Are Your Lenses Suddenly Obsolete?

Up until a few years ago, if you purchased a quality lens you could be sure that with proper care it would continue to perform well even as you upgraded your body in the future. After all, bodies decay and glass lasts. However, with the sudden influx of high-resolution cameras and the seeming resurgence of the megapixel war, some are asking: “Can lenses keep up?”

When the Canon 5DS and 5DS R were released, Canon also released a list of recommended lenses “for getting the best from” the two camera bodies. This caused a minor uproar in the photography community, particularly since some notable Canon lenses weren’t on the list. The implication was, of course, that the lenses could not resolve 50.6 megapixels within the 35mm full-frame format. The camera had apparently outgrown the lens. This was potentially huge: as technology progresses, bodies are rendered obsolete and the cost of upgrading is a generally accepted part of the profession. However, a good lens is just that: a good lens. Never before had the proposition been introduced that along with the body, our lenses must be upgraded as well. Some argued that indeed, the ultra-high resolutions were simply too dense for some lenses. Others argued that it was a marketing ploy to encourage photographers to invest in even more high-end glass.

One Statistic to Rule Them All

Back in 2012, DxoMark introduced the perceptual megapixel, or P-Mpix for short. Meant to obviate the need for the MTF chart, it distilled the sharpness of a camera/lens combination down to one number: the number of megapixels you were effectively getting from the camera and lens. It essentially measures the maximum resolving capability of a camera/lens combination. For example, my 5D Mark III has a 22.3-megapixel sensor, but when paired with the EF 100mm f/2.8L Macro, the combination has an 18 P-Mpix rating, meaning somewhere in the system, I'm losing 4 megapixels to imperfections. DxO claims this measurement “bypasses the problems inherent to MTF,” namely readability, use of understandable units (which I would argue falls under the former category), and a canonical correspondence with human vision.

As an aside, I take anything that comes from DxOMark with a healthy grain of salt. Here’s the problem: they refuse to publish detailed methodology. There’s a reason science is an open community: science is hard… like, really hard. That’s why we publish, critique, evaluate, question, test, test, test, retest, and replicate results before they are accepted as fact. The “because I said so” method simply would not fly and it’s why I highly suspect I see sparse mention of DxOMark when I search peer-reviewed academic journals, despite the scientific importance and industrial reach of the study of modern optics. I own and use both the 5D Mark III and the EF 100mm f/2.8L Macro, and honestly I have a hard time believing I’m “losing” 4 megapixels with that combination. Alas, without some hard equations and procedures to look at, I can’t tell you much more. I’m not saying DxOMark is wrong; I’m saying that without more information, we can’t decide if they’re right or wrong and unfortunately, that undermines their relevance in my humble opinion. Talking about “resolution” in photography is a deceivingly complex proposition and if I don’t even know the working definition we’re using and how it was arrived at, I can’t hope to have a fruitful conversation.

For example, consider this skeletal outline of how a medicinal study might operate: construct a hypothesis, devise a method that fairly tests the hypothesis, conduct the experiment with many, many repetitions to detect outliers and unforeseen phenomena, analyze the data using appropriately chosen statistical methods that both illuminate patterns and minimize bias, publicly publish both the methodology and the data in a peer-reviewed context, receive feedback, repeat and refine dozens of times, and finally, if your hypothesis has been verified many times by multiple, independent resources, it is promoted from hypothesis to fact. This is how science works. If I'm presented with data with a scientific air to it, I expect to see this sort of stringency in its genesis.

18 MP or 22 MP? I'm perfectly happy with my 5D Mark III and EF 100 f/2.8L Macro either way.

The Camera or the Lens?

