NikonGear'23
Gear Talk => Lens Talk => Topic started by: chambeshi on January 22, 2020, 07:50:21
-
This article should fan some debate! As it evaluates resolving performance of F-Nikkors on the "Official List" Recommended for the D810 and D850. Basically only 5 out of 13 zooms tested by Nasim Mansurov (Photography Life) 'crack the nod' for resolving above 36mp ie when used on a D850 and Z7. One of the veterans is the 14-24mm f/2.8G.
Quite a few G primes don't make the grade. But to my mind, the premium on high resolution depends across different genre(s) and allied uses of the images. For www based shots of travel etc, a 24-120 f4G does very well. And for quite a few of us, there are more than a few classic Nikkors including the 'Cream Machine' aka 85 f1.4 AFD have unique roles on a D850. And I still find the couple of AIS lenses still perform well on the D850: Namely 28 f2.8AIS and 55 f2.8AIS Micro-Nikkor.
https://photographylife.com/nikon-dslr-resolution
"...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."
"...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…" [italics added]
-
Confirms what many have observed, that many of the old lenses does have very good image quality on the D850 that's very nice know!
-
The D850 is more forgiving for lenses than the D800E and D810.
I wonder why the 50mm G 1.4 and 1.8 are missing in the list.? (And the 17-35/2.8 is ignored)
-
In the 'test' they look at resolving power, not the rendering of the lenses, quite different things ;)
I remember comparing the 14-24mm f/2.8 AFS to my 17-35mm f/2.8 AFS on D3 and D3X - Resolution especially in the outer region was far better on the new zoom, but the central part and the overall rendering was very nice and similar for the two lenses if used around f/5.6
50mm f/1.4 AF-D and 80-200mm f/2.8 AFS are also not on the list but for sure high resolving lenses with amazing nice rendering on D850
105mm f/2 DC AF-D the resolution score is not up to 36 MP according to the test, but IMHO very sharp while retaining nice smooth tone transitions and of course Bokeh on D850
The list is nice as a baseline if and when you need sharp, high resolving power, but that's it ;) All the other characteristics you need to look elsewhere to find ;)
-
In the 'test' they look at resolving power, not the rendering of the lenses, quite different things ;)
I remember comparing the 14-24mm f/2.8 AFS to my 17-35mm f/2.8 AFS on D3 and D3X - Resolution especially in the outer region was far better on the new zoom, but the central part and the overall rendering was very nice and similar for the two lenses if used around f/5.6
50mm f/1.4 AF-D and 80-200mm f/2.8 AFS are also not on the list but for sure high resolving lenses with amazing nice rendering on D850
105mm f/2 DC AF-D the resolution score is not up to 36 MP according to the test, but IMHO very sharp while retaining nice smooth tone transitions and of course Bokeh on D850
The list is nice as a baseline if and when you need sharp, high resolving power, but that's it ;) All the other characteristics you need to look elsewhere to find ;)
Tendency is to rate Rendering as the poor/distant relation to Resolution. More pragmatic to set one's own criteria for contrast and sharpness. And here we have many choices, even for high-resolution FX cameras (ie > 36mp)
-
I think there is an underlying (and very common) misconception (or approach*, if you will) about resolution, MTF, etc. For any lens, using a higher-res sensor will improve the overall quality, if perhaps only slightly. There seems to be a common thinking that you want/need very high MTF at Nyquist frequency... Besides triggering aliasing, this pushes for very big lenses (the lens has to have much higher resolution than the sensor) that possibly compromise other aspects of rendering. Very high-res sensors reveal the 'true character' of a lens and might show more pronounced differences e.g. center vs borders. But on low-res sensors we would never see the center resolution on this level. We might not like the differences between center and edge, but the next generation of sensors will also reveal the same thing about today's best lenses that now (viewed with today's sensors) seem to have very uniform sharpness across the frame. In any case, this dynamic keeps lens manufacturers busy. Back when the D800 came out, I was told by the owner of a photography store that Nikon doesn't have any lenses available that can be used with the camera. I had to laugh -- but he was not the only one to spread this message. ::)
For me, I currently don't want to deal with files bigger than 24mp... I have run into aliasing with some recent very sharp lenses. This would go away if I used a denser sensor - but I would have to bear the cost of large files for the other 90% of images that do not need the additional pixels. I also enjoy some old lenses e.g. Nikon Ai-s with which these problems do not appear.. but with every new generation of cameras, all lenses seem to get slightly better. IMHO it has never gone backwards if you look at the entire image. So I tend to hang on to my lenses.
