LensRentals.com

NEX-7 Lens Imatest Results

Posted by

For the last 3 years now, I've been running my mouth, basically saying something like 'I think the NEX series is a camera that could give great resolution if there were any decent lenses for it'.  When the NEX-7 came out there was now an even better resolving sensor, and for the first time at least a few supposedly good lenses to put in front of it (if you can get them). A lot of people preferred high quality Zeiss or Leica M mount lenses on an adapter, and Sony was smart enough to put nice peaking filters and other manual focus aids in the camera.

But some of us (myself included) like to shoot native-mount lenses, or can't afford the price of Leica glass to put in front of the camera. The NEX-7 is physically set up in a way that allows us to mount it to our Imatest set up, and we happened to have most of the new lenses in stock one day, so I thought I'd take advantage of the situation to do some resolution testing. (Because I know someone will ask, we can't test many of the small cameras because mounting to our tripod system blocks the memory card door. It takes 10 to 20 test shots to align the test setup. But if you have to unmount from the tripod mount, you have to start over. So it's not possible for us to test some camera systems.)

The usual “don’t read this stuff and go insane” cautions apply: this is a test of lens and sensor resolution done at a distance of 10 to 30 feet, depending on focal length. The results might be different at 4 feet or 400 feet. Focus is done manually and bracketed so it’s not a test of a camera’s autofocus ability. The tests use a controlled-lighting test-target with a hardware mounted camera and Imatest software at ISO 400, so photos taken in the dark of night or hand-held for 12 second exposures will give slightly different results. It doesn't measure autofocus speed, lens size, manual focus feel, bokeh, color rendition or anything like that.

The Lenses

Still the results are interesting and I suspect none of you would pick one of the eventual resolution co-champions out of the lenses we tested:

Leica 50mm f/1.4 Summilux on a Fotodiox adapter (picked because it's the highest resolution 50mm lens we've tested on other bodies).

Sony 18-55mm and 18-200mm E-mounts zooms both tested at 30mm because that was our original point of interest. That was probably the sweet spot for the 18-55, but may have handicapped the 18-200 a bit: it seems better around 50-90mm. But certainly it's better at 30mm than at the long end.

Sony 16mm f/2.8; 24mm f/1.8; 30mm f/3.5 Macro and 50mm f/1.8 E mount prime lenses.

Sigma 19mm f/2.8 and 30mm f/2.8 E mount lenses.

Unfortunately we did not have a 55-210 Sony zoom in stock at the time of the tests.

Imatest Results

The values are for MTF-50 (which correlates with fine detail resolution) measured in line pairs / image height. The first (higher) number is the MTF 50 measured at the center of the lens only. The second (lower) number is the average of the MTF-50 measured at 13 points including the center, 4 corners, 4 sides, and 4 mid points.

The first graph shows the numbers for each of our lenses (zooms measured at 30mm) with the lens wide open. This test isn't a level playing field, since some lenses are f/4 wide open (the zooms) while the primes are at anything from f/1.4 to f/3.5. But since some people tend to shoot every lens wide open, it may be useful to them. Please note: On Monday, May 7th I retested a number of Sony-Zeiss 24mm f/1.8 lenses because the initial two we tested seemed strange, and were very close in SN (it's all we had in stock at the time). I've since repeated that test with different copies and the new results at f/1.8 are much better. I've changed the results below to reflect the better results -- here's a good example of tests of one copy of a lens, even two copies, is never enough to base your decision on. We've examined the copies from the original test carefully and other than the softness at f/1.8 (which is gone by f/2.2) we still haven't figured out what problem they have.

MTF 50 at Widest Aperture


Lens Max Avg
Sony 16mm f/2.8 665 490
Sigma 19mm f/2.8 870 740
Sony-Zeiss 24mm f/1.8 740 590
Sony 18-200 f/4 855 680
Sony 18-55 20 f/4 855 640
Sony 30mm f/3.5 940 725
Sigma 30mm f/2.8 985 805
Sony 50mm f/1.8 675 602
Leica 50mm f/1.4 627 545

We can level the playing field a little bit by shooting the wide primes at f/2.8. That gives us the numbers below. It's somewhat more meaningful: the zooms are still f/4 and the Sony 30mm Macro at f/3.5, but everything else is shot at f/2.8 here. Notice the results for the little, reasonably priced Sigma 30mm f/2.8 are in bold. While some lenses are close to as sharp in the center, nothing is nearly as good in the corners. Impressive.