Anyway, back to lenses. The question of what out-resolves what is a complicated one; it depends on pixel pitch (which in turn creates a transitive dependence on sensor size), control of various aberrations, the limits of human perception, the ever-present and increasingly important (as megapixel counts skyrocket) phenomenon of diffraction, and various other factors. Suffice to say that smaller, sub-full-frame sensors may be at or past the line with some combinations. That is to say, put a lower quality EF-S lens on a Canon 7D Mark II and you may well be getting less megapixels than you paid for. This is because pixels tend to be packed more densely on smaller sensors (and thus, their poorer low light performance), meaning that a lens must be able to resolve finer detail to avoid essentially smearing information across the pixels.

On the other hand, full-frame users should be thinking more about diffraction limits and control of aberrations. Part of why full-frame glass is so much more expensive than its crop counterparts is not only because there is physically more glass, but controlling aberrations becomes more difficult as the image circle grows larger. As I discussed in my article on the 5DS and diffraction, as sensor resolution goes up, the aperture at which diffraction effectively becomes a legitimate concern drops (becomes wider). The 5DS pushed the threshold of this into legitimate working photographer apertures. As we see megapixel counts continue to rise, photographers will have to consider how this affects them, particularly those who work at smaller apertures and require high levels of detail (landscape and macro photographers, take note).

Practical Considerations

There’s also an added complication independent of the physics: we don’t normally see lower level glass paired with higher level bodies. This in itself has normally kept a certain level of separation, but now, as megapixel counts approach unforeseen levels, that gap is being closed. And so, we have to ask: if you buy a ultra-high-resolution camera, will even your good glass be rendered obsolete? Well, I can’t answer that for two reasons: first, it depends on your definition of “obsolete.” If you have a 50 megapixel sensor, are you willing to take a hit of 5 megapixels before you call a lens archaic? How about 10 megapixels? That’s up to you. Second, at this point in time, we don’t have data obtained from a publicly available, verified method to make that call. My suspicions say that indeed, the threshold of degradation may be starting to cross into lower level professional lens territory, but they're just that: suspicions.

Nevertheless, I leave it up to you now. Do you own a Canon 5DS, a Sony a7R II, or a Nikon D810, and have you noticed this phenomenon? What sort of hit in resolution would you be willing to take? Or is this all marketing to push newer lenses? Or do you want to see more transparency of technical data in photography before you decide? Let us know your thoughts in the comments.

Log in or register to post comments


Lauchlan Toal's picture

At some point, will lenses even being capable of making the most of a sensor? A lens can only be so perfect, especially the relatively small lenses used for DSLRs. Medium and large format cameras can get away with huge amounts of resolution, but with correspondingly huge image circles in their lenses. Physically, I'm not entirely sure that a lens can be made that resolves 90% of a, say, 600MP sensor. At the very least, design and manufacturing would have to be radically changed. Perhaps even a migration to synthetic diamond elements to allow for wider apertures than glass (due to the high refractive index) that would help combat diffraction limitations.

Tim Foster's picture

Obsolete to who? If you didn't need the resolution before, you don't need it now just because the sensors are better.

Anonymous's picture

If only Canon would explain with detail what's "getting the best from the EOS 5DS and EOS 5DS R" I think we could confabulate some more, aside from that everything is just vague speculation. I think maybe in the future these ultra high megapixel cameras and lenses will be like having a whole new system, like when you move from crop to full frame or mirrorless kind of thing.

Anonymous's picture

I agree. Canon's list doesn't make a whole lot of sense. There are some very obvious omissions from the list and a couple of odd inclusions. Also, from all of the tests I've seen, more pixels improve any lens, it's just the matter of how much they are improved.

Sergio Tello's picture

Betteridge's law of headlines.

Zachary Will's picture

It's also sort of an odd way to measure lens. For video that would mean that any lens that resolves more than 3mp would look the same in a 1080p video which is almost every lens.