I agree about the potential of Z mount to give us very good optics without the massive size/weight, borne out by some examples already, such as 50/1.8S vs. Zeiss Otus 55/1.4 (as some tests seem to indicate, I don't have first-hand experience with the Otus).
-
(https://live.staticflickr.com/65535/48064362433_44f42b7f2c_k.jpg) (https://flic.kr/p/2gehqq2)2019-06-14 059-1 (https://flic.kr/p/2gehqq2) by longz[url=https://flic.kr/p/2gehpWM](https://live.staticflickr.com/65535/48064360853_0ae3b50aec_k.jpg) (https://www.flickr.com/photos/longzoom/)2019-06-14 059-1-2 (https://flic.kr/p/2gehpWM) by longzoom (https://www.flickr.com/photos/longzoom/), on Flickroom[/url], on Flickr. The Z7, 80-400G, wide open, crop of 100%. LZ
-
(https://live.staticflickr.com/65535/47969378126_f208a7fd18_k.jpg) (https://flic.kr/p/2g5TATy)2019-05-30_29-1 (https://flic.kr/p/2g5TATy) by longzoom[/url[url=https://flic.kr/p/2g5Tht3](https://live.staticflickr.com/65535/47969316152_96229bf75e_h.jpg) (https://www.flickr.com/photos/longzoom/)2019-05-30_29-1-3 (https://flic.kr/p/2g5Tht3) by longzoom (https://www.flickr.com/photos/longzoom/), on Flickr], on Flickr. The Z7, 80-400G, wide open, crop. LZ
-
(https://live.staticflickr.com/65535/46893869394_2b2d33a8f0_k.jpg) (https://flic.kr/p/2erRkBy)2019-04-07_7-1 (https://flic.kr/p/2erRkBy) by longzoom (https://www.flickr.com/photos/longzoom/), on Flic(https://live.staticflickr.com/65535/40651560463_53465be37a_h.jpg) (https://flic.kr/p/24WeSDv)2019-04-07_7-1-2 (https://flic.kr/p/24WeSDv) by longzoom (https://www.flickr.com/photos/longzoom/), on Flickrkr The Z7, 80-400G, wide open at 400mm, crop from the extreme corner of 200%. So needless to trust some/every test, ( i think, personaly), because you will test your OWN COPY of the lens. Try it, make your own conclusion. Good luck! LZ
-
Having read many such articles over the past 20 years, I more and more believe that the scientific evidence is mostly supported by revenue. Trying to make susceptible buyers buy lenses they do not need.
My better half has been using a Nikon D800 for about 5 years now. She uses mostly the Sigma Art primes. 35mm f1,4 and 135mm f1,8. Both lenses are tested with high resolving power.
I just bought a D800 two days ago, and will mostly use lenses like the Nikon 55mm f1,2, 105mm f2,5 and 35mm f2 Nikkor-O. I decided to buy a 36 Mp camera after using the lenses on Kristinas D800 and seeing how they perform there.
From those personal experiences, the least sharp lenses are the 55mm f1,2 (wide open) and the 300mm f2,8 Ai-S. All other lenses do not only appear sharper, but, like the 4 element 135mm f2,8 Nikkor-Q.C actually are perfectly sharp when zoomed to 1:1, which is closer examination then most viewers will perform on the resulting images.
Optical constraints is another term for "We want your money". So far I have not seen evidence of lenses for the Nikon F mount not being perfectly sharp on cameras with an F mount. Maybe I'm just lucky though, and buy all the good samples out there
-
Having read many such articles over the past 20 years, I more and more believe that the scientific evidence is mostly supported by revenue. Trying to make susceptible buyers buy lenses they do not need.
Agree...these are businesses, and need to sell stuff.
