the above list (last updated in August of 2017) and include the newly announced lenses, including the AF-S Nikkor 500mm f/5.6E PF ED VR and the AF-S Nikkor 120-300mm f/2.8E FL ED SR VR. Both should provide sufficient resolving power over 36 MP.
Third party lenses also did poorly, with most lenses (even high-end Sigma Art series) failing to provide sufficient resolving power to take advantage of over 36 MP of camera resolution. There were some exceptions though – the Tamron SP 15-30mm f/2.8 and the Tokina 24-70mm f/2.8 AT-X PRO FX did quite well, although just barely over 36 MP. Sadly, most others that I tested resolved below 36 MP.
Compared to Nikon Z
Nikon has been quite busy in the past couple of years making lenses for the new Nikon Z mount system. As expected, each one of the Z mount lenses has been specifically designed to yield very high resolution. Based on my lab tests of the current Nikon Z lenses, every single one of them (including zoom lenses) can resolve up to 45 MP and beyond. This shows that the newer lens designs have been specifically optimized to provide exceptional amount of detail, something most Nikon F lenses cannot do. This has to do with modern vs older design – new lenses for both mounts are specifically made to take advantage of high resolution, whereas any lenses released prior to the Nikon D810 are going to suffer.
Conclusion
With less than 30 total Nikon lenses that are actually able to take advantage of the modern 45 MP sensor on the Nikon F mount, one would wonder why Nikon should even try to go beyond 45 MP sensors at this point. In fact, although a number of lenses in the above lists were able to resolve more than 36 MP of detail, some of them did that barely, only at very small apertures. Zoom lenses were mostly poor, only shining at very specific focal length and aperture combinations. If I were to exclude those lenses, the list would have been even smaller. Some of the modern third party lens options can surely fill the resolution gap, but that’s for those who are willing to do the research in order to find the very finest, or alternatively, shell out a lot of money to get the exotic Zeiss Otus lenses.
Sony has released its A7R IV with a 60 MP sensor, but the company has more glass that is optimized for such high resolution, because it is simply more recent. Whereas Nikon (and I am sure other companies like Canon and Pentax) have many older lens designs that are simply not good enough to take advantage of modern high-resolution cameras.
Personally, I see no reason for Nikon to continue the megapixel race for its DSLR cameras. Even 45 MP is overkill with the current stable of Nikkor lenses, as evidenced by this article. This is another reason why Nikon is pushing for its Z mirrorless cameras, since it is developing each lens from scratch, specifically aiming to create lenses that can take advantage of the latest high-resolution sensors. Whether we like it or not, Nikon will be forced to transition to the Z system in the future, if it wants to stay competitive with other camera manufacturers…
To me, going with ultra high-resolution sensors is not very meaningful when there is limited glass to support it. Instead of the next 60 MP sensor on the next Nikon DSLR, I would love to see a camera with a 36-45 MP sensor, but have the ability to stack and average images in-camera. A camera that can properly do focus stacking, pixel shift and other important advancements that are actually useful for us in the field. These to me are far more meaningful than another bump in resolution.
I haven’t done any testing but your results seem to be spot on for several of my F-mount lenses. I use the 24-120 and 16-35 and have noticed that resolution of these lenses on my D850 and Z7 are just lack luster. I just couldn’t explain why. Now I do and will keep the list of lenses above to avoid putting under resolving lenses on my higher resolving bodies. Testing some of the newest Tamron lenses like the 35-150 and 100-400 might be helpful.
What’s the diff putting on these lenses on higher mp d850 vs 810? Would it be the same ?
This article is not all that helpful, especially for people new to photography. So much of the content here is targeted to people learning, which is great. But this is a bold claim to make and it’s gonna confuse people. I think there’s more than one reason to buy a D850 but taken at face value this article could be perceived as an argument against it.
You know, you don’t *have* to comment.
The article is good, and accurate. And helpful to many. It matches what many of us long time Nikon shooters have seen, and what we’ve been warning about. It’s good to see this all summarized. Thom has posted similar observations, also correct.
Michael, yes, a number of lenses that did otherwise great on older bodies are struggling with high-resolution cameras like the D850 and Z7. The detail is simply not there, no matter what aperture. Newer lenses are not going to have this problem.
The problem is not the number of pixels, but the size of the photosites, in that order, cameras with a 30mp aps-c sensor would have to have absolutely fantastic lenses, just like mobile phones over 15mp would have to have fabulous lenses, and not quite, the lenses don’t have to deal with a lot of pixels, but with their size, Nikon wants to push customers to the z system and you’re supporting
I can add that I have a D810 and a D850, all my lenses, even the oldest ones, get more details on the D850
Hi Victor, it is a releif to see someone say it. The thing is, when we are shooting on this level of camera body, are we really shooting Jpegs? I strongly suspect most are shooting RAW. Assuming I am mostly right, then, the post processing, and quality of software, and experience of the photographer using that software has quite a lot to do with the final quality of the image. Then there is the final image size of the print to consider. I suspect up to 20 x 16″ prints, it would be well nigh impossible to ‘see’ this issue, especially with prudent post processing. As always it comes down to what the image has been created to do. If you are proposing going bigger than 30 x 20″ prints, then maybe a better lens could have been employed, but other than that! Don’t forget that Nasim has taken a lot of time to come up with this information, and, in the right context, maybe it would be of some importance to you. (ie, very large prints, or small areas cropped out of main images and printed big). In those cases, then you would be wise to heed his warnings – that said though, I have the 14-24mm f2.8 and I have the 24-120mm f4.0 and quite honestly, I often find the images from the ultrawide don’t come up to the standard of the 24-120mm. It could be technique on my part, but I do find large prints off the 24-120mm are pretty exceptional while the ultrawide sometimes dissapoints on edge of frame. I have come to the conclusion that my own personal copy of the 24-120mm was a lucky purchase hit, and it was constructed with parts which are VERY close to (or even right on) the designers specification, and my copy is just delivering images which are a tad better than those from other samples of the lens, owned by others. I am the sort to change lenses quite frequently, but I can’t see me saying goodbye to my 24-120mm unless other factors arise – new camera body perhaps. When you think of it, no two examples of the same lens are ever going to be identical down to the smallest atom! Even Nikon must have some batch variation going on, so do as I have, if you find your own lens and body is delivering, don’t go near the camera shop! As they say, Dear Lord, lead me not into temptation!
furthermore that I said, of the list of lenses that can solve more details than 36mp, only one is among the 3 best lenses from Nikon according to the DXO website, and includes lenses that are ranked in 10th place or worse, includes for example the 24- 70mm2.8 and excludes the 85mm 1.8 lens, perfect nonsense, at 24-70mm 2.8, it really gets a lot of details, but the level of sharpness is still far from the 85mm, any of them, even the 85mm 1.8D
Hello Nasim, great job, thank you! One question: if money is not an issue, which 5 lenses (from all manufacturers) do you recommend to use with the D850, which exploit the full potential of the 45 MP sensor Nikon D850?
Peter,
1) Nikon 28mm f/1.4E
2) Nikon 20mm f/1.8
3) Nikon 105mm f/1.4E
4) Any of the Nikon super telephotos
5) Sigma Art primes (not zooms, maybe with the exception of the Sigma 14-24mm f/2.8 Art)
12-24 f/4 Art is very good with D850 under 20mm, the long end not as sharp as wide end but IMO better than Nikon 14-24 2.8
Thank you!
i have a D810 and many of the lenses on your list that meet or exceed 36mp performance. I am only confused on one choice. I have the 24mm f1.8G and purchased this because it out performs the 20mm f1.8G based on many review sites including yours. Center sharpness is about the same for either, but the 24 gives better mid frame and corners. So why is the 20 recommended over the 24 on your list above?
Steve
Steve, I had the same thought and I actually owned both at one point, but sold the 20mm and kept the 24mm, because I found I used it more often. My main lens is the 400mm f2.8 FL and second is the 70-200mm, I only* also own the 24mm f1.8 and 58mm f1.4G’s. I don’t really like zooms, but the 70-200mm is so good, I don’t mind it. I had a D850, but it was simply too much camera for me and I sold it and have never looked back. I even for awhile thought I missed it a little and decided to add a cheap used D800E and ended up selling that too lol. My favorite camera of all time and my workhorse is the D4s, actually two D4s bodies and I have D7500 that also use from time to time if I need more reach or even more detail. Even though I own the TC-14E III and it works flawlessly on my 400mm lens, I’d prefer to just switch to the D7500, before I add the TC. If I need a lot of reach I use both and I’m at 840mm f4, which actually behaves more like 840mm f6.3 though, which is still amazing bokeh and detail. Back to the 20mm and 24mm thing though, I think what Nasim meant is the 20mm f1.8g is slightly sharper in the center and he’s just taking the highest number recorded in the testing, so his copies of the 20mm beat the 24mm in the center. I honestly have no stake in this stupid MP war or game anymore I’m glad I found that out, kind of the hard way, but I have. I no longer care about the amount of megapixels, but rather the quality and the D4s probably has one if the greatest sensors ever made. There is just something magical about the images it produces, almost like buttery smooth, yet sharp where it matters and with lenses like my 400mm 2.8 VR FL…files are ridiculously gorgeous. I do make prints once in a while up to 13×19 and sometimes a little bigger on canvas, but not often. So 16mp is a sweet spot for me as my most of my images are viewed online or in news paper print or magazine prints. I’m a photojournalist and therefore things like low light performance, build quality, FPS, etc. are more important to me than anything else and the D4s is more than adequate. I have access to the D5 through my work and I also bought one when it first was announced, but I prefer the D4s over it. The D4s has almost two stops better dynamic range and only a half stop worse high ISO to the D5 and I discovered no Nikon camera, the D5 or D4s does more than 10 FPS with G-type lenses so other than better AF, the D5 to me is not much of an upgrade. I think I will be happy with my D4s bodies for a few years more and someday maybe I’ll buy a D6 or two used D5’s, who knows, but I love my lenses. I tried so many lenses over the years, it’s not even funny and the four I have now, I found to be very best out of the bunch. Again I own the 24mm f1.8G, 58mm f1.4G, 70-200mm FL and 400mm f2.8 FL. To me lenses are more important than camera’s and I’d way rather have a $3000 lens on a $500 camera, than the other way around.
Are you saying that these lenses perform worse with a high-resolution sensor or that they perform the same but you can see no improvement on a high resolution sensor?
Thank you for this very interesting article, nobody writes about these facts as clearly as you
This article perfectly exemplifies why I usually come here, read the headline of the most recent article and leave. There’s rarely a point and almost never any information that would benefit MY photography life. :-(
Pat, I am sorry that you found our articles that useless. If there is anything we can do to improve, we are always open for feedback.
Hi Nasim. Don’t fret over this and do not feed the trolls. You (and the whole PL crew) present great articles with tons of useful information. Well done and thanks very much for all you offer the photographic community. Cheers
Totally agree ! Keep up a great job Nasim , Spencer and all the rest of the crew.
// T.
I do not see these results in my 800 series cameras. I nearly exclusively shoot primes, and i DigiTech nearly every camera system out there. And besides the Sony’s the nikons show more detail than Canon or fuji. Far too much detail. To the point where i can see the peach fuzz on a young girls face that i cannot see with my eye. So… I reject this assessment. If anything the nikons are too sharp with too much resolution. I also shoot them in 14bit uncompressed raw. There is tresting also that says Canon sensor cannot resolve f1.2 so… There you go with all that.
Perhaps you could suggest an article that you WOULD read?
How about DX ? Pixel density on D500 is on par with D850…. same conclusion? I am sure almost all older glass is inferior to the sensor, even the old exotics as 500/4 and 600/4… Modern superzooms are surpassing these great lenses on resolving power….
I don’t know what “superzoom” you’re referring to, but I highly disagree with that statement. I have owned every Nikon super-tele other than the 800mm and super-zooms will never come close. Sure maybe a 200-500mm lens at it’s best aperture and focal length might match center sharpness, but not overall sharpness, no. Also there is so much more to image quality than center sharpness and a 400mm f2.8 VR is always going to smoke a 200-500mm lens in IQ. Sure each new Nikon super-tele prime is a little sharper than the last generation, but you have to remember both Nikon and Canon have had super-tele primes down for a long, long time. In my opinion even the older 300mm, 400mm, 500mm and 600mm primes from as far back as the mid to late 1990’s will outperform modern “super-zooms!” The only lenses I can think of that would describe what you’re saying is the 180-400mm f4 FL Nikon or the Canon 200-400mm f4 IS and neither of those is a super-zoom. Also there is severe sample variation on most super-zoom lenses and while some are pretty sharp, other’s are only acceptably sharp, where as ALL super-tele lenses are made to extremely high specifications and tolerances. I don’t see any comparison in overall image quality between even an older Nikon or Canon 300mm f2.8 or 400mm f2.8 lens and a modern super-zoom. Center sharpness at it’s best aperture and focal length on a zoom maybe close in rare cases to a super-tele prime, but again that’s only in the center and center sharpness is only one of countless things that matter to image quality.
I respectfully disagree, this article is extremely useful. Some people will always find something to complain about no matter what…
Nasim this article is very useful and important for uses. Just one question. The older 24-70G 2.8 is better than the 24-70 Vr? Can you confirm that ? It’s much more expensive and later too…
The older one said to have slightly better resolution in the center while having worse corners. This article only takes center into account, that’s why the newer one fares worse in this test.
Thank you, Ilya!
I believe that the older G lens is a little sharper in the centre, while the VR lens has more even resolution across the frame.
Dejun, I did not have a chance to expand on that, hoping that our readers would read the review of the 24-70mm f/2.8E VR instead. The older 24-70mm f/2.8G is very sharp in the center, but at the expense of mid-frame and the corners, due to heavy field curvature. The newer 24-70mm f/2.8E VR fixes the mid-frame and corner performance issue, at some expense of the center. Since I provided data mostly for the center sharpness, it makes it look like the 24-70mm f/2.8E VR is inferior. For me, it is not – it is a more balanced lens of the two, with much more even performance across the range.
Thanks 😊 Nasim
Thank you Nasim for the interesting read. What I’m wondering is how you measure “resolving power”? If I remember correctly, the softer a lens gets, the lower the contrast between neighboring lines (and every test sets a cutoff point on the contrast, maybe like 50%). But this could be counteracted with sharpening in post, meaning one could very well recover the required “resolving power” in post. Of course, if contrast is zero, there is nothing to recover anymore..
So my question is: is this article just relevant for jpeg shooters that want maximum sharpness without any post processing, or an I mistaken? Thanks!
Good question, I second this – please explain how do you measure „resolving power“! The „proof“ in the essay seems to be not direct and thus – as I (not) understand it – interesting showing, but somewhat incomplete? Thank you in advance!
Robert, I don’t measure resolving power – I cannot quantify differences to a single digit, like some other sites do. I simply see that some lenses cannot provide more detail on cameras with more resolution.
It is quite simple. If you take a picture on a D810, then take a picture with the same lens on the D850 and look at things like hair or even finer particles at 100% and see no additional detail, then the lens cannot provide sufficient resolving power on a higher resolution sensor…
Thank you Nasim, will try it. I still am at a loss: shall I believe it? I am no technician, geek, nerd or whatever, so probably I just have to try.. 😀 Somewhere on the net I read once, that resolving power is only one part of sharpness and overall image quality, but did not quite understand. OK, lets test… At least, I have accepted already, that the earth is a ball… 😉😂. Luckily, I have already acquired Z7 and S primes, so my stitches are in no danger. Those big mountains are interesting with as much detail as possible on a big print. Cheers and thanks once again!