MTF 50 at f/2.8


Lens Max Avg
Sony 16mm f/2.8 665 490
Sigma 19mm f/2.8 870 740
Sony-Zeiss 24mm f/2.8 960 735
Sony 18-200 f/4 855 680
Sony 18-55 20 f/4 855 640
Sony 30mm f/3.5 940 725
Sigma 30mm f/2.8 985 805
Sony 50mm f/2.8 806 732
Leica 50mm f/2.8 920 730

We can level things out a bit more by shooting all of the lenses at f/4. Most of the lenses would get a bit sharper still at f/5.6, especially in the average number. Notice the Sony 24mm f/1.8 has gone from easily the worst resolution to among the best (at least in the center) by stopping down a bit. It really was much, much better at f/2.0 so I would consider that the real maximum aperture for that lens. (We checked a second copy and it was exactly the same - it's possible that we had two copies that were bad at f/1.8 and not at f/2.0, but I really, really doubt it. I'll check a couple of others when they get back from rental to be certain.) Really only the Sony 50mm and the zooms are lagging behind in the center, and the Sony 16mm pancake does, of course, still suck.

MTF 50 at f/4


Lens Max Avg
Sony 16mm f/4 915 646
Sigma 19mm f/4 920 765
Sony-Zeiss 24mm f/4 1000 795
Sony 18-200 f/4 855 680
Sony 18-55 20 f/4 855 640
Sony 30mm f/4 940 725
Sigma 30mm f/4 1040 840
Sony 50mm f/4 885 835
Leica 50mm f/4 1020 870

Comparing Lenses

Imatest is measuring the resolution of the lens-camera combination (as opposed to an optical bench, which measures the lens' maximum using collimated laser or LED lighting). We chose the Leica 50mm f/1.4 Summilux because we've tested it on a lot of other camera systems and know the lens is as good as it gets from a resolution standpoint. That the Sigma 30mm, a $200 lens, is keeping pace with it is really amazing. Obviously it's not really a wide aperture prime, with a maximum aperture of f/2.8, but it gives world-class resolution. The Sony 24mm f/1.8 is not the bargain the Sigma is, but from f/2.2 on it has amazing center resolution, although it lags behind the other two in the corners. The Sony 50mm f/1.8 appears to be tuned like the Nikon 50mm f/1.4 G: it's not quite as sharp in the center, but keeps good resolution across the entire front of the lens. Notice it's average resolution is up there with the 3 best lenes, although the center isn't quite as sharp as the other primes.

The Sony 30mm Macro may be better at Macro distances (should be actually), but the Imatest setup requires us testing it at about 15 feet distance. Very decent also describes the Sigma 19mm f/2.8 which is clearly far better than Sony's 16mm pancake. The zooms, well, they're about what we thought: OK lenses for when you need a zoom.

Comparing Systems

It's always a little complex to compare Imatest resolution between cameras with different sensor sizes, since the numbers we generate are measured in line pairs / image height, and image height differs.  I hesitated to do it at all, because it will inevitably lead to 716 posts about whether it's most appropriate to multiply the resolution of the crop sensor of divide the resolution of the full frame, and by what amount. But my thought is to keep it simple. If I take a picture of a building with whatever camera I'm shooting so that it is the same size on the image I print, then image height of the sensor doesn't matter.

In other words the printer doesn't care what size the sensor that took the picture was if the pictures were the same. Given this 'Roger Rule' the results with the NEX - 7 and good lenses are about what we'd expect: resolution at f/4 with a good lens is about what we see with a Canon 5DII or III (1050 / 800 or so LP/IH) with a good lens, not quite as good as a Nikon D800 (1200 / 1000 LP / IH). So shooting with one of the better lenses  listed above you can certainly get all the resolution you should ever need.

If you want to shoot a zoom, though, things aren't so pretty. For example, a Nikon 24-70 f/2.8 shot at f/2.8 gives around 890/830 jumping up to 940 / 875 at f/4. A Canon 70-200 f/4 resolves at 920 / 845 at f/4. The Sony zooms are both about 855 / 660, which means the corners of a good Canon or Nikon zoom are about as sharp as the center on the Sony E zooms. Those are more expensive zooms, no question. I'm not trying to compare apples and oranges. Simply stating that the NEX system doesn't offer a high quality zoom and giving some examples of what a high-quality zoom's numbers would be like.

But for prime lenses, at least, we're now getting some good quality, native mount lenses to put on NEX cameras without an adapter. My hat is off to Sigma for making a couple of really good lenses at amazingly good prices. Well done! Hopefully as the user base of E mount cameras increases we'll start to see more third party options.