If I want to make digital shots on my D800/E that are reminiscent of that which is achievable with an 8x10 view camera, I will need a better lens than the average old manual focus Nikkor (28/2.8 AiS and 55/2.8 AiS excepted)...and I'll need a tripod and careful manual focusing.
I shoot mostly handheld, which reduces my D800/E to the equivalent of roughly a 12 or 16 megapixel camera due to camera shake. I'm fine with that. My next cam will be 24 MP, max.
-
Optical constraints is another term for "We want your money". So far I have not seen evidence of lenses for the Nikon F mount not being perfectly sharp on cameras with an F mount. Maybe I'm just lucky though, and buy all the good samples out there
It's indeed in many cases not the optical constraint. Different mp's just ask for a different technique. Which is also valid for relative high mp's (24) on a DX camera.
-
Agree...these are businesses, and need to sell stuff.
If I want to make digital shots on my D800/E that are reminiscent of that which is achievable with an 8x10 view camera, I will need a better lens than the average old manual focus Nikkor (28/2.8 AiS and 55/2.8 AiS excepted)...and I'll need a tripod and careful manual focusing.
I shoot mostly handheld, which reduces my D800/E to the equivalent of roughly a 12 or 16 megapixel camera due to camera shake. I'm fine with that. My next cam will be 24 MP, max.
There are actually quite a few more exceptions, including the 16mm f3,5 Fish Eye and the 135mm f3,5, which is a 4 element lens with exceptional sharpness, but a few other flaws. In the necxt few days I will try out the selection of lenses I have to see how they hold up, but I expect that I won't see a lack of resolving power in most of these lenses at 100% pixel peeper view.
-
There are actually quite a few more exceptions, including the 16mm f3,5 Fish Eye and the 135mm f3,5, which is a 4 element lens with exceptional sharpness, but a few other flaws. In the necxt few days I will try out the selection of lenses I have to see how they hold up, but I expect that I won't see a lack of resolving power in most of these lenses at 100% pixel peeper view.
Yes, yes...I didn't mean to imply that the 28/2.8 AiS and 55/2.8 were the only high-performing older Nikkors.
I will try to make comparison between my 135/3.5 AiS and the Zeiss ZF 100/2 when an appropriate subject presents itself.
-
I don't think the article is saying these lenses are rubbish, or that you need to replace your lenses.
What it's saying is that there is no point is buying a higher-resolution camera if those are the lenses that will be paired with it.
So in fact, he is saying that if you are happy with your current lenses, and they are not on the list, then you don't need to buy the higher resolution camera.
-
I don't think the article is saying these lenses are rubbish, or that you need to replace your lenses.
What it's saying is that there is no point is buying a higher-resolution camera if those are the lenses that will be paired with it.
So in fact, he is saying that if you are happy with your current lenses, and they are not on the list, then you don't need to buy the higher resolution camera.
Some of my lenses that are not on the list, will cover the next 60+ MP sensors of the next generation. LZ
-
The article seems to ignore the fact that the final system resolution is the convolution of lens resolution and sensor resolution. In other words, you get more resolution in the end product by either using a higher resolving lens, or a higher resolving sensor, or for highest gain, both.
There is no such thing as sensor outresolving a lens or vice versa.
-
--
There is no such thing as sensor outresolving a lens or vice versa.
If that were true, we would never be bothered by moiré or aliasing artefacts.
-
I don't think the article is saying these lenses are rubbish, or that you need to replace your lenses.
What it's saying is that there is no point is buying a higher-resolution camera if those are the lenses that will be paired with it.
So in fact, he is saying that if you are happy with your current lenses, and they are not on the list, then you don't need to buy the higher resolution camera.
You have three sentences all saying three different things. I don't agree with parts.
Sentence 1:
Yes that is what he says,
Yes it is true
Sentence 2:
Yes that is what he says,
No - a higher resolution camera can benefit older lenses. This is what the solitaire attests to, and I can confirm
Sentence 3:
I'm not sure if this is what he is saying, but that doesn't matter much.
I believe that buying a higher resolution camera makes a difference with lenses on and off his list.
-
If that were true, we would never be bothered by moiré or aliasing artefacts.