PS
Those tables are very practical: I often cannot change lenses in bad weather, so now with your tables I can pair the right lens with the right body before the hike and be done. Very useful for me, Nasim!
This is how Imatest (which Nasim uses to measure sharpness) defines it – www.imatest.com/docs/sharpness/
That being said, Nasim described in this article an easy way to tell if the lens could resolve past a lower resolution camera when comparing 2 specific cameras (a lower and a higher resolution body). If the measured resolution doesn’t increase from the low to high body then the lens isn’t able to resolve past the lower res body. Ideally you would then measure an even lower res body to see if it was taking advantage of the original low res body.
Thank you Preston, very nicely put.
Thanks Preston, I think even a guy like me can understand your explanation!
Stefan, measuring “resolving power” is not an easy task, not something I would recommend to try for any of our readers. There are too many variables in play. I shared the data from testing different lenses on D810 and D850 bodies, and these are the conclusions I was able to draw from testing many samples of lenses over the past few years. In regards to how Imatest does it, I am not going to go into details, but the link that Preston provided, as well as many other documents from Imatest can help clarify how the software actually measures MTF. Sharpening in post can help, but if you do not have the data to start with, and it just looks like blur, what is there to sharpen? If you shoot with a lens that is good enough on the D810, but not good enough on the D850, what is the use of the additional resolution? You end up with larger files, but the amouint of detail remains the same. Sure, it can help with things like moire, but is that a good enough reason to continuously move up in resolution? I would rather see manufacturers provide other more useful features, than continuously push for higher resolution.
And no, the article is not relevant for JPEG shooters. It is relevant for everyone.
Hello Nasim, great job, thank you! One question: if money doesn’t matter, which 5 lenses (from all manufacturers) do you recommend to use with the D850, which exploit the full potential and power of the 45 MP Nikon D850 sensor?
There is a kind of photographic snobbishness in the article which I encounter from time to time. Fortunately there is less of it in the digital era, but it is annoying when it raises its head. This article, and a few others I have read, implies that some older lenses are not suitable for cameras like the D810 or 850. Yet these lenses, which have performed admirably on many older cameras from the digital era, or dare I say it, the film era will perform just as well on modern high resolution cameras. Probably there will be some improvements owing to additional features on the newer cameras. You do not have to discard your old 24-120zoom from the nineties. It will still work and produce the same beautiful results in 2020. If you have loads of cash you can upgrade to an over priced over heavy Sigma Art thing, but I would suggest It would be better to be realistic in you expectations.
I think the point (as delineated in the title) was that chasing Mpix in the camera body is likely not adding value to the end result of photographs taken with lower resolving lenses. That’s not to say (and Nasim didn’t say) you cannot use lower resolving lenses on a high Mpix body and get a picture you’re very happy with. The higher Mpix body may have some other feature that adds value (focus stacking, Improved dynamic range, better auto focus, …) which may make the picture better than would have been captured with an older body. But there is value to the purchasing public in measuring lenses “recommended” by the manufacturer explicitly to exploit the higher pixel count of specific camera bodies that in fact do not achieve that goal. If I feel I need (or just want) more resolution – this article also let’s me know I’m going to need to invest in lenses as well as a body to realize that one small (pun intended) aspect of what “can” make a great final image.
As if lens resolving power has anything to do with the making of a good photograph.
Fully agree
Making good photos doesn’t have anything to do with high resolution cameras, either. But the topic of Nasim’s article was different.
You can’t just throw around wild unsubstantiated claims like that without offering up some evidence to back it up. .
“Wild”, “unsubstantiated”
?
It’s technical, but maybe not useful learning the technical jargon for regular photographer, but if interested, here is an article…
luminous-landscape.com/do-se…ve-lenses/
Tom, this article is not about photography, or art, or good photographs. It is about camera manufacturers unnecessarily pushing for more resolution, when most of their lenses cannot take advantage of them. Ever heard of camera phones that have insane resolution like 100 MP? Yes, it is a marketing gimmick for the most part. A smartphone with a 12 MP 1/2.3″ sensor will likely provide the same amount of detail as a 100 MP 1/2.3″ sensor. You end up with a lot of useless pixels on 100 MP, because the lens in front of the smartphone is simply not good enough to resolve that much detail. This is what the above article is about.
Fair enough, Nasim. :)
You know, you don’t *have* to comment.
Since I am using a Nikon Z7 with a Z 24-70 2.8 I see a lot more moire effect then I had with the D750. Is this also has to do with the sensor? How can I avoid or reduce it?
Don’t photograph things with fine, repetitive detail in them…
You’re welcome.
Jean, please see my moire article for more information. This has a lot to do with the fact that the Z7 does not have a low-pass filter that blurs fine detail, like your D750 does.
Thanks Nasim! A very clear article!
One more question left, does a Z6 produce less moire then a Z7 or does it not matter at all?
Jean – be sure you really have moire before you worry too much. Moire often appears when you are downsizing a sharp, high resolution image to a smaller size. That might be on your computer screen, for social media, etc. But you might not have moire in the original image when you view it at 100%. Often just viewing at a different size is enough. In a print, you could use a different resizing algorithm.
Hi Nasim,
thank you for the article which should have been written ages ago – f.ex. when DX-format was introduced by Nikon. Modern DXs have 20-24MP chips and since their introduction I have wondered, do DX-lenses meet that requirement. My own experience with only a few of those lenses is no, they do not. Perhaps some day you write about that, too. And what about MFT-format lenses?
My photography concentrates on astrophotography with ordinary camera optics and what is said below has to be seen in that context. Cameras ranged from Nikon’s D70s to my current “stagnation” with 24MP D750’s. For the lenses I do not have MTF-numbers but instead experience. For 25 years when buying a new/second hand lens I have had it tried and wisely so – coma, ca and vignetting are more important than what usually is emphasized in lens tests.When I had my transition from film to digital, then for a couple of years I had 40+ Nikkors and 10+ Sigmas ranging from 8mm to 600mm. Here are some topmost thoughts.
Almost all D-series lenses were poor to my use (mainly due to huge coma). One exeption was Nikon’s 28/1.4D Aspherical, which – so I was told – had two different constructions/batches and of those only one was a keeper and still is in semi-active use. However, even this better one does not quite match with Canon’s legendary old aspherical 24/1.4. Another “old lens” exeption (with reservations) was Nikon’s 14-24/2.8, which works nicely but had a strange “vignetting”. It was not only in corners but illumination was more like a expanding circular wave, which was practically impossible to correct in post. (Yes, three rounds of vignet corrections might result in a more evenly lit background sky.) One exception to those old days was Sigma’s 28/1.8D Aspherical, which to my mind was one of the best lenses ever quality and price consídered (huge vignetting though). When Nikon 14-24/2.8 was compared against Samyang’s 14/2.8, Nikon was nearly a stop faster in astrophotography. Still, Samyang lenses (14mm, 35mm and 135mm) are OK in resolution, have acceptably low coma and keep acceptable sharpness quite nicely over whole sensor, and surely can be recommended for tighter budget. Actually Samyang’s lenses compare to Nikon’s legendary Noct Nikkor 58/1.2 which neither was super sharp (not even in center) but maintained good sharpness nicely over whole sensor and had only very minor coma.
With modern lenses my favourites are mostly Sigma’s primes: 14/1.8, 20/1.4, 35/1.4, 50/1.4, 85/1.4 and 135/1.8. Good in night sky astrowork. And yes, skip 24/1.4 and can’t afford 200/2 or longer. On Nikon side I like and still use a lot DX-lens 10.5/2.8 fisheye – hugely better than old FX-lens 16/2.8D still in use. Also very satisfied with 8-15mm fisheye zoom, which allows astrophotography with open aperture. Number of zooms is used in normal daylight shooting, but sure enough some should be updated. Also would enjoy checking some expensive Nikkor’s, but as there is no way to purchase them, better to forget that.
A couple more final experiences. 20+ years old Sigma’s 300/4 Apochromatic has less CA, shapness is almost equal to but contrast is lower than what Nikon’s AF 300/2.8 ED produces. And finally: for a daylight walk around camera Lumix FZ1000 is overall excellent and its lens is amazingly sharp.
Oh yes! 20+ years ago one Sigma’s zoom broke into parts and I found it disturbing. So I prepared a letter to old Mr Yamaki telling my concerns. Surprise – he answered and provided a new copy. Obviously also younger Mr Yamaki has inherited his great attitude!
Yes, a word on DX lenses would be greatly appreciated — both Nikkor and third-party.
Based on this article, very few lenses can resolve past the 20.9 MP D500 and D7500 (the lowest res DX bodies). The pixel size on that sensor is equivalent to a 48.8 MP FX body. It seems Nikon was aware of this since they intentionally developed this lower res sensor rather than use the existing 24.2 MP DX sensor from their other bodies. This let them achieve higher practical fps due to smaller file sizes without apparently giving up any resolution for the vast majority of lenses and/or shooting situations. And theoretically with bigger pixels they could get lower read noise (better high ISO performance, etc). For what its worth, the 24.2 MP DX bodies are equiv to a 56.5 MP FX body in pixel size.
Thank you for the feedback Pekka and for sharing your experience with lenses.
Regarding DX lenses, I think it is clear that most FX lenses are not up to task with such high pixel density, which is why we often see less detail when using those same lenses on high-resolution DX cameras. However, this does not mean that manufacturers cannot design proper lenses for APS-C – they can make lenses that resolve more detail. Look at many MFT and Fujifilm X-series lenses. They are good examples of solid lens design that is good enough, especially once corrected in software. Unfortunately, Nikon always neglected its DX lenses. We have a bunch of zooms, superzooms, and very few high-performance primes. That’s sad, especially when one sees the potential with other manufacturers (again, Fujifilm and MFT). Thankfully, there are third party manufacturers that have realized this gap and have been making excellent lenses…
A very nice and useful text, thank you very much!
This means they have no lenses at all for their apsc cameras 24 MP and above
No, but for the new Z50!
Smart move stay with 20 MP to get the owners of old F-lenses to stay with the brand!
Muhammad, please see my response above to Pekka…
Hi Nasim,
like your argumentation regarding Nikon. Convincing. But then your statement: “Sony has released its A7R IV with a 60 MP sensor, but the company has enough glass that is optimized for such high resolution”. Where´s the proof of that?
Best,
Michael
I will have to second this. Meanwhile there are a lot of Nikon F that I’m completly aware that will just not make it to the past 45Mp race, I see a lot less options if you are going Sony and you ask for 60MP. It would be nice if you do this same comparison with the Z-mount lens vs the equivalent FE-mount lens and you use the Z7 and the A7RM3 and A7Rm4, to check how many lens manage to get more detail going from A7Rm3 to A7Rm4 and how the equivalent Z-mount get positioned.
My point is that the initial “batch” of Sony lens were are all seriously lacking of resolve power, same thing that happen with the old F-mount designs and even the best of best of that Sony has managed to produce just sit in the middle between the F-mount old design and Z-mount. I would also bet that the D850+70 200 2.8 FL will resolve more detail than the Sony A7Rm4+70 200 2.8 GM after looking at Roger Cicala’s detailed analisys of all the 70 200 in the market.
At the end of the day you do have a “holy trinity” that you can use with your D850 and make use of its resolving potential and a lot of primes and zooms that cover past the 200mm limit of the trinity, so I will have to disagree that Nikon F-mount should stay below 36Mp.
Michael, I have been testing a number of lenses on the Sony A7R IV (as well as older A7R-series cameras), and this is my assessment of some of their lenses. Sony also has a bunch of poor lenses (especially some of the zooms, like the 24-70mm f/4), but many of the primes in the recent years have been excellent. Keep in mind that Sony is a relatively new player in camera market (talking about A7-series, not Minolta), so it has pushed a lot of lenses at the time when camera resolution was going up fast (the first A7R already had a 36 MP sensor). So unlike older players like Nikon and Canon, many of their lenses are already optimized for high-resolution modern sensors.
You had reached the conclusion that Nikon shouldn’t make more than 36Mp based on that older designs aren’t good enough, now tell me how the f-mount lens that Nikon has released in the last 5 years has performed so far? then tell me how many lens does have Nikon including their own plus 3rd party that does well on 36Mp+ cameras? The reality is that you have plenty of f-mount options and each single of the Nikon f-mount full frame released during 2015-2020 was a winner.
I believe this article is flawled because you are clearly just looking at half empty glass of water and assuming is completly empty. If you say “Why Nikon Should Not Make DSLRs Over 36 MP” and you pick old lens that you know are going to fail your are twisting the data to support your assumption instead of providing unbiased data. According to your arguments any camera maker that has old lens designs that has low resolving power should stop trying to get better instead of get better at both lens and sensors.
All your assumptions are completly false unless you can prove that Nikon is uncapable of creating f-mount lens that can resolve +36MP and you discarded yourself that theory at the moment you admited that there are several lens that are already resolving more than that.
I will finnish with this:
Sigma 14-24 + Nikon 24 70 VR + Nikon 70 200 FL
I believe those 3 lens are enough to prove that your asumption is wrong and Nikon can safely create DSLR with higher pixel count and be safe counting with the lens it already has, without even considering the ones that can come in the future.
The A7RIV is about more than increased resolution. It focusses almost as well as an A9; improves ergonomics and weather sealing and allows substantial cropping (often reducing the megapixel count and more often than not, potential issues at the extreme edges of the frame).
And yes, GM lenses are easily resolved by the A7RIV.
It has been proven through testing that many of the older GM lenses do not fare well with the 7r3 and many more fail to resolve with the 7r4. I wish I had the link to post here now, bit will find it and post it later on for your reading pleasure.
Here is the link I mentioned above for the testing done on the A7r4 :
sonyalpha.blog/2019/…ony-a7riv/
This article serves as a good lesson to all the pixel hunters who invested in substantially more expensive high MP cameras with no real improvement in image quality. I am quite happy with my 24MP sensor D750, and only when broken beyond repair, will I upgrade to the very similar D780.
Nasims research seems to indicate if you pick the right lenses you get a resolution advantage, but I’m right there with you as a d750 owner. Nikon rebuilt most of mine for free when the flash capacitor blew and took out the camera a year ago. It was well out of warranty. Talk about amazing customer service!
When my d750 dies I’ll see if I go mirrorless or d780. The d750 captures such amazing images I don’t see a reason to upgrade. It’s be nice if we could get a list like this that differentiates between 24mp and 36mp.
Well the Nikkor AI / Auto S Micro Niklkor 55mm f3,5 performs very well on a d800 and even on a Z7…
Hi Nasim,
It´s not only that there aren´t many decent lenses for bodies with so many megapixels, also many users of Nikon APSC (I´ve D7200), which we dont want or need to change cameras, we dont have decent lenses of the brand for our equipment. A few years ago we still had hope, but with the arrival of the Nikon Z we have to rule out that possibility.
Regards
Javier, if Nikon does the same thing with the Nikon Z APS-C cameras, it will repeat its history with DX / FX cameras and only deliver crappy zooms and superzooms. Nikon should really push to take advantage of APS-C cameras, by providing high-quality lenses for them. Sadly, that train is already gone for DX…
Dear Nasim,
I think there is typo in the third graph: it says the D810 is a 45MP camera (not a big deal).
I have the impression this article is flawed for a few reasons:
-There is not even a hint at the improvement in color resolution (which is not limited as much as luminosity, having 1/4 (?) the resolution)
-I don’t know the details of Imatest, as I’m more familiar with MTF curves, but, when considering the the 24G test graphs, the f/16 case jumps out: the lens scores higher on the D810 than the D850, which is absurd. Unless Imatests tests at the pixel level (and not the overall image) or something happened while getting the data. The 28E graphs don’t have that issue.
