An Update and Comparison of the Sony FE 70-200mm f/2.8 GM OSS

About a month ago, Roger posted his MTF bench results for the Nikon 70-200mm f/2.8E FL ED AF-S VR (TL;DR: it’s optically superb), and down in the comments there were some requests to compare it not only to the Canon 70-200mm f/2.8L IS II in the article, but also to the Sony FE 70-200mm f/2.8 GM OSS. There were MTF charts linked:

These are not very good looking results, especially for a $2600 lens, but Roger wasn’t sure if they were correct results. In his comments after these charts he said:

Let’s kind of keep this here for right now – I haven’t published it because I’m still a little uncertain about the results. Sony has suggested that a change in cover glass thickness might improve the results some. Not dramatically, but some, off-axis. This lens also has to focusing motors and we have to focus it electronically via a camera to test. I’m not absolutely certain that ‘setting it at infinity focus on the camera’ and ‘manually focusing on a an object at infinity’ are absolutely the same. So take these lab results with a grain of salt. On the other hand, they do seem to agree with what we see in real world results.

I try hard not to put out anything until I’m just absolutely certain our results are correct. We’re doing some stuff here that’s pretty cutting edge, honestly. No one does 4 rotation MTFs, for example. I’m pretty certain these are good results, but not absolutely certain. So I’ll post them in this discussion on my own site but I’d rather not see them reproduced all over the internet yet.

Someone suggested we try all three 70-200s on their respective camera bodies and shoot the same detailed scene with each, then share the files. But that would involve three different cameras and wouldn’t really be an apples to apples comparison, would it? Roger suggested he have one of our photo techs shoot all three lenses on a Sony camera with adapters, admitting that adapters add another variable, but there would be some good practical implications.

So I did just that. I took all three lenses, the Canon 70-200mm f/2.8L IS II, the Nikon 70-200mm f/2.8E FL ED AF-S VR, and the Sony FE 70-200mm f/2.8 GM OSS, and I shot our test chart with all of them on the Sony a7R II. For Canon, I used a Metabones T Smart Adapter IV, and for Nikon, I used a Novoflex Nikon to Sony E adapter. I manually focused all three lenses to get the most consistent results, and here’s what I got:

Full Resolution Examples are Available Here

If you view the charts at 100%, you’ll see that they’re consistent with the MTF charts Roger generated for all three lenses. The Sony just isn’t that great, and that’s really disappointing considering the price tag and how long customers have had to wait for that lens to be available. But if you need f/2.8 and working autofocus, it’s really the best option out there for Sony mirrorless cameras.

Author: Joey Miller

I’m Joey. I love cameras, especially old film cameras, and I can’t remember the last day I didn’t take a photo. Digital cameras are great, and they keep me employed, but I also still like processing my own film. I’m stuck somewhere in the middle. I shoot every single day, no matter what.

Posted in Equipment
  • Magnar W. Fjørtoft

    Less than 50%? Are you sure? Your own MTF measurements? In which respect? All aspects of image quality? Please explain! Sample photos are great! Or else I might think you are just trolling.

  • David, our protocol is fairly simple, but has been very effective. We confirm ‘lens focused at an infinity object reads infinity focus on camera’. We then mount the lens to the bench and using an attached camera set focus to infinity leaving the camera on to maintain that focus while we test. That has, to date, worked fine. It’s very reproducible in repeated tests of single copies, in batches, and every way we have looked at it. That includes running field curvatures, which are both reproducible and reasonable.

    I have a tiny concern that the algorithms for this lens might actually be something like “Focus to infinity, then adjust compensating element to sharpest focus”. I have no reason, other than the MTF results for this lens failing to meet expectations, to think that. But I can’t be certain there’s not that, or something else I haven’t thought of, going on.

    The following is my speculation only: I think the early group of lenses not meeting expectations with most being slightly decentered, followed by a lack of supply of them, might indicate there was a manufacturing problem that is being corrected. That’s the reason I haven’t spent more time on this until we get newer copies and repeat our tests with those. I asked Joey to look at this for me, hoping it would clearly prove the optical bench results were wrong or right. But it didn’t prove either, at least to my mind.

  • Brandon Dube

    …yes? They receive power from a camera just the same whether they are used on a camera (as for the charts here) or on the MTF bench. The 70-200 GM is not the first lens to feature two focusing groups – many true cinema lenses have two, at times even three focusing groups. The Leica SL 24-90 has two focusing groups as well, and its performance is stellar both on camera and on the MTF bench. I do not think there is a real problem here, only that Sony is unable to build their 70-200 as well as Nikon can build their newest one. I don’t think the MTF indicates the lens is very soft at all at 200mm it is comparable to e.g. the Nikon 105 micro, Canon 200/2.8 II prime, or even Nikon’s new 24-70 at 70mm. None of these have a reputation of being particularly soft.