I'm no expert on optics, but moiré is the lens outresolving the sensor, and the sensor making the best of it? Or did I get that wrong. Because I remember moiré being a serious issue with the D100 and D70 back in the day
-
Moiré is one of the expressions of "spurious" resolution due to insufficient spatial resolution of the sensor (compared to what the lens delivers of transferred contrast towards the Nyquist limit). In that sense, the lens can definitively "outresolve" the sensor.
A sensor "outresolving" the lens would yield increasing amounts of mush, as the lens cannot render detail with sufficient transferred contrast at high spatial frequencies.
-
Most modern lenses will be able to handle high MP just fine, provided your willingness to stop down.
I'm sure many have the 50mm/1.4 AIS lens. Close than down to F5.6, the sharpness and detail is amazing on my D810. Wide open, that's another story, not so great that is.
There are many cases where a lens out-resolves the sensor or vice versa. The science that goes into this stuff is almost esoteric, it's very messy and confusing, especially since cameras have their own algorithms to handle the photos, and on top of that, programs to read and edit will have a say in the final result as well.
Monochromatic sensors yield superior sharpness and resolution for any lens compared to colour sensors, all lenses designed for the VIS spectrum will produce the best MTF under green light.
Debayering must be accounted as well.
In microscopy, it's important to match the camera to the microscope setup's resolution. ie we don't want the camera out-resolving the microscope setup (softness, mush), or vice versa (moire!). However when it comes to non-monochromatic sensors, stuff such as debayering and the camera baking raw files (not a true raw) must be accounted for as well. An objective might be calculated to match a 24MP fullframe sensor, but we need to account for tolerances, so I would suggest a 36MP instead. That's also why I haven't upgraded my D810, 36MP is all I need, 45MP is too much. My microscopy setup can't handle that, as it's matched to around 22MP on FX.
Read more here: https://www.microscopyu.com/tutorials/matching-camera-to-microscope-resolution
Too add to the mess, there's also the case of resolution drop-off. There's many lenses that are very high resolution, however they suck when coupled with an FX sensor since the corners are just horrible. An example would be the 5x NA0.28 HR objective lens I have. It has an NA equal to my 10x objective, it stomps all the 5x lenses I have in terms of sharpness and resolution, this however only applies to the centre, perhaps just 10% of it. Anything outside of the centre is basically unusable. However I can also push this lens to 10x, making it almost as good as my native 10x. This is why lenses designed for smaller sensors have higher resolution, they need to resolve smaller pixels, a more complicated design is required. Don't be surprised at those massive M4/3 or Fujifilm DX lenses, they need that size.
Or tl;dr, don't worry about this stuff, modern lenses are good enough and will handle that 45 of 60MP FX sensor just fine. Knowing this information won't make better looking photos. Framing is more important than the lenses' ability to capture lines printed on glass.
-
With how few people really have any need for >~12mp (myself included the vast majority of the time), I find the resolution concerns people have to be generally amusing. I can't wait to hear the cries for more MP and sharper lenses once the triple digits have been attained.
"I viewed a portrait at 100% and the surface texture of my subject's eye boogers lacked acuity, unacceptable!"
-
With how few people really have any need for >~12mp (myself included the vast majority of the time), I find the resolution concerns people have to be generally amusing. I can't wait to hear the cries for more MP and sharper lenses once the triple digits have been attained.
"I viewed a portrait at 100% and the surface texture of my subject's eye boogers lacked acuity, unacceptable!"
Don't show it to your subject, (assuming) she will want to smack you after seeing the skin. ;D
-
With how few people really have any need for >~12mp (myself included the vast majority of the time), I find the resolution concerns people have to be generally amusing. I can't wait to hear the cries for more MP and sharper lenses once the triple digits have been attained.
"I viewed a portrait at 100% and the surface texture of my subject's eye boogers lacked acuity, unacceptable!"
After seeing results from a 12 Mp D3 and a 36 Mp D800 back to back for a few years, I do see one advantage in having those extra pixels, eve if it is mainly to produce 12 Mp files from those 36 recorded Mp.