-For the older lens, while the center sharpness indeed doesn’t improve by a lot, some midframe/corner get a mile better :
24G corners at f/2.8 improve by 40% (1419 vs 1987). At f/4 by 23%, at f/5.6 by 17%.
-In general, mounting the same lens on higher resolution camera does produce sharper images: the system MTF is the product of Lens MTF and Camera MTF. at the end of the Lens Rentals blog article named More Ultra High-Resolution MTF Experiments (they test newer designs, so it doesn’t apply to this article as much), the author clearly writes in layman terms. Stealing a single sentence: Let say we were using a crappy kit zoom lens with an MTF of 0.3. With the old camera (whose MTF is 0.7), we get 0.3 X 0.7 =.21. Let’s spend a fortune on the newer, better camera, and we get 0.3 X 0.9 = 0.27. So our overall system MTF only went up a bit (0.06) because the lens really sucked.
This the theory, the reality might be a little different, but still.
-There’s also the fact that sensor manufacturers are moving towards higher resolution sensors: it might simply be more ecomical for Nikon to follow suit, especially if the sensors in the Z and F cameras end up being the same.
Besides, other than the increased file size, I don’t see why larger images should be worse. More detail allows for more effective downsampling or noise reduction and reduces moiré, among other things.
Regards
Giovanni, thank you for the feedback. I published the charts late last night and they did have some issues, which I have corrected.
Regarding the Imatest results, I added a note on the 24mm f/1.8G – I basically used the best results from 3 different samples on the D810 and D850, so the results look a bit different at particular apertures.
The f/11 and f/16 results were basically an error on my part. Since the 24mm f/1.8G has focus shift, I had to refocus at f/4 on the D810. I did the same on the D850, but forgot to include the data from the refocused shots, which resulted in inferior numbers. I went back, re-ran the tests and updated the chart – please refresh to see the correct version.
I also realized that one example is probably not good enough, especially one that used the best of three different samples, so I went back and added another test using an excellent sample of the Nikkor 85mm f/1.8G. The data on both D810 and D850 was from the same lens.
In regards to Imatest, shooting a test target is no easy task and it is never exact science. You could measure the same lens, refocus and retest it, and come up with completely different results. The way I measure lenses is very complicated and time consuming, but here is how I do it:
1) Try to get at least 2, preferably three copies of the same lens.
2) Use identical setup, using the same continuous lights, as recommended by Imatest.
3) Use a high-resolution test target capable of resolving beyond 50 MP (I have three test charts, some are film and backlit).
4) Before each major test, measure light output at different chart distances (center, mid-frame, corner, up and down) to make sure that it is consistent.
5) Take a quick shot in AF while being perfectly parallel to the test target. Run through Imatest and make sure that the lens is not decentered. If the lens is decentered, see if I can move the camera left or right to compensate. Sometimes it works, sometimes I have to move on to the next sample.
6) Move macro rail to mid point (0). Focus on the chart using AF at widest aperture. Tweak focus in live view magnified at 100%, if needed. Go back -10 steps on the macro rail, then take pictures at each step. Once I get to zero, go forward +10 steps, one step at a time.
7) Copy the files to my PC with Imatest, run all images through it, assess the graphs.
8) While running the tests, I can see which point on the macro rail provided the best sharpness results. Go back to that point, run through the whole sequence, this time stopping down from maximum aperture to f/16.
9) Repeat the first process twice more, third time only using manual focus. Sometimes I have to repeat this more than 3 times, especially when something is potentially wrong with the lens sample.
10) Pick the best data from all the tests.
11) Assess the lens for focus shift at different apertures. If there is focus shift, repeat the above tests stopped down by 1 stop (f/1.4, f/2.0, f/2.8, f/4.0, f/5.6).
12) Combine the data from different tests and apertures, then copy the best results for each apeture to a separate folder.
13) Run final Imatest with the best of the best. That’s what goes to the charts on the lens tests.
14) If it is a zoom lens, repeat the whole process for each focal length. Zoom lenses can take hours, sometimes days to properly test.
There are many other factors in play, which I am not going to write about here, but hopefully you get the idea…
Testing lenses is a very complicated process, something that takes many hours and the results are not always consistent. I have exchanged many emails with Imatest engineers, as well as Roger Cicala on this. Roger moved on from Imatest to an Optical bench system, which has its advantages, but makes it a real pita to test some lenses. He does not test lenses at stopped down apertures, so there are pros and cons to both methods.
As I always tell our readers, it is a good idea to look at multiple resources that tests lenses and come up with conclusions. I am a human, and I can make errors, or have bad samples, just like everyone else out there :)
Regarding this: “besides, other than the increased file size, I don’t see why larger images should be worse. More detail allows for more effective downsampling or noise reduction and reduces moiré, among other things.”
Downsampling won’t add any additional details to an image, if the lens is not able to resolve that detail in the first place. Noise reduction is also not an advantage, if output size is the same. Moire is the only advantage I can think of. Think about all the disadvantages – larger files, more processing power required to work with those files, backup size, etc. I would much rather see cameras with amazing features, such as ability to average multiple images, or perform pixel shift, than to see a 60 MP+ camera body…
Hello Nasim, I cannot agree.
If i use a Nikon J5 with some lenses of mine it is clear that the center can do more than 100 MP.
Off course you need good lenses to do that. But you can just buy them for reasonable prices. My best lens is the Sigma 40mm f1.4 ART. Already at f2 it shines in this respect.
Pieter, please re-read the article. Of course there are amazing lenses out there!
I understood the article; the J5 sensor is similar to a central area of a 150MP FF sensor; More lenses with a cross, for instance the Nikon 24mm 1.8 G will do at least 100Mp in the centre Without a bayer structure sensor it would do even better.
The corners are of course a different story. I cannot check those. But even the 85PCE and 45PCE lenses do at least 75MP shifted. ( shifted on a J5 means the full FF frame)
So some of your findings may be partially a sensor quality and not to the lensquality.
If i could i would have send in the images so you could see for yourself.
Hi Piers
I too still use J5. Could you please write up a list with lenses are especially suitful for J5 beside the Sigma you already mentioned? Thank you in advance!
Sorry, Pieter of course! I apologize…
+ 1 re the Sigma 40 1.4 Art. It’s a beast even wide-open.
Wow. I came into this article ready to argue, but your argument and evidence is very convincing.
Now I’m really wishing Nikon’s sRAW support didn’t suck so bad. We’re all storing 45mp images for no reason.
“We’re all storing 45mp images for no reason.”
May I suggest to speak for yourself? :)
First, no one forced “us 45 MP storers” to go for a 45 MP sensor. Second, if I want less resolution, I either use a lower res camera, or use DX mode. I found myself very often not pixel- but detail-peeping, and before I can do so I need high MP images.
A while ago I decided to go 3rd party lenses since Sigma came out with so many, often better lenses. I replaced the 85, 24, 14-24 Nikkor Gs on which I was so proud of, and with others I didn’t take the deviation to go Nikkor first.
The S-lenses are a new game, but … it’s not only the lack of some FLs, it’s also the lack of mechanical focus, the lack of f/1.4 and I found my Sigma’s to still have a more pleasing bokeh. and also are up to 45 MP…
I agree on not being forced, but after reading the article I’d rather Nikon had stuck to a lower res for the D850, because I love all of the other features of the camera over the D810 (which I also have).
The article also notes that the Sigma Art lenses aren’t even resolving at 36 MP (“with most lenses (even high-end Sigma Art series) failing to provide sufficient resolving power to take advantage of over 36 MP of camera resolution . . .”). I too use primarily Sigma A and S lenses, and I can believe that claim, although I’d love to see actual test results.
And even if my Sigma lenses did resolve 45MP, I still use a couple of Nikon lenses regularly as well (particularly the 70-200 f/4, which I love) and they affirmatively do not resolve that high. It’s not like I’m going to swap cameras just to use the 70-200 f/4.
No Nasim, it’s not as simple as like if a lens is unable to resolve 45MP, it provides no information extra if compare to shooting with a 24MP body. Instead of looking at charts, it’s easily confirmed by taking real world images. Whichever lens u mount on a 45MP body, stop down to F8, you will see more details/information than with a 24MP body.
This is an awesome quick review list, but it’d be great if it could be extended down to 24mp to know if certain lenses don’t resolve even at 24mp…if that’s the case for any of these
Nasim,
Great article!
I was very impressed when I compared Z7 with 24-70 F/4 S lens and D850 with 24-70 F/2.8 VR and found that little S lens is much better in every possible way (an 24-70 F/2.8 S lens is even better than F/4 version).
Thank you for your time and very useful information.
Nick
I had the same experience testing a Z 7 with 24-70mm f/4 S lens against D850 with a 24-120mm f/4 G lens. A lot more keepers with the Z 7. Some of it can be attributed to the S lens being better than the G lens, but part of the reason is the built-in 5-axis VR of the Z 7, I believe.
Overall I need to digest this test. Some of my lenses score very well or quite well in the DxO Mark tests, and here they fail (85mm f/1.8 G and others). If the conclusion is that a D850 with a given lens can’t print bigger than a D800 with the same lens, then my next camera might well be a Z series camera.
Michael
Sorry, but this piece reads to me like a Tony Northrup type of article. What do you mean by “can’t resolve”? The lens is either sharp (shows a lot of detail) or soft (less detail, blurry, unclear). It has nothing to do with the lens, it’s a sensor issue.
It seems pretty clear from the article and the example. For some lenses, the MTF charts do not significantly improve going from 36->45mp, while for others they do. In other words, the amount of detail in the final image does not increase (much) even though the sensor resolution increased.
You cannot see improvemens in MTF charts because lens manufacturers still measure 10 and 30 lines/mm today (actually calculated), although the sensor resolution has increased overtime. MFT is about contrast in (fine) structures. Think of 10 mm being tree stems, then obviously the higher pixel pitch sensor “cannot” resolve more than an older/weaker lens.
To assess differences between old and new lens (designs/calculations) you better test with finer details, for instance MTFs with 40 and 60 lines/mm.
For those understanding German, check this video www.youtube.com/watch…38;index=4
Very good lesson for all of us (incl. Nasim) by a true lens expert.
Still waiting for the quad Bayer 200mp FF sensor to better handle moire. A sensor that out resolves the lens by an order of magnitude at its native resolution. No more digital noise. At least film grain looks natural, digital noise is horrific.
Highly objective but interesting my Nikon 70-120 f4 does quit well along with my Tamron 70-200 on my Nikon d 850. So don’t give in and go out and buy a bunch of mew lenses over one article enhanced by sales personnel.
But does your 70-200 f4 do noticeably better on a D850 than it does on a D810? Mine doesn’t, and I’ve it quite a bit on both.
This is a very interesting article with some rather controversial conclusions. Nasim is the first person to make such claims to my knowledge. Thus he needs to explain his test procedures and the criteria for his conclusions in much more detail. He shouldn’t expect his readers to accept such claims without more rigorous justification.
I, for one, would like to see his results reproduced by a few other testers before I accept them completely.
Reasonable comment.
I’d also like to see an explanation of the reported result of lower resolution being measured in the first example at apertures f11 and f16 on the D850. That does not stack up with the conclusion that the resolution is limited by the lens at about “3300” (whatever that number may actually signify).
Wouldn’t it just be simpler to draw a horizontal line that represents the Nyquist frequency in the magic number domain on the graphs? That could save a lot of folk getting hot under the collar about removing their anti-aliasing filters.
My personal experience with my 25 f-mount lenses on my D800 and D850 support Nasim’s claim. I know that my comments do not have the scientific rigour that you are looking for, but it may be all you can get. I would look at Thom’s website, he has touched on many of these issues. I think he will respond to emails.
This is not new stuff.
The lenses are not digital. But the sensor is digital. Lenses pickup analog signal and presents it to the sensor to convert to digital. The lenses have limitations stipulated by laws of optics. Here is an article that is fairly technical if one is interested…
luminous-landscape.com/do-se…ve-lenses/
Hi Nasim, thank you for your thoughts, I like and appreciate your blog, in fact I confess at it having swayed some of my buying choices. However, your results in this particular post seem odd and inconsistent with practice.
Just two examples: the linear difference between D810 and D850 sensor resolution is about 12%. How then to explain the 25% differences in Imatest results shown in the graphs for the 28mm at low f-numbers, presuming they refer to MTF50?
And for high f-numbers, how to explain the fact that the D810 comes out ahead of the D850 with the 24mm, despite its larger pixels?
These results do not seem right and perhaps suggest errors in the collection and/or interpretation of the data.
Keep up the good work.
Jack
Hello Jack,
I don’t know how Imatest works but, if the numerical result is for the whole image, a 25% increase is in line with the increase in MPx (36×1.25=45). However, that is be the maximum improvement obtainaible in theory, supposing a perfect lens. I have made a similar remark, and some others, in an above post. No mention of colour resolution is a big one.
Regards
Hi Giovanni,
I understand how Imatest works, the increase is linear (sqrt(45/36)=1.12) so a 25% increase in MTF50 is implausible unless there were differences in the technique at the time of capture (e.g. focus) or processing (e.g. sharpening). Regardless of the data, I believe this title and conclusions are not supported both in theory and in practice.
As for the effect of a Bayer CFA on resolution, you may be interested in this article:
www.strollswithmydog.com/bayer…sharpness/
Jack
Jack, please see my responses to other people, including Giovanni, in regards to how lens data is measured, in addition to my response regarding Imatest response to sharper lenses (no, it is not linear as you assume). I have been in touch regarding this issue with Norman and Henry Koren at Imatest and I sent them sample RAW files from the 28mm f/1.4E lens, which they found to be pretty mind-boggling, as they have never seen this perfect of a lens before. I encourage you to reach out to these people and ask all the relevant questions.
You are welcome to challenge me on any of the presented data, and I’m happy to share RAW files that you can run in Imatest yourself, if you somehow think that I misrepresented anything. You are also welcome to put a high res test chart, put consistent lighting and measure the performance of the 28mm f/1.4E yourself using two separate bodies. This article is aimed to provide helpful information to photographers about most lenses (including ones specifically recommended by Nikon) not being able to provide sufficient detail to take a full advantage of a high resolution sensor. There are exceptions, and I specifically talked about them in this article.
Of course there will be people to whom most of what I said does not apply. If one carefully picks their lenses, buys top of the line glass and makes sure that they cherry-pick good samples, then they will be able to take advantage of 45MP+ sensors. But for the vast majority of photographers out there, and as you can see from the comments here, they are happy to slap on their 16-36mm, 24-120mm or a superzoom like 28-300mm on their D850 body and think they are going to have great results, when their lenses can’t even perform well enough on a 36 MP body.
People like you, who carefully curate their gear and understand lenses and sensors represent less than 1% of the general population. I interact with many photographers, including those I meet in the field, and most of them have zero clue that there is even such a thing as lens resolving power. This article is written for them.
Understood Nasim, and appreciated. As I hope it is clear I like your blog and I mean my comments to be constructive.
I don’t assume anything and when I said that the resolution goes up linearly I was responding to Giovanni: sensor resolution as measured by Imatest with the slanted edge method, increases about linearly with pixel pitch, not with the number of pixels in the sensor. In this case it goes up about 12% and if you are using the same good lens and good technique that sets an upper limit for the increase in MTF for the system. Feel free to ask the Imatest folk or anyone else to confirm this, I will not take it personally.