Felix Wu's picture

Canon certainly has its own agenda with their release of 5ds/R cameras...lenses! That's where most people invest most of their money in a camera system. As we know a lot of the legendary primes/zooms have been around for years and haven't been updated, so what's the best way to obsolete those lenses and let photographers of this world to upgrade all their lenses? Let's release some high MP bodies! Then 5ds/R was developed. Canon didn't push for dynamic range performance, but one thing they did pretty good was the improvement on sharpness and detail ( the sheer effect from high MP body). So with such high resolving power the public instantly recognise how much more they could capture with this new line of cameras, and after that they need lenses! As people slowly upgrading their legendary lenses Canon is living wealthily thereafter...

In short like someone mentioned above, if you didn't need high MP before you don't need it now. The need for high MP depends on print size and screen resolution. If neither of those is required then you can stick to your current system and make nice pictures.

Francisco Eduardo de Camargo's picture

My dear Alex Cooke ... Show images on a computer tele does not mean much, right? The night all cats are gray and the computer screen everything is beautiful, but ... you would be interesting to show the image on a printing miaor size ... I'm talking nonsense?

Ralph Hightower's picture

Physics doesn't change, but technology creates change by lens coatings and aiding in lens design. Measurements are the best way to quantify lenses. DxO , since they don't reveal their measurements, relies on Voodoo technology.
Take a camera I shoot with from the 1980's, Canon F-1N; what it produces is dependent upon the lens and the film.

David Sklenář's picture

Great article, enjoyed it, but honestly I have to say I don't care at all :-) I have been used to photograph with an old 550D + old nifty-fifty and I was fine with it. Now I shot with 6D + 35 Art + new 50 1.8 stm and the details are still awesome, I would say I have more detail then I have ever needed :-) I shoot fashion, engagement and lifestyle, I use high ISOs and I have never felt bad about it... :-)

Ricardo Jakulica's picture
Prefers Film's picture

Honestly, I couldn't care less about the latest high resolution sensor. For my website, the widest image is 550 pixels. And what I post to Instagram is not even twice that. I still shoot raw, but my library is really just taking up space until I need to make that rare print. All my lenses are either Canon L series, or Canon primes, with the exception of one Sigma. They won't be obsolete any time soon.

Lis Beattie's picture

It's not always the megapixel count that makes the difference. Nikon's D3200 and D7100, despite being vastly different in quality, have the same megapixel count (24). However, the 18-55mm kit lens from the D3200, although acceptably sharp on that camera, is soft on the D7100 even at it's optimum aperture. The difference is that the D7100 has no low-pass filter. You're getting better resolution out of the same number of pixels, and the lens isn't quite up to it. Nikon's new 18-55mm kit lens takes this into account, and is apparently acceptably sharp when paired with cameras lacking low-pass filters.

Jeff Gelzinis's picture

I find this discussion really interesting, but I have a hard time wrapping my head around the concept of "Losing MP". I would love to see, as the author alludes to, some image examples of what a 5 or 10 MP "loss" looks like compared to a "full resolution" image, on any lens/sensor combination. Does this currently exist anywhere? Is DxO mark publishing these sorts of images along with their P-Mpix information?

Undrell Maholmes's picture

Ha Ha! I bet film is starting to look real good to all of the "all digital" shooters huh? I feel that it's just another marketing ploy.

John Skinner's picture

This topic always gets me laughing !

We've come from a world of pushing 400 ASA films to 1600 and up.. grain as large as salt. Now?

People are whimpering about just how much of a gnat's butt hole we can see from 1 mile away.. Really?

We have capabilities of making billboard size ads from regular FX size sensors, and people keep on about effective mega-pixels versus body mega-pixels.. The world has completely gone off their collective rockers. Someone needs to 'stop the madness' and start to focus more of the craft... less on the hype generated by companies like Canon who seem to want to change a knob, and call it a 'new model'.

This constant comparing and upgrading, then on to some dolt's opinion on sharpness, focus speeds, glass coatings. It's enough to drive a person to drink. If your one of these guys always looking through specs and over your shoulder at what;s coming up? You could be one of these idiots.

The vast majority of people are just amateurs or pro-ams doing the odd job or birthday party/wedding. The vast majority of people post to web only.. Isn't it time we just get over all of this?