High ISO noise. Some might say, the D3s is a lowlight beast, made to be used at high ISO, but the extra pixels of the D800 allow for even higher ISO speeds to be used. When the files are reduced to a lower resolution, most of the noise is reduced to a level where it is no big issue.
I found myself using my D3 at no more then 1100 ISO and started losing detail as soon as I used higher ISO speeds. The noise produced wasn't really bothering me, but the loss of clarity really was an issue for me. With the D800 I now go up to 2000 ISO and still feel that there is plenty of clarity in small details, after reducing file size a bit.
In that sense, the files I get from the D800 are far better then those from the D3, even if I resize both files to 12 Mp.
The D3 had better white balance and color accuracy however. Files from the D3 hardly needed editing in that area, where the D800 files need quite some tweaking to get the colors right
-
All this talk of higher resolution, more Mp's and whatnot, not only amusing but in my view a lot of BS too.
1/. you don't buy a new camera with more megapixels(eg. D800/810 -> D850) in isolation!
It's bought as a system in itself. The D850 will have a better(cleaner) ISO, which turned out to be the case. It'll have better AF, It'll have better buffer clearing times, it'll have faster frame rates .. etc.
Maybe some folks buy the higher resolution camera for the fact that it has more pixels, but the reality turns out that it has better 'everything else' factor that comes with it.
So the photographylife article is pretty much a load of hot air, not really realising the reality of buying a so called "higher resolution camera"
Articles like these are usually best ignored so that the creator doesn't attract any more attention than they deserve!
Then you get articles from other sources that complain when Nikon(or any other manufacturer) .. DON'T!! .. make next gen products with more pixels!
To my mind the issue is not the number of pixels(more or less), products, manufacturers. engineering, gear(ie. lenses or otherwise) that users have .. it's simply that such articles should just be ignored so that the authors don't get any more airtime.
I've gone from 6 to 12, to 24 to 36Mp, and have never seen any real issue with any sub par, or inadequate lenses on higher and higher res camera models.
It's not that the image quality is actually reduced in any way! it may not be a quantum leap of extra pop or wow factor if that was the expectation, but the notion that Nikon should stop making >36Mp sensored cameras is stupid!
May sound harsh, but it's the reality of it.
Nikon should make whatever they feel is necessary to sell as many of that product as they think they can. If that means a 100Mp Fx sensor D850 replacement, this is Nikon's call to make, not another internet based armchair expert!
This then becomes my option(should I choose) to have one or not.
Basically, Mansurov should just stop chasing a dream he could never hope to achieve .. and that is for Nikon to make for his own pleasure a product based on his expectations!
ie. if he doesn't want more than 36Mp sensor product, why upgrade to a 45Mp one!
He needs to realise that his choices are for him to make and not for Nikon to enforce upon him.
:P
Anyhow, I'm looking forward to the 100Mp D850 update to go with my Hanimex 70-600 f/4.5-5.6 consumer lens from 20 years ago! ;D
-
High ISO noise. Some might say, the D3s is a lowlight beast, made to be used at high ISO, but the extra pixels of the D800 allow for even higher ISO speeds to be used. When the files are reduced to a lower resolution, most of the noise is reduced to a level where it is no big issue.
It's not only the lower noise, but also the better colour (Color fidelity) and luminance at high ISO where the D3s excels. Despite the downsizing advantage of the D800/850 the D3s clearly wins *(should it be a contest). With some minor color chrominance corrections ISO 12800 gives clean results.
-
So I actually read that article...
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.
Do I need to elaborate on how wrong and stupid this statement fundamentally is?
Yeah, I won't.
Then he makes a list of lenses that Nikon recommends for their D850, crossing out ones that can't handle 45MP as he claims to counter his previously made statement. Many lenses need to be stopped down which substantiates to my statement of "provided one's willingness to stop down".
His test provides us with 2 conclusions:
- Higher MP yields more resolution for any lens
- Newer lenses are able to yield better MTF numbers on higher resolution cameras
I honestly did not expect the 14-24 to do that well, it's almost a decade old, it was designed well ahead of its time.
His conclusions are pretty baffling. "No major improvement on 45MP, therefore the lens can't resolve 45MP". That's just fallacious at best.