I don’t doubt your files produce those plots. Therefore I would suggest that perhaps something else is at work. For instance equipment (out of spec lens), technique (vibration, focusing) or processing (you do know that LR/ACR applies a whole bunch of lens and camera specific corrections under the hood, with all sliders at default, including sharpening and noise reduction right?).
I believe the data in this post does not reflect theory and practice, bringing into question its premise. I don’t have the 28mm nor the two bodies you used so I am not able to put together a counter example. But one can’t escape physics.
There are of course diminishing returns as pixels get smaller but no hard limit at 36MP, not even close. Happy to explain further if anything I said is not clear via email.
Keep up the excellent work,
Jack
A hint to one of the variables that may be at play here:
www.strollswithmydog.com/nikon…sharpness/
Great article Nasim!
Hi Nasim, good article. I agree with and/or think the content is reasonable, with one exception. I think you are giving Sony more credit than it is due. While they deserve a lot of credit and have some great lenses, they are “generally” not quite as good as the new Canon and Nikon lenses. Plus there are lots of duds.
Now if I was a Sony user, this is small stuff. I would not change mounts over this.
However, your content does not support the conclusion that DSLRs over 36mp should not be produced. As your list points out, there are several lenses that resolve more than 36mp, several of which I have. There is only one DSLR in the Nikon lineup that has more than 36mp, the D850, which I have.
So I think your headline is wrong.
And speaking to the great content of the article, of my roughly 25 f-mount lenses, I would say that I cannot perceive a difference with a little less than 20 of them between my D800 and D850.
Jeff, I wasn’t referring to the quality of Sony lenses. Of course there are good and bad lenses out there. I wouldn’t expect high resolution output from a superzoom, no matter what brand. But Sony and Zeiss did create lenses that are optimized for high-resolution sensors from the day the system was born, because the first mirrorless camera was a 36 MP A7R. It had a boatload of problems, but it also had a ton of resolution, from the get-go. Canon didn’t have a high-resolution camera until the 5DS R. Nikon did not have a high-resolution camera until the D800. Both have been making lenses specifically for high-res bodies ever since, but Sony made more compared to these two, in the same amount of time. So Sony has a lead for high-resolution lenses, and will probably maintain that lead for a while. Canon and Nikon have started their new mirrorless mounts, and those have a very bright future for upcoming high-res bodies, because each lens is optimized for sharpness.
Very interesting article Nasim.
I basically agree, but it would be useful to see what this actually means when you enlarge a small section of a test picture using one of the lenses that resolves ‘just’ 24mp taken on a D750 and a D850 (or a D810). Ignoring noise/diffraction (i.e. shoot at F5.6, ISO 100), would the D850 picture show any more detail just because it has more MP, or would the detail available be exactly the same?
Thanks!
Burghclerebilly, if a lens does not have the resolving power to provide more detail beyond 24 MP, how would extra detail be born with 45 MP? Take 1 pixel in Photoshop, enlarge it to 4. Do you have extra detail?
Also, unless a 24 MP sensor lacks an AA filter (which blurs details), it would be meaningless to compare the results from the D750 to the D850. For a proper comparison, you need a 24 MP sensor without AA filter, a 45 MP sensor without AA filter (D850) and a lens that cannot resolve beyond 24 MP. I can find the latter two, but not the former.
Agreed, but still would be useful to see the difference between a D810 and D850 on screen (no AA filters) with a weak and strong lens if possible, rather than just lab-test findings, just to sanity check the article.
Is the reality that the 25% extra resolution (36->45mp) would only be visible if you cropped tighter than 5.5% of your original image on a full-HD screen (2mp) or tighter than 22% of a 4K screen (and that would need to be the dead centre of your frame to see the resolution), so it’s a fairly academic exercise? As others have said, it looks like Nikon’s 45mp selection is mostly based on what is the latest hi-res sensor available regarding BSI/ISO performance, rather than absolute resolution, because they know there are very few situations where you can actually see the resolution difference regardless of your lens, unless you like taking pictures of the dead-centre 1”x1” of test-charts.
Click bait with pseudo-scientific results for support. I am underwhelmed. Sure, it’s the same test with I assume the same lenses but I find the results unimpressive because of the lack of description of methods used. How does Imatest work? How do you ensure your results are in fact comparable? Have you changed your target? Have you changed your lighting? What distance are you using? How do you focus? How do you ensure proper focus?
I think this discussion is worth having and I appreciate the attempt, but before making such claims I think you need more explanation.
Good questions, but perhaps you can suggest a different “more scientific” result.
Sean, don’t care for clickbait, never published an article to get the clicks. That’s not what we do at PL here, period. Not sure when you accuse me of providing pseudo-scientific results, when I provided detailed data with numbers to back up my claims. If you believe any of the data is incorrect, please feel free to do your own tests and challenge anything in this article.
In regards to your questions, please take a look at my reply to Giovanni. You will find the answers to all your questions there.
Thank you Nasim,
I am not a pro and I admit I haven’t read all of the feedback. Isn’t this really about what you can see in a print? My impression has been the the limiting factor in the lens vs. camera sensor debate is really the fact that printer output is no better than 300 dpi? Also, with the objective of producing outstanding prints, it’s composition, and photographic technique that win recognition, isn’t it?
Your impression is incorrect.
Print resolution (say 300ppi – not dpi) determines the physical size of the print.
It has nothing to do with lens or sensor resolution except in the following respects:
An image file from a high Mp sensor will produce a file capable of being printed at a larger size at a given output resolution because it contains more pixels.
A high resolving lens will produce a sharper print than a lesser quality lens at the same print size.
By the way, DPI is not PPI. They are frequently confused.
PPI is the output resolution of the printer – typically 300ppi for Canon printers and 360ppi for Epson.
DPI is the resolution of the print head – typically 1440-2880dpi – and is the number of ink dots laid down by the printer on the paper.
Hope that helps.
Correction: DPI is the resolution of the print head – typically 1440-2880dpi – and is the number of ink dots laid down by the printer on the paper – per inch.
Thank you Betty for helping out. Excellent response as usual!
Ur a FUCKING JERK, learn how to take a picture instead of this stupid bullshit.
We all seen level of Nasim’s work, please share a link with examples of your pictures, so we can all evaluate your awesome picture taking skills…
Wow, that’s useful! What a great reply, full of terrific insight and use for future photographers.
Bill, you need a chill pill…
Cursing doesn’t help. If you don’t agree, fine. No one is forcing you. If you disagree with the article, say so and layout your reasons, so that others will benefit from the discussion. A sane discussion will benefit everyone including you…
If that is all you are able to contribute then please don’t contribute at all. We don’t want that sort of thing on here. Hopefully – requesting! – this whole post will be deleted. Do I speak for the vast majority of readers…?
The above comment is supposed to be a reply to the profane post at the top but has appeared out of sequence. Hopefully the meaning is clear!
If anyone is a jerk here, it’s you, Bill.
And what about Canon? Are their lenses so much better then those from Nikon?
Urs, I have not done extensive testing of Canon lenses, but I can tell you right up that many of the high-end Canon lenses, such as the Canon 24-70mm f/2.8L II USM are inadequate for cameras like the EOS 5DS R. I tested 3 different samples of that lens before I gave up – the results were subpar, especially towards the edges of the frame. Camera manufacturers like Canon, Nikon and Pentax did not produce glass in the past to specifically optimize for high-resolution sensors, so very few lenses will be able to resolve enough detail.
Yes, take a look a Christopher Frost’s YouTube Canon lens reviews. Virtually all the lenses he reviews show problems when put on high resolution cameras, especially with implied high-resolution DX cameras of 24mp, which equate to 60mp on full frame (Canon 1.6×1.6=2.56). As well as struggling with sharpness, such high resolutions introduce more chromatic aberration and are incumbered by diffraction at lower f-stops. So glad I kept my D3S!
Even many ancient lenses produce enough contrast at high frequencies to cause loads of false color artifacts and other aliasing artifacts to the image. As long as aliasing exists in the capture there simply is not high enough sampling rate, i.e. enough pixels. Much more pixels are needed for proper sampling.
The alternative would be to use thick AA-filters on top of the sensors, but cause of the simplistic pixel-peeping reviews of the internet that’s hardly acceptable from marketing point of view.
I agree with Any Nyymi. Oversampling improves image quality, even with lenses that don’t bump up the MTF numbers. In general, digital files based on a Bayer mosaic at less than very high MP are full of aliasing artifacts, which affect image quality in a variety of ways. Here is one explanation: diglloyd.com/blog/…pling.html. Here is another: theonlinephotographer.typepad.com/the_o…nough.html
That is why camera manufacturers have created sensor-shifting routines using multiple captures to overcome the image quality weaknesses of 45, 52, 63MP and even 100 MP sensors. This may only interest some photographers, but it’s a real issue for some of us.
So why not push to resolve the issue with aliasing artifacts instead of just bumping up resolution? I would much rather see a 36 MP sensor with pixel shift that works (and supported properly by pp software), than the next 60 MP or 100 MP sensor. Look at Sigma’s Foveon sensors – they provide exceptional amount of detail at pixel level.
Well, that’s kind of a different argument, isn’t it? Right now, high resolution sensors provide the best way, in most situations, to overcome the image quality weaknesses of existing Bayer sensors, which are in almost all of our cameras. Sure, the whole equation will change when and if 16-image pixel shift is perfected, even for moving subjects. Or when and if Foveon sensors are manufactured with much higher resolution. I’d like to see those things happen, too. But those sensors will be sitting in whole new generations of cameras, somewhere down the line. In the meantime, it seems to me that the premise of your article should come with a pretty serious caveat.
The most important question people should be asking is how important is an imatest measurement on a flat surface when most lenses are not flat field? Second more important question people should be asking is why are they worrying about even greater resolution when not actually using it? Third, and lastly, is the question of directed interest: why are people obsessed with the results resolution/test charts vs taking pictures of something interesting? I sort of know that one: it’s easier to proclaim winners than it is getting an interesting photo.
Good questions TC.
Turtle Cat, I simply shared data from what I was able to gather, and I will leave it up to each person to decide how important that data is. To me, it seems like going with ultra high-resolution sensors is pointless when there is very limited glass to support it. Instead of the next 60 MP sensor, I would love to see a camera with a 36-45 MP sensor, but have the ability to average stack images. A camera that can properly do focus stacking, sensor shift and other advancements. That to me is more meaningful than another bump in resolution.
Very good points TC. There is very little need for more than 24mp for most photographers, including most full-time professionals I know. The benefits of new cameras today, including the D850, relate to handling, speed of use, reliability, and hopefully one-day praise the lord, good connectivity!
This is a very good article and nails the point on the ongoing “upgrade” push. We have reached a stage where people spend a lot on bodies and buy low cost lenses. I have said for some time that the 24/36 mp sensors with improved Low Light performance with the D5/D850 processor, focus and metering systems will be a great DSLR success. The D780 offers a great deal for the Video shooter but the still photographer got a raw deal. The D6 will offer the same. Just a lot more for the video shooter and the still shooter gets faster frame rate. The Z6/Z7 has same scope for improvement in my view the demand for 2 card slots. Focus system will greatly improve if Nikon can improve the Black out time of the EVF and the start-up delay. I find that my 300 f2.8 vrii with TC1.4 is sharper and clarity better on my Z6 than my D810.
Concise, objective, pragmatic information. Much appreciated!
This article is an “eye opener”; it was a surprise to see how well some of the tested lenses performed with the D850 and others did not. I use 2 of these “recommended lenses” on my D810 bodies and they perform very well. Looking back: I had no explanation for some of my former lenses working well with my D700 but somewhat disappointing after upgrading to my D810.
I wonder if the 500PF should be on that list of recommended lenses.
This also means the D500 at 20MP on APS-C is also about the highest they should go.
Jason, yes, and I highlighted that in the article.
I am heartbroken and I cannot take a photo until this is resolved.
A+++
LOL, good one :)
So much hate for Sony lenses here, it makes me think people are referring to a-mount lenses? Most Sony FE primes have class-leading sharpness (reference the 90mm
f/2.8 G macro, 24mm f/1.4 GM, 55mm f/1.8, etc) and most zooms compare well with even Canon’s best. Some, such as the 12-24mm f/4 G and 16-35mm f/2.8 GM, are even class-leading. Sample variation has been an issue but recently releases have shown that Sony is improving in that regard.
Regarding the assertion that Nikon needn’t bother using higher resolution sensors in their DSLR’s, that’s rubbish. Sure some f-mount lenses don’t resolve well but 1) many do, 2) committed f-mount shooters are demanding these new sensors, and 3) most importantly, Nikon uses whatever sensors Sony makes available to them and the latest tech is in the new higher resolution sensors. The last 35mp sensor was released many years ago and most certainly wouldn’t perform as well with respect to signal/noise ratio and other metrics as the latest sensors do.
Thanks for writing Nasim – Excellent article
This article is being debated in DpReview under the Z forum and one point there that was made is that your improvements for the D850 and the Nikon 28mm f/1.4E ED lens seems to be greater than the linear increase in the sensor over the D810. I’m certainly not debating it one way or the other, but wonder if you have some explanation for that seeming contradiction?
Mark, Imatest does not increase its numbers linearly with higher resolution sensors. It uses its own algorithm to assess edge detail and provide the data. I have already exchanged emails with Norman and Henry Koren on this subject and they understand that some lenses with extremely high resolving power (such as the Nikon 28mm f/1.8E) might appear very sharp in the center compared to the corners. As you can see in the second graph presented in this article, the center sharpness looks 2x better than the corners, which makes it look like the corners are far worse. But that’s certainly not the case, at least when it comes to perception of sharpness when looking at images.
Hopefully Imatest engineers address this problem and come up with a way to provide more meaningful data that can correspond linearly to increased resolution.
Reasoning seems to be sound, but I don’t understand how Nikon made such long-lasting success with D7200 if they really don’t have adequate DX glasses for APS-C 24 MP sensor?!
Drazen, take a lens that cannot resolve, put it on a 16 MP DX vs a 24 MP DX camera and see if the latter yields any additional detail. Hint: it won’t. The biggest improvement we have seen with DX is not resolution, but improved dynamic range and SNR.
Mason, what interests me as a half-beginner is whether too high resolution actually degrades the pic or not; I mean whether “excessive resolution” attempts to resolve detail from inadequate lens, fails and leaves some sort of blurriness, degradation that is visible. Or not.
If not, if insufficient details resolving doesn’t degrade that image, than, higher-res body can have equal or better quality from any lens plus much better cropping ability, and in that case “inability to resolve details” has no much of practical consequence imho.
I appreciate your article and especially your comfort with controversy.
I have one question- what affect does adding a teleconverter have on resolving power, specifically with the super telephoto lenses?
Dave, teleconverters have a huge impact on resolving power. Please see the article on teleconverters for more information.
Holy cow! Just skimmed the comments. Boy, when the PL Nikon digital fan boys come out in response to a pure technical article like this, they really do come out in force. Don’t they? 🙄
Thom Hogan has a very similar article, which started from the Nikon recommendation of the D800 series lenses. See link below.
www.dslrbodies.com/lense…ses-2.html
I find a lot in the article to agree with, not including the title. My beef with the title is this: Really it comes down to what you’re doing with your camera and lens(es). As a birder, probably 90% of the photos I take are with the 500mm FL prime (a lens which Nasim says will resolve greater than 36MP), and I definitely can tell a difference between D810 shots and D850 shots with that lens. The D850 is superior, no question. So for what I use the equipment to photograph, the 45MP of the D850 is not overkill.