It's no wonder DXO has all the traffic they have... People have lost their marbles.

Tony Northrup's picture

We've done controlled comparison tests with all these high resolution cameras, and our finding is simple: more megapixels gives you more detail with whatever modern lens you're using, including your kit lens. A jump from 24 megapixels to 36 megapixels doesn't give you 50% more detail unless you're using an optically perfect lens. Typically, a 50% increase in megapixels gives you about 25% more detail (given good technique).

And, FWIW, DxOMark has a lot of really detailed information about their methodology: We've definitely found some errors in their data, so I agree that their data isn't perfect, but they describe pretty precisely how they measure.

Alex Cooke's picture

I agree, Tony, both in regards to the relative and the absolute gains. Certainly, resolution will never get worse when using the same lens on a higher resolution sensor; I think the issue in question is more if there's an asymptote that we're approaching.

I've perused the DxOMark protocol pretty closely, but unfortunately, they still gloss over a lot of crucial details and most importantly, neglect to explicitly state the equations and algorithms in use. I would be thrilled if they would release those!

Nigel Banks's picture

I remember back in the days, an article that may put perspective on this issue. If a film can resolve 115 lpm and used with a nifty 50 at 80lpm, then we have lost something already. Now the film is developed and goes into the enlarger whose lens resolves 50lpm onto paper that can resolve 11lpm, which is all good except our eyes on average only resolve 5 lpm. Do you get the point. This is where perception comes in, we may (??) be able to perceive greater levels of apparent sharpness due to contrast, but not by much. To come to the point, sharpness doesn’t matter as much as many people think except when making humungous prints, but then we have to take into account viewing distance, a wall mural needs to be viewed at 20ft. to appear sharp, at 12” it most definitely isn’t.. Instead of these pointless discussions for the sake of technology, we need to work on taking pictures and our skills at doing so.

Derrald Farnsworth-Livingston's picture

For the most part I've seen old 8 MP sensors be able to be enlarged quite well. A lower quality lens on a high megapixel body will still be preferable to a lower megapixel body in any case.

In any event I'm pleased to own the 5ds R, it's a heck of a camera and I welcome the extra megapixels for everything I do, even though I am using lenses I've owned for almost a decade on it.

The 17 TS-e on it though, wow, that's a combo!

Steve Cantellow's picture

Agree re: DxO Mark. No methods and clear potential for bias given that they want to sell you software and now, a camera. The issue you touch on is more far reaching than lenses and bodies and software - the way we make images could follow any number of paths in the next decade as the pace of change itself is... accelerating. We should keep our kits relatively light as we are bound to have to change them!

paul aparycki's picture

Good article.

Sad thing is most people approach this from the wrong perspective.

Yes nikons d810, the new canon, and sonys newest offering all have extraordinary capability with their sensors, and YES, the problem is the lenses. The vast majority of glass out there are designs that in some cases pre-date digital photography (and it's unique problems).

In all fairness, one cannot expect Canon, Nikon etc to redesign every lense in their arsenals each time there is a leap in sensor attributes. Consequently, there will always be some glass that is less than desirable, regardless, many are still very picture worthy. Both Canon and Nikon have some shining stars that are not a good match for what is happening today, still the lenses are great and many in high demand.

There has been in the medium format sector an effort to truly address this problem . . . Rodenstock produced an entirely new line of lenses expressly designed for MF. Their image circle is quite small so there is not much movement available on a tech camera, but the results speak for themselves on an MF back. Only major drawback? Yup, $$$$$$$$$$$$.

Rodenstock produced an excellent paper addressing this . . . it was available on their website . . . should still be there, or perhaps buryed in one of their brochures. If you can't find it there, check out Michael Reissmanns excellent blog, "luminous landscape". There is an article from about 7 years ago regarding the outfitting of a Linhof 679, along with a back, and subsequent lens selection . . . AND WHY . . . goes into detail. An interesting and educating read.