He never compared the result to lenses that aren't on the list, so it's not even a controlled test and thus his conclusion is absolutely useless.
I won't call a score of 3000+ in the centre on the 85/1.8G "unable to resolve 45MP".
The 28/1.8E is interesting. 4162 in the centre by F2.8 but 2214 mid-frame, whereas the 24/1.4G gets 2356.
While it's true that some lenses are resolution bound, I question the claim of "not being able to resolve 45MP". If anything, Nikon should make more modern F-mount lenses optimised for high MP, not stop making high MP bodies. There's plenty lenses on the market that handles 45MP just fine, such as the printing-nikkor/Nikon Rayfact lenses, Schneider's Xenon-Sapphire lenses and the Macro-Varon... the list goes on, the author acknowledges it with third party manufactures such as Sigma and Tamron in his conclusion. There's plenty people (like myself) who use the camera as merely a digital back, Nikon's F-mount lenses doesn't matter too much to me, it's merely convenience when it comes to ordinary shooting. What I need is a good ecosystem, clean images (ISO 64!), and ergonomics, all of which Nikon offers on their D850 DSLR, in fact it's the only one that ticks all boxes.
Nikon might not enhance their F-mount F1.8 and F1.4 lineup until the Z-mount is fully established. However if they wish to put out more DSLRs with high MP, they probably should, unless the DSLR market shrinks or something. I'm still waiting for the 24 and 45 PC-E to be modernised, which I guess will never come.
-
45MP is too much. My microscopy setup can't handle that, as it's matched to around 22MP on FX.
Isn't this statement somehow similar to this one:
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.
?
I wonder what happens if you use your above-mentioned microscopy setup with a 45MP sensor.
-
Isn't this statement somehow similar to this one:
?
I wonder what happens if you use your above-mentioned microscopy setup with a 45MP sensor.
Let's see… so my lenses are able to resolve about 22MP. Having debayering and other in-camera baking in the equation, 36MP is about enough.
Let's say my lens can resolve one pixel on a 22MP fullframe body. Now the same pixel is split into about 4, ie 1/4 of a pixel, and there's 4.
This results in essentially "digital enlargement", otherwise called "mush". So I'm wasting megapixels and there's no improvement. This analogy is overly simplistic, but it epitomises the need to match camera resolution to the lenses.
The two fields can't be compared. Nikon can just make new lenses optimised for higher MP. Objectives manufactures can, but they won't. An objective lens is intended to be used on a microscope.
Adequate working distance is required for lighting, therefore the only way to get higher NA while retaining acceptable working distance would be to make the objectives bigger, which doesn't work as they won't fit on a microscope turret anymore. One can shrink the working distance to get higher NA, this however would make lighting difficult and yield a thinner depth of field, essentially making it harder to use. "Closing down" isn't an option, the relay lens should always be wide open but it can't be too fast or else spherical aberration will kick in, along with other weirdness. The equivalent aperture after the pairing is well into diffraction territory already.
This is obviously not the case for consumer lenses, Nikon etc can just make bigger bulkier lenses with more sophisticated optical formulas. Market defines everything, there's no industrial or scientific need for massive microscope lenses optimised for large sensors and high MP.
That's why the statement stands when it comes to applications that are truly limited by either physics, consumers, or both. For commercial photography, neither is the case, therefore his statement is fundamentally wrong. Considering that most never go beyond 1:1 when it comes to macro, it's safe to say the author never considered this case nor does he care.
Moving up to a d850 will be incredibly beneficial for me, however not on the MP side. The d850 has a fully electronic shutter, the d810 only has EFSC. At 10x and beyond, shutter shake becomes an issue. The d850's tilt screen is a welcomed addition, as well as far better liveview capabilities, as I don't pair my system with a computer. I own the 70-200 F2.8E which will benefit from the far better AF engine, and my Nikon Rayfact and Xenon-sapphire lenses will pair well with 45MP.
That said, I'd much rather have a d8x0 with that newer 36MP 16-bit Sony sensor.
Hope it makes sense.
-
I found substituting the Z7 for D810 made a significant improvement for my macro work (typically 5-10X). Most of the improvrements probsbly are due to much lower amounts of camera shake, not so much the extra 9 MPix.