If you don’t have the 500mm FL lens (or any of the others that will resolve 45MP), then maybe for you as an individual there’s no point in having a camera body with a higher pixel count (so don’t spend your money on one), but that doesn’t mean Nikon shouldn’t make such bodies. They should make such bodies for those of us who benefit from the higher pixel count (and maybe refrain from marketing them to those who would be better off with a lower pixel count, but that’s too much to ask).
This brings up the next issue: Should Nikon chase the 60MP sensor? Well, let’s see if nature photographers toting the best glass get better images at 60MP than they do at 45MP. If they do, then maybe I’ll be interested in a camera with 60MP. I saw a significant improvement when I moved up to the D850 from the D810, but only in limited circumstances. In bird photography, once the subject gets out past a certain number of yards, the distortion in the air often makes more pixels moot. That is, the detail simply isn’t capture-able no matter how good your lens is or how many pixels you have or how perfect the light is. But for hummingbird-sized subjects at 15 feet, I don’t feel like I’m wasting my money using a D850 instead of a D810.
I agree with Nasim that there are things, other than more pixels, that I’d rather see in the next Nikon body. Of course I say that without having a 60MP camera to pair with my 500mm lens to see if it can resolve that much detail. IF it can, I might be interested in 60MP. Birders are always going to want as many pixels as their best lens can resolve. They buy the longest telephoto they can afford and/or lift, then they crop many of the resulting images to “zoom in” even more.
Well said, TDW. Including the title… Nasim, such „garish“ titles just invite comments like the one from Steven Green further below. He concludes things you did not test or say and jumps to grossly overstated conclusions (… any decent lenses…). Pity for the money spent on wrong gear.
Robert, I don’t think we should worry about people who do not read articles and come up with wrong conclusions. I provided factual data in this article based on my assessment of many lenses – I will leave it up to people to make their own conclusions.
There is nothing wrong with the title of the article, as I do believe that camera manufacturers need to stop pushing for more resolution on cameras when the rest of the echosystem cannot support it. I would rather see more meaningful upgrades that actually matter – better, faster and more accurate AF algorithms, IBIS, pixel shift, image averaging, simpler and better ergonomics, etc. The camera market is collapsing and smartphones are taking over. We need simple tools that work, plain and simple – not another 60 MP or a 100 MP camera that does the same thing as its predecessor.
Nasim, I think Internet sites like yours should worry – at least a bit – about people who do not read articles properly in that sense, that those internet sites are spreading information and wield influence. Therefore, I think it is definitely not only „my opinion“ and caveat emptor. Especially not, when the article represents only a rather simplified measurement of a single aspect, torn out of greater context.
I wanted to argue, but was not able, because the findings appear at the first, superficial glance really beyond doubt. My theoretical knowledge of optics is small, so I was not able to get to the root of my hesitations about this article spontanaeously. I just remembered, that this theme was worked on already and consensus was different from your „conclusion“ as it clearly is palpable in the title. Since yesterday I spent much of my time researching and I suppose you did it too. If not, just read Jim Kasson, Ctein, Lloyd Chambers, Thom Hogan, etc. on this. They do not restrict their thoughts on one single number, but an the greater „thing“, the image quality (which should be important in the end). E.g. Jim Kasson explains in dPreview forum with very few words the flaw of the method you used.
I understand you wish more features from the Z-Mount in cameras with F-Mount, now that they are here. But is it logical? Technology brought us something new, doable in mirrorless format, why to step back and ask to apply all of this on the technology which – as you otherwise write and the great public believes anyway – is of yesterday? Neither me nor – I suspect – you would be ready to pay such premium over the mirrorless price.
Take care
Agree… Sharpness is one of several variables of the ability of an optic to render a high quality image. Despite MTF presenting a synthesis of proxies, it is only one methodology to portray data graphically. This is explained rather well in a classic article on Luminous Landscapes back in 2011 (cited in these Comments earlier).
Thank you for this article. It confirms my reading from elsewhere: that although Nikon SLRs have superb senors in them, for me the Canon system wins hands down because of its superior lenses (particularly zooms). For example the 5dsr has an inferior sensor but this can be mitigated by multiple exposures. The effect of a poor lens (relative to the sensor that is) cannot be compensated for in any reasonably practical way. Also why should it be necessary? Why can’t Nikon make consistently decent lenses?
Stephen, Canon has the same issue as Nikon with its older lenses. Aside from lenses released in the past 5 years, most older lenses are not going to cut it with cameras like the Canon 5DS R. Similar to Canon, Nikon makes excellent lenses. You should re-read the article, because your conclusion is wrong…
Hi Nasim,
you certainly created a lot of stir within the nikon community :-). I liked your article very much. It very much reflects my personal expierence, too.
So I did not buy a Nikon D850 2 years ago, when I had the chance. I am not a professional and to me spending more than 3000€ on a camera seemed not reasonable, though the D850 fascenates me to this day. I also did weigh in the fact that I needed to upgrade some lenses too. So I saw myself in an kind of upgrade cycle.
Acquiring all necessary information (in terms of lens quality), then selling old lenses and purchasing new lenses. It all costs time and money. Which is for one self to decide, if that is worth it.
These were the very practical thoughts that led me to decide for a less resolving body, anyway. And the saved time and money was well invested in a lot of fun creating great images / memories.
After all, that is what it is all about, right?
Look at the China built cell phone cameras that now record 100 MP. Which is pretty close to outright fraud. The chase for extremely high megapixels reminds me of my own hugely expensive foray into medium and large format. Is sure do not miss that at all. For my purposes I’d still rate the D7500 and D500 as the two best DSLRs for ME! I do like to own a D850 and some of the best lens. I now own the Z50 Nikon and the two excellent kit lens. I have been able to do all I want with the 20.9 MP and compared to my fleet of 24MP Nikon DSLRs, like the D7200 the speed and handling of the D7500 and D500 is far more impressive.now. Also the video is better. Lastly the Z50 is a really great adventure camera. The two kit lenses are awesome. Particularly the 16-50 lens.
My experience confirms much of what Nasim said. I have the D810 and the D850, and had the D800 until I gave it to my son. My experience is that the D800 took marvelous pictures, but the D810 was an improvement, especially in handling. My lenses resolve fairly well with it. The same lenses don’t do so well with the D850, but I can still get great pictures with them. I think you have to find your lens’s sweet spot when using the D850. In general, there is a clear superiority in the quality of photos over the D810/800. I would love a few dedicated lenses for the D850, which would have the resolving power that the camera needs. But even without them, where the 45 megapixels excel most is in editing. You have so much to work with, that you can crop the heck out of a photo, and still have full size mp rich pictures. Other post processing procedures benefit too. So there is a very good reason to go for those 45 mp, if you do this kind of work.
Did the sequential upgrades from D800 -> D810 -> D850 make you a better a photographer?
In fact, they did. When I went from the D300s to the D800, my skill made an enormous leap. I had to learn to work with a full frame camera, and anything that causes you to do things differently, makes you learn and so makes you better, providing you put in the work that it takes. The D810 was pure pleasure to switch to, because it did a number of things better than the D800, and I can tell you that I loved the D800. Working with the D810 improved my abilities enormously. The D850 is a very hard camera to master, and working towards mastering it makes me a more skilled photographer. I am still working hard at achieving my best work with the D850. Sometimes when I see what I can do with that camera, I am amazed. Whatever exercises your brain and makes you try harder, makes you a better photographer.
So, in other words, the latest greatest gear makes the photographer. Right?
No, it’s the gear between your ears that makes the photographer. The camera gear just helps.
An interesting article Nasim but I fundamentally disagree with your casting out of some of my favourite lenses. In particular the 16-35mm which is an immensely usuable and convenient tool. My condition is maximum data but I also come from years of film and actually like the D500 grain. Personally I hope that Nikon do a 60MP camera. Meanwhile hurry on the 500mm pf
Charles, it is not about casting out lenses. It is about providing factual data. Like it or not, it is a fact that the 16-35mm cannot resolve. It is not a bad lens by any means, it just cannot keep up on high-resolution cameras like the D810 and D850. I used to own this lens for many years and provided very high remarks in my review. But with time, it is showing its age and its weaknesses.
Well… I don’t have the knowledge to challenge the technical part of this, but I have one question (disclaimer: it might be very stupid):
At small apertures due to diffraction, lenses lose acuity, correct? So why is the D850 still registering more detail than the D810? Shouldn’t there be a point where 36 and 45 Mpix just don’t matter ? See the D810 with the 24mm can register a score of 3300. How is it that when score drop to e.g. 3000 on the 28mm with the D850, the D810 can’t score the same?
Then a remark about the title: I disagree. Nikon SHOULD make DSLRs above 36 Mpix, for three reasons:
– there are lenses where there is a clear benefit, you show it clearly in your article
– there are also non-Nikon lenses…
– the point is not that they shouldn’t make high res bodies: they should make better lenses
Seb, yes, diffraction affects edge detail in a very severe way, as evidenced from the graphs. This is why I do not find it useful to try to measure MTF at smaller apertures like f/22. Heck, with modern high resolution sensors, I even question the point of measuring f/11 and f/16!
In regards to Imatest numbers, I am not the engineer who created the software, so I cannot fully answer your questions – you might want to take them up with Imatest and see if they can provide the answers. My assumption is that diffraction affects numbers differently on sensors with different resolution. Technically, edge detail should look the same at f/16 on both 36 MP and 45 MP sensors, but the larger image is going to spread that detail over more pixels. Perhaps the algorithm does not have a way to fully compensate for that, I do not know.
Keep in mind that Imatest measures the “system”, which combines everything from sensor data, lens sharpness and even RAW processing by the manufacturer. So if there are slight changes in the way RAW data is processed, or a slight change in the way the filter stack in front of the sensor is made, it might have an impact on the numbers.
Regarding the title, I will leave that up to each person to decide. People buy cameras based on what they know, and they should know that most Nikon lenses are incapable of taking a full advantage of higher resolution sensors. The same goes for other brands like Canon, but since I do not have adequate data for the whole system, I cannot yet say that. And yes, manufacturers should push for better lenses first, then worry about more resolution.
Seb,
Theoretically, diffraction is the only one that sets the resolution limit for a given system, and that is dependent on the f number of the lens.
This diffraction pattern is called an airy disk (bright region in the middle and concentric circles around it). As f number increases, the airy disk size, (esp the concentric circles) increase in size and start to overlap one another, affecting the overall image quality.
When ‘pixel size’ is brought into the this, as the airy disk increases in size at smaller apertures, it can vis a vis exceed the pixel size, thus affecting the image quality. Generally, as long as the airy disk size fits over couple of pixel pitches, the image has acceptable contrast levels, but when that size gets spread over more pixels, then the image quality starts to suffer.
In essence, the lens has to be designed and optimized for new high resolution sensors so as to have small airy disks at a given f number. Better design, glass and manufacturing processes should be able to address that in the near future….
I am not sure if it makes sense, but I tried my best…
There are several ways to consider acuity of a lens for one’s genre(s).
Some genres definitely need FX sensors of 60mp – and more – in FX cameras. Landscape photography is one IF huge prints are an important output.
Wildlife photographers will appreciate that much more latitude to crop – within finite limits (ie injecting more noise). Basically, it’s the quip of having more pixels on the proverbial duck. A 1:1 / DX crop of a 60 mp image = 40 / 26.4 mp versus 30 / 19.8 mp in a D850.
On the other hand, merely average MTF profiles of a collection of dependable lenses do not – indeed should not – panic photographers to trade-in established favourites and buy “all new”. Lenses are tools first and foremost. It is a truism that priorities for highest IQ vary across genres, and are subject to different factors. There are “older” F-Nikkors (eg not only 16-35 but also 18-35 G) that are still respectable landscape U-wides on a high-res FX camera. One reason is we usually shut them down to f8 or more. And many “inferior” / older lenses remain ideal for travel and our other less demanding uses.
The 58 f1.4G aka the Neo-Noct is another topical example. Apparently, it just scrapes into the “Nikon 45mp Top Tier of Lenses” (see above and the Hogan FX List). This unique Nikkor prime confuses no end of photographers and “reviewers”who treat MTF scores as the Holy Grail of Photography. But some of us know what it delivers where and when, and how. Such an optic renders the image in unique ways. Nikon’s respected engineer Haruo Sato has described the quest of his team to design and manufacture such primes (see interviews over the past few years on the Japanese site dc.watch.impress).
Thus, why some of us still choose even older lenses on the D850 and Z7, such as the legendary Cream-Machine’ the 85 f1.4AFD. It delivers superb portraits on modern sensors. The corollary? Too many geeks ignore that photography is often more art than science.
Maybe I’m rusty but I would imagine that the system mtf would behave like a convolution of the lens’ mtf and the sensor mtf. Assuming the lens’ mtf is non-zero for a particular spatial frequency, wouldn’t this imply improving sensor resolution matters? I don’t think I buy the argument presented in the article.
A pointwise multiplication [pointwise product] in the frequency domain (e.g. lens OTF times sensor OTF) equals convolution in the spatial domain.
“In mathematics, the convolution theorem states that under suitable conditions the Fourier transform of a convolution of two signals is the pointwise product of their Fourier transforms. In other words, convolution in one domain (e.g., time domain) equals point-wise multiplication in the other domain (e.g., frequency domain).”
en.m.wikipedia.org/wiki/…on_theorem
NB: The modulation transfer function (MTF) is only the modulus component of the complex optical transfer function (OTF), it ignores the phase component.
“Since the optical transfer function[2] (OTF) is defined as the Fourier transform of the point-spread function (PSF), it is generally speaking a complex-valued function of spatial frequency.”
en.m.wikipedia.org/wiki/…r_function
Great article Nasim.
I can take a crop from a shot with the D850 with a good lens and get a descent result. This is a relatively rare need. I do not blow up my photos to cover a wall. Can someone tell me why I need 46 MP? I upgraded from a D800 to a D850 for the features (focus speed, etc.), not MPs. I would have much preferred 20 MP with better light sensitivity.
This seems to be why ALL adapted lenses I have tried on my Z7 are terrible.
A certain sample variation may exist in the adapter or Z7 body, but no authority has weighed in.
John, I measured MTF on the D850 and compared it to the MTF results from the Z7 with FTZ adapter. Here is the page of the Nikon Z7 review that discusses it in detail. Basically, there is no difference between a lens mounted on Nikon D850 vs Z7 – provided your FTZ adapter has no problems. In your case, if your lenses are not doing well with the FTZ adapter, there might be an issue with the adapter. It is possible the mount on the Z7 is decentered, but if that was the case, your Nikon Z lenses would also suffer as a result.
I for one am more than slightly surprised to see a Tokina, and a zoom lens at that, on the list of high resolving ones. I’ve read many good things about its optical qualities (usually followed by “a pity the AF sucks horribly”) but I didn’t think it’d be *this* good.
A very nice article, I was skeptical while reading the title and after the first couple of tests but then I saw what you mean when talking about not being able to resolve 36 mpx or higher. I can’t say I’m surprised by the results but I admit I thought more lenses from Nikon would be able to resolve 36 mpx or more, making a super high res camera without a good amount of lenses that can use that resolution is…counter-intuitive, I guess. Seems like all manufacturers are now making such lenses with very high megapixels in mind (100max on FF? Or thereabout). I personally wouldn’t mind an increase in resolution over my usual 24 mpx but I think 36 or a bit less would be just lovely. And would probably be fine for most lenses too, which doesn’t hurt either.
Thanks for all the research you put in this, it was much appreciated. Cheers, have a lovely weekend.
Hi Nasim:
Thank you for a most interesting and helpful article, which gave me much food for thought. My question is a little off subject in that it refers to Fuji cameras rather than Nikons.