    Sony’s bayonet construction may play a role; I know a lot of plastic is used around the mount of at least the older cameras, though the A7rII may be updated. I think IBIS is also likely very bad for the alignment of the sensor, just as OIS is bad for the alignment of lenses. These factors in conjunction with the higher resolution of the A7rII and the invited comparison to the new best-in-class lens from Nikon make what is probably a pretty good lens, even if not a great one, get a worse rap than it deserves.

  • Carleton Foxx

    So how does one go about checking such a thing?

  • David Kilpatrick

    DSLRs are affected by sensor to flange alignment, AF module to flange alignment, reflex mirror alignment, secondary AF mirror alignment, focusing screen alignment and centering, further alignments inside each lens, and also in some cases sensor carriage alignment including pitch and yaw adjustment and f adjustment… guess why half the industry really loves the mirrorless revolution!

  • David Kilpatrick

    The three-legged stool principle might seem sensible – zero rock – but the mount connections are not exactly robust. In practice I’ve found four-screw flanges do often show some tilt across one of the two axes created. Six-screw flanges might have another axis of tilt but in practice have fewer problems. Sony’s original E-mount body flanges were very low quality with coarse CNC machining, poor flatness and inconsistent thickness. I’ve not examined Fuji X as closely, or dismantled them. Would you say the Fuji engineering is better?

  • David Kilpatrick

    That’s reassuring but since Sony lenses don’t have a hard infinity position, and can not be focused unless powered, I remain doubtful (not about your test procedures, but about the ability of the Sony lens to work correctly and to position groups in an absolute relationship to the focal plane, with a relative relationship to each other).

  • Brandon Dube

    Four screws is mechanically overconstrained, six is only more so.

  • Brandon Dube

    The charts are shot only on a camera. The camera and lens will behave for them exactly as they will any other scene. If there is a problem with the focusing groups on the chart, it is a firmware or mechanical issue and you should not see improvement in day-to-day images.

    On the MTF bench, the mount for the lens is electronically connected to a sony camera (in this case, an A7rII). All other aspects of the MTF bench procedure are identical to any other lens; the addition is that Sony lenses are powered, and talking to the camera. As far as they are concerned they are looking at a star (at infinity); nothing more, nothing less.

  • Carleton Foxx

    Mother of heaven! How do I test for sensor flange/lens mount register and alignment? Does this affect DSLRs as well? Or is it only a mirrorless problem?

  • Carleton Foxx

    Joey, if you really love old film cameras and film, here’s a cool idea…. Put your head together with the other great minds at LR and figure out a way to do the definitive test of film vs. digital.
    But the criteria has to go beyond the boring line pairs per whatever. You need to create criteria that measure the magic of film—the gentle handling of highlights, the mysterious ability to do soft-sharp, the velvety nature of film shadows, and the thousand other ways that film is better than digital.
    Keep it simple by pitting your favorite film camera and lens combo (perhaps medium or large format or even just a Nikon F) against a contemporary camera….
    Who benefits? The LensRental bottom line. Visitors to the site would skyrocket which, of course, translates to higher rental rates which means more money for the owners. Everyone wins.
    Plus it would be cool, and there’s literally no group of people on Planet Earth more qualified to perform such a daring and important experiment.

  • Munchma Quchi

    I tested the f4 version of the Sony against the Nikon and the Sony typically scored less than 50% of the Nikon. Hit rates and speed were also issues. The Sony was worthless. These were all Imatest results.

  • Ben Nieves

    These results lead to the question: should Sony users rely on the 70-200 F4 by Sony instead?

  • Munchma Quchi

    Roger, – where does one typically find the best versions? Thinking there are two schools – 1. initially upon release where QC is high or 2. later when production issues are worked out. How about a series testing lenses across production dates?

  • Munchma Quchi

    Every Sony I ever bought/tested and returned had some degree of sensor flange alignment that revealed itself in Imatests. On the whole about 1 in 6 was a keeper.

  • Munchma Quchi

    I love these tests where ‘these leaves look sharper than those leaves’. They always greatly influence my buying decisions especially when large amounts of money are involved. Numbers schmumbers…..

  • Albert

    I looked at the images at 100% and it looks like some decentering problem. The top right corners of the Nikon and Sony are nearly identical (close enough that no one would be complaining about the Sony), but the top left is night and day, with the Sony terribly soft, and the Nikon consistently sharp. There is no reason for the lens to be sharp in only one corner, and very soft in the other.

  • Well, it’s an ok lens, assuming the price is right. IMO the right price for this lens should be well under $2k.