Empty magnifiction (leading to 'mush') is to be avoided in any critical system. Hence the need to match (pre-, post-) magnification, optics, and camera resolution to make an optimised system for a given application.
-
1) When I bought the D800, it did not make my lenses obsolete.
2) Once stopped down, good old MF lenses usually are able to make full use of the 36MP resolution, except maybe in some corners that rarely matter. The 105/2.5 is not the only one. I think a "blind test" challenge could settle the discussion here, better than statements by bloggers.
3) My first experience with a Z7 was in a store in Tokyo, and I asked if I could try out the lens I had with me: Noct Nikkor, AIS. The result was not for me to say "gosh, my dear oldie is obsolete", but "wow, more beautiful than ever" (the Z7 sensor is *really* good, not just hi-res). Later, I repeated the experiment with an old Summicron-R 50/2 (2nd generation). And, in both cases, I did not see lots of mush in places that really matter (close to the center or first third).
If I were to buy a Nikon Z, I would really hesitate between Z6 and Z7 because of the overall IQ of the latter (rather than resolution), given the strong impression that the Z7 made with my oldies. But I'd keep my lenses ! Zeiss, Leitz Canada, Voigtländer, and Nikon AI/AIS.
Of course, when shooting big flat detailed subjects like famous paintings in a museum or antique mosaics, expectations can be very different from mine.
-
It really depends on how one defines "acceptable". The corners of my stupidly expensive objective lenses aren't acceptable to me, so I only use the middle 4 segments out of 16, which is 25% of my 336MP D810 sensor. The rest is discarded as disgusting mush. Some might feel more comfortable, some won't.
To exemplify, here's a butterfly wing:
(https://live.staticflickr.com/65535/49507079046_cfba717bc2_b.jpg)
Full resolution: https://live.staticflickr.com/65535/49507079046_ae7e4d2445_o.jpg (downsampled by about 50%)
Centre is beautiful, corners put me off big time, and that's 50%, imagine how bad 100% is. There's weird comatic issues, softness, and smearing. This is a $950 objective lens. There are better options, but I'm not made out of money. Also the better options won't have good corners either, arguably worse actually, the high NA mitutoyo HR lenses have a smaller circle of sharpness. Some will say the corners look totally fine.
Such subject wouldn't matter too much when it comes to bad corners, however some are unforgiving, such as silicon wafers/chips:
(https://live.staticflickr.com/65535/49559724711_f23d017ab0_b.jpg)
https://live.staticflickr.com/65535/49559724711_8a0fa1f32e_o.jpg
You can see edge-to-edge sharpness with no visible drops, this is because I made a panorama of 10 shots where only 2 is truly necessary to showcase the entire chip. I can tolerate soft corners when it comes to butterfly wings; however for these chips, I want the absolute best from my camera. Can you imagine not being able to see what's in the corners? It defeats the purpose of such reproductions, I cannot and will not accept that.
I would have bought the Z50, but Nikon being Nikon decided the mirrorless D500 as their marketing claims doesn't need a shutter release input. I am baffled and have no idea what went through the engineers' mind. The Z50 doesn't tick the most important feature that I always assume is on a $1000+ camera body. >:( The cute little Nikon 1 J5 I bought for $150 AUD used doesn't have an input either, it fits in my daily travel pouch. Puts out really nice images.
Would be rather simple to DIY, but I'm not spending that time to get a feature that should be on a $1000+ AUD camera to begin with. Either they make a Z5 that's the true mirrorless counterpart of the D500, or I'll stick to my Dinosaur810.
-
Thanks for the examples and elaborations. I can follow all that. If you have a lens which produces mush in the corners there is no benefit in increasing resolution. What I struggle to understand is this notion (implied or not) of somehow being worse off with a higher res sensor. Maybe there is an unstated assumption about storage space and processing. And of course if people upgrade their camera to a higher res model they can be disappointed if they were hoping to be able to do certain things. But if higher-res sensors can be made at the same price point and as long as storage capacities continue to more or less evolve according to Moore's law, there is nothing much to worry about, even if there were no new lenses from now on.
Nice images of the butterfly wing and chip BTW!