For the past 20 years I have shot with Nikon cameras. I left photography for a while to concentrate on video work, and then really missed the joy of having a photograph than can be studied and enjoyed. I decided it was time to go mirrorless, and, wanting to produce large prints, aquired the new Fuji GFX100 camera. I was very impressed by it, but it had a problem that caused it to shut down, and ultimately stopped working so I returned it. Would you be able to comment on the resolving power and quality of the Fuji GFX lenses coupled with a 102 Mp sensor? Are they good enough to make the camera worth buying?
Thanks again!
Boy Nasim, a can of worms indeed! All-in-all, this is indeed a major shift in camera resolution technology that impacts future lens development…but the sky isn’t falling…yet. Also, as you know, the D850 already has in camera automatic focus stacking–if you have the need. I’ve been shooting with the D850 for about a year and half and my work-horse lens is the 70-200 f/4. The same one I used on my D750. The combination exceeds my general photography needs and expectations. And I’m VERY picky. Case in point: yes the glass needs to be upgraded to match the sensor resolving power. So what else is new? It’s still all about skill, color, light and texture. I would feel very confident that many of your previous F Mount images shot at 24 -36 MP are no less spectacular to your latest shot on the Z with the latest glass. Let’s all take a deep breath and practice our art.
If a lens has enogh resolving power, will a dx image be as sharp as an fx image?
I see where you are going with this but the list of lenses is longer than you have shown. Nikon’s list of recommended lenses hasn’t been updated. Off the top of my head, I bet these three would qualify
Nikon 500mm f/5.6 PF
Nikon 180-400 f/4
Nikon 120-300 f/2.8
I suspect some new third party lenses should be on the list as well (like Otus) so the list of lenses that can take advantage of the higher resolution is probably closer to 50. That is plenty of reason to add high res bodies.
Mike,
All three lenses are mentioned in Nasim’s article, I’ll save you the bother of actually reading it:
“Nikon included the older 200-400mm f/4G ED VR II in this list, which couldn’t resolve past 36 MP, whereas the newer 180-400mm f/4 VR did much better, clearly providing more detail than its predecessor, even wide open at 180mm. Nikon should update the above list (last updated in August of 2017) and include the newly announced lenses, including the AF-S Nikkor 500mm f/5.6E PF ED VR and the AF-S Nikkor 120-300mm f/2.8E FL ED SR VR. Both should provide sufficient resolving power over 36 MP.”
I didn’t say the article didn’t didn’t have them I said Nikon lists didn’t have them. I also disagree with this comment:
“With less than 30 total Nikon lenses that are actually able to take advantage of the modern 45 MP sensor on the Nikon F mount, one would wonder why Nikon should even try to go beyond 45 MP sensors at this point.”
For 2 reasons. First, I think there are more than 30 f-mount lenses that can take advantage of higher resolution. Two, even if it’s “just” 30, that is still plenty and they cover a huge range of focal lengths, surely enough for many photographers. There are more lenses coming as well, so the whole argument is moot.
Hi Mr. Nasim,
I’m watching around to buy a prime lenses set for my D850 and this article surprised me.
In your previous reviews about Nikon 24mm f/1.8G ED and Nikon 35mm f/1.8G ED, the first one looked really better in MTF tests.
In this article, on the contrary, the 35mm outperforms the 24mm one, when mounted on a Nikon D850.
Did you use two different cameras to test 24mm and 35mm in your reviews (the single lens reviews) or was the same body?
If it was the same body (and the 24mm lens resulted better than the 35mm), why, in this article, the 35mm lens is able to outperform the 24mm when both are mounted on a D850?
I’m not an expert, so I’m sorry if I’m asking something trivial.
Thanks for everything!
Riccardo
Nasim, thank you for this article. You highlited very important aspect. I’m commercial photographer shooting AD, primarily cars and trucks. So, I badly in need of resolution. The more – the better. I have had a lot of usage Phase One IQ280 (renting) and Leica S (i had partnership with Leica for few years). But then I almost completely switched to Nikon and use leica S just as backup. While Leica lenses is simply best, with Nikon I can work much faster and make up to 10-12 photographs per day. For example with Phase One I can produce no more then 3-5 pics per day. Well, working with Phase One in the field with small crew is simply nightmare.
Well, I need resolution because usually smallest print size of my photographs something like 1.5 meters and largest was 8,5×6 meters. And clients still want a good level of details (it should look well from the distance of 1 meter If we talk about lite-boxes on the events, presentations and automotive exhibition).
But return to article. Currently I use Nikon D850 as my primary camera. Below is my kit and my impressions. Maybe it will be useful for somebody.
100% of my cars and trucks still photographs made with tripod, electronic first curtain, exposure delay etc. Usually my working apertures are f8/-f/13 for still. For moving objects it depends of shooting conditions, using or not rigs etc. Well I use all starting from f/1.4 to f/13
Which works:
Nikon 20/1.8G – better sharpness than 24/1.4, but I use it only when can’t use 24/1.4G because of “look” – colours and contrast
Nikon 24/1.4G – good sharpness on D810, not so on D850 but colours and contrast very good, poor geometry. But I always prefer pics from 24/1.4G over 24/1.8G and 20 /1.8G in blind test. Planing to switch from 24G to 28E
Nikon 35/1.4G – so-so on D810, not good on D850, but contrast is really good
Nikon 58/1.4G – very special lens. I like it very much, but it really not good choice for product photography because of field curvature. Brilliant colours and contrast mask not so high corner resolution well. If you know how to work with this lens – 58G produces great result, simply it is best “normal” Nikon lens in current Nikon F-line. My workhorse #2
Zeiss Macro-Planar 100/2 – was my workhorse #1 if I don’t work when raining. Very good resolution, best colours and contrast of any my lenses for Nikon F.
Nikon 105/1.4E – bought it 2 months ago and already shoot with it AD company for SCANIA. Currently it simply best lens of all my kit. Now it is workhorse #1. Very high resolution, beautiful colours, contrast and ”look” at all apertures. It the only lens I can compare with Zeiss Otus and Leica S lenses. Nikon 105/1.4E replaced my Zeiss Macro-Planar 100/2 for all shots except macro.
Nikon 300/4E PF – not good sharpness, but pics look nice. If you need to shoot moving subject it is best lens around 2000USD with very fast AF and good VR.
Doesn’t work:
Zeiss 28/2 – poor resolution in D810, great colours, but good “look”, high CA
Zeiss 35/1.4 Distagon – poor resolution on D810, good colours and contrast, high CA
Zeiss 50/1.4 Planar – bad colours, high CA
Nikon 24/1.8G – good resolution on D810, not good contrast and colours
Nikon 28/1.8 G – not good resolution on D850
Nikon 35/1.8G – good resolution but impossible to shoot in difficult light conditions because of flares
Schnieder 45/2.8 Till/Shift – very poor resolution (I think just about 16 Mpx) but great colours and contrast
Nikon 50/1.8G – poor resolution on D810, really bad colours, ugly “look”
Nikon 85/1.8 0 – ok resolution on D810, not ok on D850 at infinity, good colours. Good lens, but 85mm just not mine focus length
Nikon 14-24/2.8G – poor resolution on D810, bad colours
Nikon 24-70/2.8G – poor resolution on D810, bad colours
Nikon 24-120/4G – worst nikon lens I’ve ever had, very low resolution, ugly colours etc. But ok for video
Nikon 70-200/2.8G VRII – poor resolution on D810, good colours and contrast
Well, all these info just my experience for my type of shooting scenario. Your experience may be opposite. For example for my non-commercial doc projects I can use any lens and any camera to shoot with.
This is a useful article and makes an important point. I’d just wish for one slight nuance, however. The MTF figures for the “not able to resolve 45MP” lenses are, in fact, a bit higher on the 45MP sensor than on the 36MP sensor, and this is exactly what MTF theory predicts. The Imatest web site explains this well.
Now, the difference may be so small that it’s not visible in any practical output, but it would be good to acknowledge that nuance at least once somewhere in the article. The reason is that this is not an either/or phenomenon — i.e. there is no sharp cutoff where a lens is able to resolve well enough for an X resolution sensor, then it can’t resolve anything past that resolution.
Instead, it’s a phenomenon of diminishing returns — i.e. the magnitude of the increase in MTF gets smaller and smaller until it’s no longer practically visible. But there’s all kinds of mushiness in there, depending on the lens, the aperture you can or want to use, and the output you need. A very small increase in MTF that might not be useful to me might be useful to, say, a forensic or scientific photographer.
Eamon, very good points. In fact, I would argue that the reason for slightly higher MTF have to do with the Bayer pattern and artifacts – higher resolution sensors are going to be better in that regard. This is the reason why some techies are pissed by what I wrote in this article, because their argument is that a higher resolution sensor is always going to be better than a lower resolution sensor, no matter what lens. However, unless resolution differences are vast (16 MP vs 36 MP), most people are not going to be able to tell the difference. Few people are going to assess images at 200% zoom captured with a D810 and look at the same pixels from the D850 to see which one looks better. If detail is not there, both images are going to look like shit at 100%. That’s what matters, and that’s what I wrote about here.
What about Sony’s A7riv?
Nasim, I for one have valued your time, dedication and professionalism in researching, testing, writing articles and replying to comments from since before there was PhotographyLife.com. I found this article very interesting and worthy for any future camera equipment purchases. For me what matters isn’t 36-60 MP sensor but the other camera attributes you mentioned. Ultimately, as in selecting other electronic equipment like computers, the more important question seems to be “what do you intend to do with it?”. As primarily a nature and wildlife photographer who exhibits and sells my works regularly (average size 24″x36″), I still love the Nikon D750 and have resisted “bigger is better”. Capturing the image right in camera — composition to specs — always comes first. This allows little need to crop and the ability to enlarge up to 40″X60″ with remarkable results, even with the Nikon 24-120mm F4 G ED VR lens .
I noticed a huge difference using some F lenses ( old version ) from my Nikon D700 to current D 750
I believe that some lens design images and the aberration are amplified with more pixels.
Nikon is reducing the pixels in some cameras to avoid the obvious : more pixels more imperfection.
I believe the Z system lenses are better design to cope with ever growing pixel demands.
The idea of using “classic” lenses in new digital cameras it’s fantasy.
yep, some classic lenses do very well with classic DSLR’s (12-16Mp FX and 12Mp DX)
just as well as they did 10 years ago, but all this gear has become way cheaper since all the hype has moved to high resolution gear. That’s all good for amateurs on a budget :)
Nasim, the article title and conclusion are only valid in a world in which Nikon is the sole manufacturer of F mount lenses. There may be such a world but it isn’t the one I live in.
You touch on this with “Some of the modern third party lens options can surely fill the resolution gap, but that’s for those who are willing to do the research in order to find the very finest, or alternatively, shell out a lot of money to get the exotic Zeiss Otus lenses.”
Surely such research isn’t much of a challenge in the internet era. For those not familiar with it could I suggest the Lensrentals blog as a good starting point.
A far more useful and balanced article would have been something along the lines of “These are the best lenses to use on high resolution Nikon bodies.” Once you accept that the independent lens manufacturers such as Zeiss and Sigma often make better F mount lenses than Nikon (and I’m sure you do), then your conclusion becomes invalid, hence the accusations of a clickbait title.
He specifically mentioned Sigma lenses in the article, and how even the Art line doesn’t typically surpass 36mp.
Martin, the article title and the conclusion are valid for 99% of photographers out there, who have no idea what lens resolution means, and who use their incapable lenses on modern cameras without understanding the output. The 1% of people who carefully pick their gear and know which lens to couple with which camera can safely ignore this article, or simply draw their own conclusions from it.
Nasim
As usual, a very good article with information that many of us pros have known and put into practice. People tend to believe hype instead of doing the hard work involved in creating images that “sing!” Tech can take you just so far….
Hello, Nasim,
thank you very much for the article.
Basically I choose a camera according to how it lies in my hand and how I can handle it, how good the optical viewfinder is, etc. The sensor resolution plays only a minor role for me. But it is correct that many F-lenses are older calculations. The newer E-lenses are usually much better. But there are lenses from other manufacturers which are much better than the original Nikon lenses. I am thinking for example of the Sigma Art 1.4/40. In my experience this lens is one of the best lenses ever developed and can keep up with or even surpass all Z-lenses. I suggest to test this lens, you will be amazed! I also think very much of the publication of MTF curves. These only serve the marketing department. A comparison in real life would make more sense. With the large format lenses of that time, the MTF curves were created using a collimator. The results obtained were objective and very informative.
For me, the F bayonet system is not yet at the end, the Z bayonet system has been added.
Many greetings from Germany
A very good article describing some basic truths.
Many older (and even a number of current lenses) are not up to the task when confronted with a high resolution sensor. Newer lenses are better than older ones but only on a newer and better sensor. Given that fact, there seems little point in most people investing in high resolution cameras without also investing in glass appropriate for them. There is no more to it than that.
What is disconcerting is the number of comments from amateur ‘experts’ in optics desperate to have their confirmation bias reinforced – making ill informed arguments in support their own pet (usually half baked) theories about their own pet, underperforming lenses. Some have clearly not even taken the trouble to read the article attentively before launching into their ‘critique’. As Nasim points out, measuring lens resolution is a complex and difficult undertaking with different methods producing markedly different results. How then are these bedroom commentators qualified to question the article’s findings based only on their exhaustive reading of reviews and their own subjective (usually biased and inaccurate) opinions?
What is even more disconcerting is the disparity in the number of comments that these technical articles attract compared to those about practical photography. Technical specs, while not unimportant, seem to be an obsession for far too many, to the point where one has to wonder whether these people ever actually take photographs of anything other than test targets.
For heaven’s sake, from a purely practical standpoint, unless one is making very large prints or involved in technical work demanding ultimate resolution, these differences in resolving power are immaterial to the production of good photographs. Yes, factors like cropping and noise come into play but in general by the time a sharp, well exposed shot captured on a 45Mp sensor has been post processed, sharpened and downsized to 1200px for web or screen, it will be indistinguishable from one taken at 20 Mp.
Hi Betty
Usually I like your comments very much. This time I think you could better. Only two points, otherwise I am finished with this essay and „discussion“:
1) The title is misleading and wrong – even if e.g. your last sentence were true, stopping to make higher MP-bodies would mean to stop further research etc. Btw exactly such titles provoke readers to hasty and even harsh comments. They are unsure and unsurety gets mankind a bit out of equilibrium. It is nice to help Nasim, but this essay simply has some shortcomings and teking into account that every blogger has some kind of responsibility what he/she writes, it is not nice to come down on „stupid“ readers.
2) I am no scientist, so no (false) formulas here, but big prints from those two cameras you mention are distinguishable. I do prints between 60cm and 1m and a can see the difference clearly. Of course, it is not only due to resolving power, but heck, I think you know. Otherwise please read papers from Jim Kasson, Lloyd Chambers, etc. They are knowledgable in optics and mathe, which I am not.