  • Zé De Boni

    Wow! When I followed the link to read this article I expected the highest level of information based on my previous experience with LensRentals. The article itself is terrible, but there are very good comments by optical science experts (like David Kilpatrick) or literates. Now I am trapped in a discussion with what looks more like a circle of goblins and gnomes believers. I will try a last explanation, sorry if I may disappoint some of you.
    – The primary function of an adapter is to provide the right dstance from the lens back flange to the image plane, according to the the original design of the adapted lens, so that the lenses work exactly as projected.
    – To make an adapter or an extension tube is the easiest task for the industry, based on centuries long mechanical Engineering expertise.To talk about “residual tilt” or “a flange not square” sounds like chasing ghosts! Much more could be reputed to lens centering, but that doesn’t mean flange position and it is related to that of individual lens elements (for which adapters cannot give any good or bad contribution (except those with additional lens elements, like the turbo adapters).
    – Yes, the shorter flange distance allows new optical designs, for more compact lenses or other theoretical improvements (like on the Sony RX1 design). However there is nothing that restricts the design or use of lenses on which the last element sits far away from the sensor. There would be a problem of corner shading due to the tight margin of the lens mount diameter, but in practical terms I do not get this problem when using a legacy preset 800mm f/8 lens (on which the last element is about 5 inches far from the lens flange) on my A7RII.
    – So far, despite all possible compromises, mirrorless cameras may use the same sensors as DSLR’s. Or just the opposite, as Sony Alpha 99II is using the same A7RII CMOS (actually, full sensor+shutter+stabilization assembly).
    – Good adapters, like the ones that must be chosen by serious lens testers, have light baffling similar to those in SLR mirror boxes. In many cases their walls are more distant from the light path than the average internal wall of an SLR. The idea of “diffraction due to an extra lens flange” may arise from someone that never used such adapters, nor had the chance to inspect one closely. One more mythological entity created from the lack of knowledge.
    – There are so many adapters on the market, not only for Sony but for all mirrorless cameras, from M43 to medium format! Are all those manufacturers stupid or crooks? Just for Sony APSC I found 141 models in B&H. So let’s name some of the brands that are in this so said “suspicious” game:
    = Leica has the “Leica S-Adapter L for SL Camera”, “allowing you to mount a Leica S-system lens (DSLR MF) to an SL (Typ 601) mirrorless digital camera body” (full frame).
    = Hasselblad has the “Hasselblad XH Lens Adapter ” so that you can “use any of your current H system lenses (DSLR MF) on a Hasselblad X system camera”, which is, by the way, mirrorless MF.
    = Sony has its own line for mounting A series lenses. Sigma designs its own adapter to use Canon or Nikon mount Sigma lenses on Sony E mount cameras. More significant for those that place Canon as the supreme deity in camera and lens manufactures, this brand has now a mirrorless APSC camera system for which you may presently choose from just 6 native lenses… or you may get “the Canon EF-M Lens Adapter Kit for Canon EF / EF-S Lenses”, which “enables you to mount (any of the 82, as counted on B&H site) Canon EF and EF-S lenses onto the Canon EOS-M mirrorless digital camera”.
    To finish this chapter, below is a list of adapters that I have used on the last five years and the lenses/cameras combination that they allowed me, with satisfactory results (no visible flaws, just the limitations of each lens).
    – Mirex Tilt and Shift Mamiya 645 to A mount: 500mm f/4.5 (sometimes with a Mamiya 2x converter) on A77; 35mm f/3.5 on A900 and A99; 80mm f/4 Macro, 120mm f/4 Macro and 150mm f/2.8 on A99.
    – Sony LAEA-1 (A to E mount): Sony A 50mm f/1.4 and Sony ZA 16-35mm f/2.8 on NEX7, Rokkinon 8mm f/3.5 Fisheye on A6000.
    Sony LAEA-3 (A to FE mount): Minolta HS 300mm f/2.8 to A6000 and A7RII, Sigma Art 35mm f/1.4, Sony ZA 24mm f/2 on A7RII
    – Novoflex Nikon to E mount: Nikon 10.5mm f/2.8 Fisheye, Nikon TS 24mm f/3.5 on NEX7; Nikkor non-AI 105mm f/2.5 on A6000.
    – Novoflex T to A mount: Rikkenon 800mm f/8 on A99, A77 and A6000 (via an additional LAEA-1).
    Now I am considering buying the new Sigma EF 85mm f/1.4 with the EF-E adapter.

  • David Kilpatrick

    If the lens is set to infinity focus, but your target relies on physical sensor refocusing to achieve a closer focus, you will get unwanted curvature of the focus plane and this will result in progressive lowering of MTF towards the outer field. The lens must be set to the actual reported/marked focus of the target whether this is via an optical collimation system, or an actual physical distance. Otherwise, the very small adjustments made for both the focus and zoom mechanisms to optimise group spacing at closer distances are being ignored. Can I suggest that you try focusing the lens so that it shows and reports the actual distance of your target, and lock that in, before doing another test? I may be wrong but there’s a chance I am right and only you can check this.