-
The butterfly's wing is perfectly sharp at the upper right and lower left corners, with some small loss of the resolution, as usual. The rest is simply out of DOF, but, a very small detail is sharp, in the extreme lower right corner, inside tolerance of DOF. It looks like a very good lens, for me, at least. LZ
-
The butterfly's wing is perfectly sharp at the upper right and lower left corners, with some small loss of the resolution, as usual. The rest is simply out of DOF, but, a very small detail is sharp, in the extreme lower right corner, inside tolerance of DOF. It looks like a very good lens, for me, at least. LZ
I think you're right, I've been quite conservative with the number of steps, my D810 is reaching 200k actuations. Maybe I should just nuke it and buy a D850/Z7, who knows?
That comatic aberration is especially weird, it's probably Oof mush that the program is trying to make sense of, thereby taking specular reflections as actual detail.
(https://live.staticflickr.com/65535/49199242373_e95458e789_b.jpg)
Here's one where the edges are sharp-ish as well. I somehow overlooked newby mistakes. I'll make another stack just for the sake of it.
Not saying the lens is bad, it's $1700 for what essentially is the size of a film canister. The corners just aren't "good".
You can see an example here: https://www.photomacrography.net/forum/viewtopic.php?t=41112
These are from totally manual setups with a resolution of 0.5 microns, so there's no way I missed the best depth of focus.
Thanks for the examples and elaborations. I can follow all that. If you have a lens which produces mush in the corners there is no benefit in increasing resolution. What I struggle to understand is this notion (implied or not) of somehow being worse off with a higher res sensor. Maybe there is an unstated assumption about storage space and processing. And of course if people upgrade their camera to a higher res model they can be disappointed if they were hoping to be able to do certain things. But if higher-res sensors can be made at the same price point and as long as storage capacities continue to more or less evolve according to Moore's law, there is nothing much to worry about, even if there were no new lenses from now on.
Nice images of the butterfly wing and chip BTW!
If we control for everything, then there isn't a real benefit. If not, obviously better ergonomics etc is the benefit.
If I have a lens that can only resolve a 4x4 unit pixel, then splitting this 4x4 into 4 segments of 2x2 pixels doesn't do anything. It's digital enlargement and on a pixel level, it's going to look less sharp.
High MP cameras (D850) will always cost more than their low MP (D780) counterparts with chronology being accounted for. So the question is, would that extra $1000+ that one is expected to pay worthwhile? When resolution is bottle-necked by the lens, no. For future proofing and potential newly released lenses to at least match the resolution? Perhaps, depends on how one values money, because next several years there's going to be a new toy that will make the current one cheaper.
You're not worse off, it's just pointless as the improvement will be marginal at best. What the author of the article fails to acknowledge likely due to the mirrorless bandwagon is "new F-mount lenses can always be made". Hell, we have several dozen existing ones that pair absolutely fine already! So >36MP is not pointless and certainly not "worse off", that's unless we're talking about the circle I'm in. A Mit 10x Apo is $900 new (works well with the D850), the HR version with an NA of 0.42 is $8000+, it couples well with a sensor that has 3x3um pixels (Sony a7R4 has a pitch of about 3.7um).
However at 20x, all this is turned upside down. The 20x with an NA of 0.42 resolves about 6.1x6.1um pixels, the D810 works pretty well with 4.87x4.87um, how about higher? Not so much. One isn't worse off, there's just close to zero improvements image-wise. If we go to Mit's 50x NA0.55, then it's 9x9um, the "just don't zoom in" category. It can be said that one is better off cropping into the 20x 0.42 and keeping that at least $800 for a used copy that might not work at all. Unless there's a commercial and scientific need for big fat objectives with adequate WA and high NA, all at actually acceptable prices, the D810 is pretty much topped out when it comes to image quality.
All this changes with the introduction of pixel shift though, which is capable to circumvent this resolution matching fiasco and pop out beautiful detail, even on lenses that can't resolve its pixels. So at the end of the day, higher MP is still justified regardless of the author's rationalism or cherrypicked charts. Oh, downsampling is always an option too, I downsample my 36MP shots into 18MP ones. Goodbye mushy corners.