Please, no bad feelings, take care. For me, this essay was beneficial because it roused my interest and I am studying the essays I mentioned – first the one Jim Kasson referenced in dPreview forum, afterwards the essays from Ctein, and then something from Lloyd Chambers… We shall see, whether it goes…
Take care (once more), sorry for my English, my mother tongue is different and the language of the county I live is different too 😉😃
Robert,
1) No, stopping to make higher resolution bodies does not mean stopping further research. This is what many people wrongly assume. Most APS-C cameras have pretty much stopped at 20-24 MP, does that mean that development of 20-24 MP sensors has also stopped? Of course not. Why not spend the R&D resources to make superb 36 MP sensors before we move on to 60-100 MP sensors for the sake of increasing pixels, when most lenses won’t perform? If I were very picky when grading lenses for this article, over 80% of them would not pass, especially most zooms. If a 24-70mm can only exceed 36 MP at 24mm stopped down to f/5.6, but cannot resolve at other focal lengths, it still got a “pass” in my chart, but for my personal use, I would find it to be inadequate for a high-res body. Also, keep in mind that this article was not written for the 1% of people out there who carefully curate their lenses for their camera bodies. The general population buys into the high-resolution game, get sold by blogs and influencers who speak about amazing detail they see in their images, and recklessly buy cameras like D850, only to pair them with a 28-300mm superzoom that can’t resolve even 24 MP. Most people do not understand the importance of having a strong lens that performs. I take everything I write very seriously. A smart person who reads into what I write, perfectly understands the premise of the article and the message I am trying to convey, without focusing too much on details and the title. Heck, look at some of the comments – people posted them without actually reading the article!
2) I think you need to re-read what Betty wrote about printing and final output.
Hi Nasim!
Thanks for your answer. Point taken, still thinking the title is too… Toll on (over)simplification or whatever. Never mind, can live with it. About people not reading an article but commenting it: Pity, partly virtue of Internet, one can just let loose frustrations or whatever, without responsibility. Maybe write shorter articles? (Joking only).
Yes, Betty wrote „..for web or screen..“, I have overlooked it and some others would certainly too. Aiming for kind of absolute conclusion and quickly restrain it.. Catch for those „quickreaders“. Again, no problem: Would I shoot for web or screen, I probably would not use D850, Nikon 1 would be sufficient..😉
Take care, Robert
Btw those papers I started to study are pretty difficult to read and understand… That‘s life.
Am I the only one here who feels Nikon have cheated me? If their plans all along were to hike the Mpix of their camera bodies, why were, and are, they still selling the lenses crossed out on Nasim’s analysis? As a non professional but still very keen enthusiast, I have spent thousands on the lenses I now own, and to find that Nikon are leaving us behind is frankly quite shocking and disloyal. The lenses should have been re-jigged to offer the resolution that will be required on the higher Mpix bodies now launching. If this is truly an issue, why did Nikon launch the Z7 with an F adaptor at all? Clearly most lenses are inappropriate for use on the Z system if the resolution is simply not of a high enough standard? I do feel very strongly that I have been ripped off, to some extent. These are not life changing sums of money, but, they are pretty substantial when you compare the cost to, say, golf, or fishing or any other equipment oriented hobby. Nikon should be offering trade in deals to those who find themselves with fairly new and expensive lenses which do not cut the mustard on the Z system! As things stand, Nikon appear happy to let their customers get ripped off – not a great business philosophy, long term, is it?
If you’d run a business and a lot of R&D (and sometimes more PR power to gte the stuff out) would be necessary, would you care to take old equipment back and loose money on selling new stuff for reduced prices? Does that feel like a good business model? Who you think has to finance your idea?
You have nice Nikon lenses, cool. You made some pictures you really like and can KEEP on your hardie or have it printed – I don’t know how many golfers can keep their one-in-a-hole for a lifetime and even fish doesn’t get more tastier over time when you keep it after fishing.
How many of your pictures would get significantly better, improve their emotional impact to the viewer with more resolution and sharper lenses?
Do you feel more ripped when a new camera comes out and is sold after few months with a signifcant rebate – or you try to sell it second hand and can hardly get more than 50% of the original price?
They came out with the FTZ adapter because they apparently are not up to the task to pump out a full lens system at the start. Not because they think – what a lot of amateurs do – that the old glass is already good for newer tech. Do you think Canon users feel less ripped? They get loads of great glass but just mediocre bodies / sensors.
Joachim, I think you missed my point completely, maybe it was the way I put it. What I am saying is that, for the past few years in the run up to the Z range, Nikon KNEW. Yet they still kept prices high, and did nothing to improve these lenses, and now, they are doing nothing to support their customers. The second part of my point was that, people invest in a system, and, that is a two way street, or ought to be. If Nikon want to have customers coming back, then, it is pretty important to at least keep them in the loop – which they abjectly failed to do. Don’t misunderstand me, I bought my gear at a fixed point in time, and added some lenses along the way, at other, different, fixed points in time. The last one, just a month before the Z range was announced. I was, and am, happy with the gear, but had I been forewarned of the move to the higher resolution lenses, maybe I would have held on to a couple of thousand and gone on to buy them now, instead. I can’t see why you reject my sentiment about Nikon ripping us off, because, that is exactly what happened, Fortunately for me, I compose my shots carefully, and the need for cropping is minimal at best, leaving me with excellent results. I still can’t help feeling that if I had just held on a few months, I could be enjoying images with potentially much better resoluton, to take forward into the future.
As to Canon, maybe they are doing it right, by having better glass and improving their bodies & sensors – I don’t know, because I am not invested in a Canon system. All I can say is I feel very let down by Nikons lack of loyalty to regular customers. It is not like a one time purchase or product. If they want people to come back to buy more, then their business philosophy is clearly in need of a revamp.
I’m eternally grateful that Nasim took the trouble to produce this information, but unless I change bodies and some lenses, it can’t help me now! It will serve as a guide should I decide to change in the future, but the law of diminishing returns means I am unlikely to put thousands of dollars into this, again, for a slight improvement in resolution, which can only really be seen on huge enlargements anyway.
Well, Autofocuscross, you were suggesting Nikon to give their customers something like a “trade in old stuff, get new stuff”. I was just asking who will finance this kind of sales?
I like Nikon for their bodies. My era of buying Nikon lenses pretty much ended with Sigma’s 35/1.4 Art. I had some troubles to adjust AF on Nikon lenses, I think it was the usual combination of a bit upper tolerance on the Body AF, a bit of lower tolerance of the lens and a lot insecurity of how to AF best from me. After I tried, I exchanged the 24/1.4G, 50/1.4G, 85/1.4G, 14-24/2.8G, 24-85whatever (horrible kit lens), 70-200/4G. For each of that expensive lenses the Sigma version was at least as good as the genuine, but mostly better – and cheaper. So, I felt the rip-off a bit earlier than you – but was happy I didn’t loose too much, as in most cases I could use the money I got for the Nikkor to buy a better Sigma.
And I heard every bloody stupid reasoning against Sigma, unreliable AF, slow AF, not as weather-resistant, not working well with the body. Fact is, there’s no weak lens in their Art line, but some pretty hefty resolvers. If I would continue to get F-glass, I’d replace the 50 and 35 against the 40/1.4 or try to live with an unloved 28/1.4.
But I’m not loosing a single tear about a DSLR anymore. I understood you feel ripped off and I would not recommend you “go Z instead” as I’m under the impression that some solutions on Z-bodies were done with cost reduction in mind – and only that. Also, the lenses are sharp, resolve a lot of details but feel cheaper than they actually are. And each comes with a new filter diameter.
It might as well be that I turn my back to Nikon and go L-mount or Sony, just to enjoy Sigma glass. Also because I’m under the impression, the last firmware update for my Z-bodies was doing something weird to them. They act unreliable to a point I’m not used on any Nikon before. In very few situations they didn’t focus or behaved weird like the play button not reacting, EVF not switching over to LCD. Altogether I’m in the process of let go Nikon. Everybody will have an own opinion but to me it appears as if they don’t listen and don’t learn. They commit only half-heartedly to ML. Bring a 58 Noct which will just sit in the showcases of some collectors. That was the last confirmation to not take their ML attempt seriously, and I will not go back to DSLR.
For those of you who have read all the comments posted so far, here is a link to show the same outcomes for Sony lenses mated to a sony A7r4 body. You will see that many older GM lenses are not resolving as well and even a few surprises such as some G Zoom lenses performing better then the GM zoom lenses.
sonyalpha.blog/2019/…ony-a7riv/
Hi everyone,
I own a Nikon D850 and I have been quite concerned with this article. I did not expect that my 24mm 1.8G was not up to expectations with my camera.
Anyway, I was thinking of selling it and buying another zoom lens that had that range.
Do you plan to make a list of lenses that take advantage of the high resolution other than Nikon? There is not much information on the internet about this and I think it would suit us all very well.
If it’s “only” for MTF: Roger Cigala of Lensrentals is doing tests of batches of lenses (usually 10) and has collected the results in their blog: www.lensrentals.com/blog/…/restests/
First page currently shows some extremely high frequencies like 200lp/mm, but from page 3 you see the “normal” comparisons at stadard frequencies.
And the guys of Digital Picture do have comparison tools: www.the-digital-picture.com/Revie…Crops.aspx which is interesting as they use partly Lensrentals’ batch test results as well.
I am new to this but it seems to me pixel density is the real issue. If that is the case, then I wonder about APS-C cameras with 20+mp sensors since upscaling to full-frame would yield a sensor of around 50mp which is above anything you tested. In other words, it seems like the sensor density of APS-C is even “harder” on lens resolution. Don’t know if this makes sense, but again, I am new to camera technology and its inter-relationships.
I’ve been using Nikon for about 35 years. Today I’m dividing my project work between Fuji (X-series/medium format) and very occasionally, still Nikon DSLR’s (the reason why I kept on Nikon was a better native support of flashes with Nikon CLS but that’s a no issue today) A lot of Nikon glass was disappointing me already a while on higher resolution sensors – not blown away by sharpness in the corners and the CA you have to life (mostly green or purple edges) with in quite a few Nikkors is often beyond something you can accept in a today’s concept. I’ve never worked with the Z but I see very little light in this system as it is a complete ‘rebuy from scratch’ if you want to exploit the native quality of the system. The adaptor-story is purely to blend in the system in an existing F-market. Regarding the use of Sigma Art, I can argument for that too. The Art series are optically good performers but in the real world the 35 and 50mm are extremely heavy lenses, the AF is not as reliable as in the Nikkors and mostly and they are no optical miracles either (I find them a bit over-hyped). With the offer of so many excellent MILC’s and related glass today I’d never persue that particular path again . For me – and see that as a personal, not hard standpoint, Nikon F has had its best time, it was already over the top five years ago and look at the second hands market, a lot of people are selling off their entire F-gear. In favor of… mostly not Nikon (unfortunately), but I may assume Sony. And that’s where it went wrong – if the Z-series should have been just a mirrorless F five years ago, also offering newer and better standard glass without all the known artifacts, I’d likely never had gone away from them because that would have been a true migration path…
I disagree very strongly on the parts of “overhyped Sigma, not better than Nikon glass and more unreliable in AF”, but your conclusion of “too late with their mirrorless system” I only can agree on. Nikon always was on the slow side with development. Some diehard Nikon users say “you need to wait longer but once they come out with it, it’s just great”. I don’t think this to be true anymore – if it ever was.
I also think, a 120-300/2.8 at 4 times the price of the much longer on the market Sigma – good luck in finding buyers, maybe the people ordering the mindboggingly stupid and useless Nikkor Noct still have some dosh to burn.
And even if the 70-200/2.8 S will be one of the few zooms to catch up with high MP sensors: backorderlists will be very long, like with their 500/5.6 PF which is always “not in stock” although the best to say about it is it’s lightweight – f/5.6 is usually not enough. And the 70-200/2.8 S will not improve the fact that there’s still no professional body showing a real commitment. I expect even the D880 or what it’s name will be, will appear before they come up with a fast, reliable and capable pro ML body – while Sony is already offering it’s second gen α9 Nikon still has to improve a lot of stuff.
Hi Nasim,
thanks again for this one and I agree with the doubts about constant rat racing for better figures that can be put in a brochure – like sensor resolution – rather than thoroughly trying to getting the best out of what is already there. In the past I had the chance to work directly for and with the Japanese people of the group that Nikon belongs to and had the privilege to look behing the scenes in PLC development and manufacturing. There the things go in the direction you would expect from a Japanese high tech emterprise: Contantly aiming for the perfect solution and basically forget things like 80:20 rule … or at least make it a 99,1 : 0,9 rule … They simply don’t have word “sample variation” or random sample testing in their dictionary :-) … But that’s beyond what we can change anyway.
However, your article is putting up a question that I can imagine going around in many people’s head working below Pro level and considering themselves more as a enthusiast or serious hobby photograher.
The results of your comparison shows that a higher resolution sensor doesn’t necessarily lead to better results in resolution tests if the lens can’t keep up, so that increasing the sensor resolution does not give you (significantly) better results with this particular lens. Inverting this thought means to me that a particular lens that does not give splendid results with 45 or 36 MP could still be a top choice for 24MP sensor and making it unnecessary to soned lots of money for new glass, before actually getting a new camera with it. It would be very interesting to see a comparison between the max. lens resolution requirements depending on the sensor resolution from a practical perspective based on your existing test results. This would give many people on the mid range and lowerend of the scale – which might even be also on a budget – a good guideline for actually reading your test results and translate them more easily for their current camera.
Hey Nasim,
A fascinating read, thanks for putting in the time and research to create this! I’ve read through the article and the comments and I’m still trying to wrap my head around all of this. By a lens not resolving past 36mp, that pretty much means that it cannot provide the detail to match the high MP count, right? Say you do shoot with a lens on a D850 that can’t resolve to its level, are you just giving up potential detail? Are you harming the images coming out of the D850 by matching it with one of the below-par lenses? I guess I’m trying to fully understand the downsides to shooting these lenses with a D850.
I currently shoot with a D750 and Nikon F100 (with all G/ai-s lenses as well as a Sigma 35/1.4) and have been looking to potentially jump to a D850. I’m trying to discern the impact shooting these lenses would have on that body. Since I shoot a lot of film still and love the F100, it’s harder for me to make the jump to the Z system and have an even more crowded camera bag. Would the D780 make more sense?
Would appreciate your wisdom on the matter!
Have been a frequent visitor this site a good while, typically I agree with, enjoy and learn from most of the articles posted. Have been an avid user of Nikon and photographic other systems for 40+ years, I strive to have good photo technique even if the subject matter and composition may not be the best.
This article does not quite seem right. I have known for a while that many of the lenses I have and use do not resolve as much detail as the D810 sensor can record. Some lenses don’t come close, those lenses that obviously fall well short are sold or relegated to the back shelf. Other lenses I use may not be resolution champs, but render in a useful way that may be different from a clinically “sharp” lens.
Nikon should keep up with (or beat) the competition with both sensor development and lens capability, not stop camera development. They have to keep up (or better, innovate) to remain competitive. Nikon could and is improving the lens lineup, maybe too slowly. I would point out that this site has recommended Nikon lenses to be used on the D8XX series cameras that “now” do not have “enough” resolution.
I think, it’s a shame to use high MP sensors just to record all shades of blurr caused by (too) low-res lenses. Innovation for me would be here to get a sensor which can record high MP (unlike Nasim’s pixelshift idea, as it is useless when motion is involved), but also can switch 4 or 9 pixels together to make one pixel group, able to collect more light. So, having in one body the choice to either shoot at very low light or very high resolution would be something worth to be called “innovation”.
Let’s say the test is done right and a lens “only” resolves 36MP. Then you would need at least 144MP+ to resolve that without any aliasing or moire. So there is a pretty big argument for further increases in resolution.
And AA filters are not the answers because they blur all frequencies meaning they will hurt resolution.
Also, modern high ISO performance doesn’t depend on pixel size anymore. So why not go for the 200MP sensor to resolve everything the lens has to offer without artifacts and then simply scale to a normal resolution? Maybe even do that in-camera so you get a “better” 36MP RAW, so you don’t suffer the storage cost
If a lens measured today with anti-aliased sensors doens’t resolve them, how you think, things would improve by just multiplying 36 MP by 4?