  • David, all are valid points. Our bench setup actually does exactly what you suggest; the lens is electronically connected to an A7Rii throughout testing, maintaining focus at infinity according to the camera. It’s worked accurately for all other FE lenses and Sony’s engineers have come here in fairly large numbers to look at it and ooh and ahh. BUT, and it’s a reasonably big but, we’ve never tested a lens with dual AF motors separately controlled like the 70-200 GM. I’m not absolutely certain the camera’s report of infinity focus is absolutely accurate in this case, although every internal test suggests it is. But that’s why I’ve never made an ‘official’ blog post about it and put so many disclaimers on these MTF results.

  • David Kilpatrick

    As a result of testing Milvus lenses on adaptors on A7RII – and a few earlier tests – I’ve found that Sony bodies can vary at least ±0.1mm in mount register, and most adaptors vary this much too. As the Sony mount has only four support points, it can tilt easily (Fuji’s six-point system, and the Alpha A-mount six-point, prevent this). What I found with the Milvus 15mm test was something I’ve suspected when using floating element ultrawides and zooms before, even a .05mm register discrepancy can throw the centre to edge field correction out. The article on the 70-200mm implies that the lens had to be set up in focus and then remounted on the test rig, not the method described in the link above.

    I now have one A7RII body where I know the register collimation is as exact as I need for lenses down to 10mm and longer with very wide apertures. This was not the case with my A7R which needed a Tough E-mount to get correct register. I use Rosco Cinefoil to make shims. Longer lenses are, in general, less sensitive to errors in the mount and body but zooms can be very sensitive to internal alignment. I don’t intend to buy this huge lens as for me it defeats the whole point of the A7/FE system but the MTF figures look far from what any designer would want see.

  • bdbender4

    I think some of the comments miss the point. The only formal test was MTF for the new Nikon 70-200. Joey took some informal shots to follow up. I draw two conclusions:

    1. If I was in the market for the Nikon lens (I’m not) I would buy it.

    2. If I was in the market for the Sony lens (I’m not) I would wait a while until what seems to be early teething QC problems get worked out.

    For the record, I am just a private citizen with no affiliation with any camera company. Or with LensRentals, for that matter.

  • Omesh Singh

    Residual tilt due to the adapter will affect corner performance. If if there is sag then the top two corners will be behind center focus and bottom corners will be in front of center focus. Depending on field curvature of the lens this could have positive or negative impact on corner performance. If you add in some famous manufacturer decentering then all bets are off in terms of what one can expect in the corners. With adapted glass you can only assess performance in a small area around the selected AF point.

  • Omesh Singh

    I’m also a Canon shooter. In all honesty I chose the Tamron 70-200 VC + 135L for less money which gives me two options for events/portraiture to use on two different bodies (or having backup to cover the same medium-tele range) The 135L is magic SOOC.

  • UnSean

    I agree Sony looks a mess with soft points all over and varying from side to side with focal length, so either the camera is moving on it’s mount relative to the target or the lens has a moving decentering issue with focus shift, which is my opinion the problem.
    The good news is that at certain points it matchs the Nikon which looks excellent.

  • Jekabs

    Who knows, but the initial LR E-mount test result article says the lenses are electronically hooked up to the camera when mounted in OLAF, so that the focus can be adjusted ( ). If so, this is unlikely an issue of OSS parking, as the lens is powered on and thinks it’s on the camera. Also, since the lens is powered on, the AF actuator voice coils are engaged and the focus group should not be able to slip out of position.
    Geometry suggests that a similar misalignment of mount on lens or body, or sensor itself, would cause more tilt effect with a mount with shorter flange distance. Perhaps this is why Sony seems to be having more tilted lens issues than the DSLR competition, or even smaller sensor mirrorless competition?

  • Nado

    There is absolute degradation of quality on EF lens adapted to FE. It has been well documented. Not using a native flange/mount you never have the same mating from lenses to camera. Using an adapter the lens mates to third party adapter then adapter to third part camera, two specific contributors to degradation of image quality is a flange not square to a sensor let alone the diffraction passing through an extra flange mount. Absolutely less performance. If you cannot interpret this, obviously I am not able to explain properly if that is the case.

  • David Kilpatrick

    Roger, maybe this info is there, but how do you maintain the power/status to fix the OSS mechanism in a neutral (off) but not parked position if you are not using a Sony body? Same for the focus and any related object distance compensation?

  • David Kilpatrick

    OSS parking. Bet you they find this is the issue. Wonder how they test, for example, a Sigma OS lens where the parking is dramatically off-centre unless the lens is mounted and powered up?

Follow on Feedly