I’ll wait for a 24 MP Foveon-sensor which should be released within the next 2 years. That makes 72 MP equivalent to Bayer patterns. :)
First, the D810 and D850 both have NO low pass filter.
Now if you can resolve the lens with 36MP you will need at least 4x (2x on each axis) the resolution to record it without aliasing according to the nyquist shannon sampling theorem.
This way you get all the signal the lens produces without suffering from anti aliasing or moire. You can then apply a proper low pass filter digitally when downscaling.
24MP + Lens = 82% resolution of the lens +moire
36MP + Lens = 100% resolution of the lens + moire
36MP + AA Filter + Lens = <100% resolution of the lens + no moire
144MP + Lens = 100% resolution of the lens + no moire
A very small pixel pitch is a big problem with digital sensors, especially bigger ones.
The photosites function best with light coming to them from a straight 90 degree angle.
Remember how Olympus advertised the advantage of the four third sensor as lenses more easy could be made telecentric?
With steep ray angles not all of the light hit the intended photosite but spill over to the photosite beyond, in the direction of the ray angle. The mandatory inclusion of filter stack and cover glass further distort the path of the light ray.
The consequence is lower resolution and contrast in areas outside the centre of the sensor.
Despite the resolution increase that 144MP sensor could with many lenses have lower resolution at the edges and especially in the corners than a 36MP sensor.
Adding to the above smaller photosites produce more noise. So it will mostly be useful with low ISO and on a tripod.
Nasim, I have the deepest respect for you and your work. It is always top notch and your findings are well-founded and really helpful. This article is very interesting and as such, no exception.
This sentence caught my attention:
“The exotic super telephotos obviously shine, but surprisingly, some of the older lenses like the Nikon 400mm f/2.8G VR, or the Nikon 600mm f/4G VR were already not good enough for 45 MP sensors.”
This is a surprise, indeed and I have been curious as to why they never made Nikon’s list. We all “know” the 400mm f/2.8 G VR and 600mm f/4G VR to be very sharp, right? I venture to say that Nikon’s marketing departement may have something to do with this. The previous 300mm f/2.8G VR is optically identical (including the coatings) to the current 300mm f/2.8 VR II that made the list. Thus, based on optical merits alone, the older 300mm f/2.8G VR should have made the list as well. I know it went out of production in 2009 and it begs the question of how far back Nikon should look? One generation is what they did for some of the lenses, so why not for all of them? Adding the previous 300mm f/2.8G VR, 400mm f/2.8G VR and 600mm f/4G VR to Nikon’s list might hurt new sales as there are used lenses in circulation that are equal, or not far behind, the current models (that are otherwise improved). If that is the case, I would think the marketing department has been involved in that decision.
Should the 400mm f/2.8G VR and 600mm f/4G VR be on the list? I believe so and I base that conclusion on test data from DxOMark and LenScore where both of them did better than the 500mm f/4G VR (and much better than the 300mm f/4E PF VR LenScore also tested, which made the list).
DxOMark’s test concluded that the 400mm f/2.8G VR resolved more detail than the 400mm f/2.8E FL VR (33 vs 32 on a D800E). (If the D800E would reveal more detail with the 400G than with the 400E FL, would it not be reasonable to believe that it would also hold true for the D850?) Thus, it seems like the 400G is not that much “weaker” (DxOMark has it ahead of the E FL and LenScore the other way around, 1290 for the 400G and 1466 for the 400E FL) for it not to make the list. The 600mm f/4G VR, that also did not make Nikon’s list, fared better in Dx0Mark’s lab (24 vs 22 on D800E) and achieved higher resolving power test results when tested by LenScore (1319 and 1212 respectively) than the 500mm f/4G VR G that made the list.
This lead me to assume that some of the older lenses never made the D850 list for other reasons that their optical merits alone. Perhaps this is also true for the AF-S 300mm f/4D as well?
I think the simple answer is that Nikon, with some exception only included current lenses in stock when the list was published. The exception is the old AF-S 70-200 2.8 VR. As that type of lens is the bread-and-butter lens for many pro photographers an many probably still use the first version it may have been included to inform these photographers that their lens would be fine if they needed to replace their current camera wit a D850.
The AF-S 70-200 2.8 VRII was probably still in stock and therefore also included.
As for the super tele lenses these are made in small batches and shipped quite soon to markets around the world. No big stock are kept. So when Nikon introduce a new version they usually have no stock left of the previous version. The one exception here seem to be the AF-S 500 4 G.
So yes, with the exception of the first AF-S 70-200 2.8 VR Nikon only included lenses they had in stock and wanted to sell. Therefore you are correct there should be more lenses making the list if also older lenses had been included.
A remark is that already the first version AF-S 300 2.8 practically uses the same optical scheme as the later VR lenses. Nikon simply added VR to one lens group. That said later lenses can still be slightly better with slight tweaking of elements.
I also read comments that the first AF-S 400 2.8 is just as good or even slightly better than later AF-S 400 2.8 VR versions, but clearly worse at the sides. But if one anyway mostly will use the lens in crop mode that matters less.
The old manual Ai-P 500 4 is also said to be very good, comparable or maybe even slightly better than later AF-S 500 4 G lenses. But if it would make the list I don’t know.
Thanks for the great article as usual, Nasim!
We all get what we paid for, if it is the case indeed, I just need to make a slow and painless transition to ML to get higher quality image, hopefully by then the super tele line up is ready for the Z system.
A better title would be: “Why consumers should not buy Nikon camera more than 36 mp…
JANUARY 29, 2020 AT 10:59 AM
i have a D810 and many of the lenses on your list that meet or exceed 36mp performance. I am only confused on one choice. I have the 24mm f1.8G and purchased this because it out performs the 20mm f1.8G based on many review sites including yours. Center sharpness is about the same for either, but the 24 gives better mid frame and corners. So why is the 20 recommended over the 24 on your list above?
Steve
Nasim,
I really got a lot out of this article. You make some very important and practical points.
I do have a question: thinking in a reverse manner, are there lenses with resolving power that lower MP sensors just can’t realistically take advantage of? To put it another way, taking the 24 MP sensor of the D750 or Z6 as a basis for discussion, are there Nikon or other lenses whose resolving power is so great that the sensor simply isn’t up to the task? If “yes”, then Wouldn’t this also suggest that there is a point past where chasing resolution is not a practical expense? The resulting recommendation being don’t purchase lens XYZ for use on a D750 or Z6 because the sensor can not take full advantage of the lens.
Appreciate your comments,
Dave
The simple answer is no. The ideal is having lenses with greater resolving power than the sensor. Only then can the sensor be used to its full potential.
Besides, resolution is only one of many lens characteristics. And usually one buy a lens based on specification, not resolution.
Good points EnPassant. What I was wondering, though, is there a point at which the price of improved lens resolution becomes impractical if it is to be used on the 24 MP sensors of the D750, D780, and Z6? Perhaps I am missing the point, but intuitively it seems as if one could equally chase lens resolution to an unnecessary extreme insofar as their use on 24 MP sensors go.
Is it worth it? That is the eternal question. Since it involves personal opinion, needs and how much is available in the bank account. I also do not know a formula or program that could give an answer. But anyone writing such a program that actually worked could become rich!
For an amateur, who seldom print and look at that print from a distance and mostly watch photos on a screen it is usually not worth spending a lot just to get a lens with better resolution using a 24MP sensor as most modern lenses are good enough, especially at their best apertures.
Differences in resolution can then mostly only be seen with pixel peeping.
Pixel peepers may of course disagree.
If I would make an exemple, although I don’t have a Z camera, I read that the 50/1.8 S is a fantastic lens compared to Nikon’s all previos 50mm lenses. So in this case I would say it is worth it buing the S lens for a Z6.
But, provided the owner think it is okay using the FTZ adapter, it may not be worth it buying the S 35mm and 85mm f1.8 leses if the owner already have Nikon’s or Tamron’s F-lenses with same specification, or is thinking about moving to the Z system just for the lenses.
Another exemple is when I bought a car a couple of years ago.
I was considering between a combi/estate wagon/station car (or whatever name is used in different parts of the world!) in slightly large middle or big size.
Equipment was very similar with plus and minus for each.
Practical size difference was 20cm in length, both exterior and interior.
Prize difference was around 6-7k €.
In the end I came to the conclusion the medium car was big enough for my needs. Only in some odd occations, that may never occur, could the extra length be useful. So it was mostly about more space and comfort in the backseat, a lesser consideration since so far I only had passengers in the front passanger seat! Also a bigger car comes with several downsides because of the size and weight. But for some people it is still worth it.
Great that the 50mm f/1.2 Ai-S, that lens has something special about it. Def one of my all time favourites and go-to for night time street photography. Good work as usual, thanks for the article.
And then there’s noise.
Gordon Laing’s YouTube review of the A7R IV with its bonkers 61 mpix shows that over 400 ISO the resolution advantage over the 42 mpix R III is lost in noise.
First of all thanks Nasim for publishing value based facts. One question regarding the performance of 24 mm focal length. The Nikon 24 1.8 is equal or slightly better at the D850 compared to the D810 regarding sharpness. The 24-70 VR is worse than the 24 1.8 when reading your tests of These lenses. At the D850 however the 24-70 VR seems to outperform the 24 1.8 (@ 24 mm and w.r.t. sharpness).Can you Share the MTF data that supports that ? I’m still struggling with a Zoom beeing sharper than a Prime ….. Best Regards and Thanks
These tests confirm what I have seen at lensrentals blog (before they moved to optical bench testing) and opticallimits who also tested the same lens on low and high MP sensors.
While some lenses actually can make use of the increased resolution wide angle lenses have problems. Most of the resolution increase is coming in the centre of the picture according to the test. But for the areas outside the improvement is marginal. In the corners at wide apertures not only the 24 1.8 G but also the acclaimed AF-S 28 1.4 E is worse on the higher MP sensor!
How can that be? I think it is the smaller pixel pitch causing trouble with steep ray angles.
This is also why I would have liked the D780 to have 36MP. Because I had already come to the conclusion that is the best compromise resolution for a full frame camera (used for photography), at least with modern AF-S lenses. I hardly have a need for more resolution, but sometimes would like to crop a bit without losing to much resolution. I also much prefer the smaller size and weight of D750. Although D780 grown a bit it is still the same weight. But as it looks now the only 36 MP option seem to be a pre D850 camera. So one has to look into the used market. At least it will be cheaper. It looks possible to buy both a D800 and a D810 camera for the same price as one used D850 camera. And without rebates for a new D780 one can buy three used D800 cameras for same or lower price!
While there here are some talk about problems with the lens testing program not matching real world results there have always been a conflict between lab tests usually made of a target at a defined distance compared to real world photography that can be made at all distances from closest focus to infinity.
Hi Nasim,
Sorry if this is a naive question for the level of discussion taking place here, but could you please give me some counsel related to this topic? I have a three-part question, and I’m hoping your answers gives me some practical advice as well as clarifying my understanding of this whole issue. I’m really just looking for
the simplest possible answers here.
First, if using a “middling” quality F mount lens (such as a 24-120 f/4, or 85 f/1.8) on a Z7, would there be any benefit to image quality in reducing the camera’s sensor resolution down from Large (45mp) to Medium (25mp)?
Secondly, would that same F mount lens produce a superior image, in any way, if placed on a
D750 set to its top resolution of 24 mp? Assume I’m shooting the same scene.
Finally, does the answer to either question depend upon the available light of the scene (low light vs. more light)?
I enjoy this website, and appreciate the amount of work you must be putting into it. Thanks.
Slightly off topic…
Stay with the native Z lenses.
I have the Z7. And, I can definitively state there is a difference with the native Z lenses and 3rd party lenses. Sometimes subtle sometimes major.
At the moment, I have a Sigma 135mm 1.8, Nikkor Z 24mm 1.8s and Nikkor Z 50mm 1.8s. All good—more in a moment.
But I just recently had a Tamron 24-70mm 2.8 G2 and a Sigma 14-24mm 2.8.
The native lenses completely knock it out of the park! The difference makes me feel good about the purchase of the Z7. (I really danced around between a Sony or older Nikon large body—so glad I didn’t make that mistake.)
Oh and I realize it’s been beaten to death, but the resolution and sharpness on the Z 50mm 1.8 is the best I have seen on ANY lens in my life with only one exception: Fuji medium format 45mm. The Nikkors are astounding! (I’ll eventually fork over the $$$ for the Z 70-200mm and 24-70mm). Maybe. I’m starting to think how great a prime 20mm or prime 135mm Z would do on my Z7!
My 2 cents.
Dear Nasim, the head of this article reads:
“Why Nikon Should Not Make DSLRs Over 36 MP”
However, the expressed assumption is misleading.
First; there is no lens that fails to resolve all the pixels from a D850 that yields worse results on a D850 than a camera with less resolution. If a lens fails to resolve all the detail, it is because it lacks the capability of high resolution. A superzoom or a so-so zoom like the 24-120mm 4.0 will not yield the best possible results on a camera with lower resolution either.
The issue is therefore not the high resolution of the camera, but the failure of some lenses to resolve sufficient detail in order to fully utilize the high-MP cameras like the D850. Nikon does have many DSLR lenses that demonstrably can resolve sufficient detail to do so, and there are other manufacturers that also offer lenses with high resolving power.
The point is that for the best possible technical results, the lenses have to be chosen carefully. That would apply to all cameras.
Just out of curiosity, how many MP did you find the Ai-S to resolve?
I tried and could not find your list of recommended lenses for D800E. Do you have one? For some reason, that will be my top digital body.
Or maybe your or other readers know about such lists from others, please share. I know Nikon’s list and Thom.
Thanks!
Marcel
1. You only need one lens to benefit from the additional resolution to make it worth it to some. Anyone with the 500 mm FP lens will want a sensor to take advantage of it. I’m planning to get that lens, but wouldn’t bother if the pixel density were not up to it.
2. The pixel density of 36 MP on a FF sensor corresponds to 16 MP on a DX, 5 MP on an RX100, and less than 1 MP on iphones and many video cameras.
There’s no question these other sensors benefit from higher pixel density, so why not a FF? The DX and FX even share many lenses. Of course these other cameras only need the resolving power over a smaller area, but still a FF camera would benefit for cropped photos.
3. I’m skeptical of your results because in my comparisons of cheapo lenses (Nikkor 70-300 AF-P, Sigma 150-600) with a D850, I can easily see the individual pixels degrading resolution (in favorable situations). This is true even with the kit (18-50) lens of a Z50 (which has similar pixel density).
The diffraction limit at F/4 doesn’t limit resolution until about 1 micron, which would correspond to 1 GP on a FF camera. And the effective resolution is not determined by the worst contribution when they have comparable magnitudes. They will all contribute.
4. This is anecdotal, but switching from a 24 MP D610 to the D850 made a very noticeable difference in sharpness of my pictures. (Even if I correct for rationalization due to money spent:)
5. Even if the higher pixel density does not benefit sharpness, more pixels gives higher dynamic range: 4 pixels effectively adds 2 bits of dynamic range compared to one pixel 4 times bigger.
To clarify the comparison between the D610 and D850, what I mean by noticeably better sharpness is really more cropping latitude. In fact, when I had the D610, I kept a DX camera (D7000) for additional reach, because it has a higher pixel density. Using a DX with the same lens is just cropping the same image, but if the pixel density is higher, you can crop even more before the pixels are visible.
The pixel density of the D850 is similar to that of the D7000, so I sold the D7000.