 

Roger Cicala

Lensrentals.com

April 2012

 

40 Responses to “NEX-7 Lens Imatest Results”

Christopher Campbell said:

Roger, I have a large system of Canon primes and just recently purchased an NEX-7 as a compact camera, and really appreciate your doing this testing. I am slightly confused, however, about the "Sony 24mm f/1.8" lens. By that I assume you are referring to the Zeiss ZA 24mm f/1.8 E-mount lens? I too have found that stopped down to about f/4 on the NEX-7 it gives a Canon 5DII with a Canon 35/2 lens a run for its money, but it's a bit shocking to see how badly it fared at f/1.8!

Promit said:

Roger, sorry if this is a stupid question but regarding the memory card door issue: why can't you just use an Eye-fi card? That would allow you to pull the calibration images without touching the camera at all, let alone dismounting it.

LensRentals Employee

Roger Cicala said:

Hi Christopher,

Yes, the ZA 24mm f/1.8 is the one. It was a surprising finding - not that it was so soft at f/2.8, but that it sharpens up so nicely by just f/2.2. I thought at first, despite the fact the lens looked good, that it might be decentered. But the second copy behaved exactly the same way. I still want to check more copies to confirm, though.

Roger

LensRentals Employee

Roger Cicala said:

Promit,

That would be a good solution, if we didn't have so much trouble getting Eye-fi cards to send files consistently. Might just be us, but it's incredibly frustrating to have it quit halfway through 2 hours of testing and have to start all over.

Samuel H said:

Wow, that sigma 30mm f/2.8 looks like a great deal!
My "low budget, nearly pocketable" recommendation has been the same for years: EPL-1 + Panasonic 20mm f/1.7
It looks I could finally update it: NEX-5N + Sigma 30mm f/2.8
But I still have to find out about bokeh on that little Sigma... Any comments about it?

Also: about the card slot thing
The solution is to put one of these between the tripod and the camera:
http://www.ebay.com/itm/Leica-Leitz-Table-Tripod-Ball-Head-for-M8-M9-KGOON-/251051206013?pt=LH_DefaultDomain_0&hash=item3a73d1657d
(maybe a cheaper one would work too, I'm sure you have lots lying around)

RussellInCincinnati said:

Interesting that you don't really get a huge benefit from stopping down the Sigma 30mm F/2.8 to F/4. Wonder how all the lenses would do at F/8.

jim thomson said:

Sigma appears to have hit a home run with the 30mm f2.8. Great resolution at a low price.

Bart Zoni said:

Hats off to Sigma, really. I purchased the 19mm. They created a $200 'perfectly unoffensive' lens. It provides good contrast that isn't harsh but lacks a destinctive signature, usable resolution across the field at all apertures, neutral color, out of focus areas that are smooth but not buttery delicious, AF that is a bit deliberate but accurate, files that can take a little sharpening abuse, plastic finish that is not cheap to the touch. You don't look at the images and say WOW but you look at them in context of the price and say WOW!

Chris said:

I also made a comparison of FD lenses versus kit versus SEL50.
In my test, the SEL50 improves already at f2.0 in the center, but slowly improves at the corner from 1.8 to f4. The kit zoom is very good in the center at 24mm, but wideopen the corner sucks. This is confirmed by your test.

Still, what apparently cannot be put into plain numbers is the contrasty, almost sparkling color rendition of the SEL50. I have shot some cars and the chrome fenders look spendlid.

With the kit zoom, even under best conditons, there is no such "sparkle" :-)

Chris said:

Ah, something I forgot to mention: On the Nex 5N in P-mode, the Program will rather raise the ISO to 1600 before leaving the setting f4 & 1/80 sec. In extreme, it will select f2.0 but in my experience, never f1.8. That looks like Sony knows where the lens performs best.

Zachery said:

Interesting results. Having shot with most of these lenses, it's always fun to compare your real world experience with some hard facts measurements. One thing that's easy to forget is that the Imatest tests don't consider color specific corrections or color reproduction. It's easy to forget as well that Zeiss is best known for color handling. What's more, 24mm seems to be notoriously difficult a focal length to make well, just look at tests of all the full frame 24mm lenses to see what I mean. Being smaller in diameter doesn't make the optical formulas any less difficult to concoct. Finally, even though the numbers seem similar, 24 and 30 are worlds apart already as the shorter the focal length, the more significant the difference between them.

Actually, that last point is a reason why you might consider owning both the E 24 and DN 30mm lenses in your kit.

Personally, I love the images produced by the E24 and E50mm. The 50mm has especially wonderful bokeh and color reproduction and it even handles harsh light fairly well. The 24mm just makes sweet satin out of just about everything you point it at. I can corroborate with my personal experience that ƒ/1.8 is not worthwhile. Use it if you intentionally want softness, otherwise stick to 2.2 or so. Generally when I'm walking around with it, I'll leave it on 2.8 if it's dim light, or 5.6 otherwise. 5.6 is its sweet spot.

Now, if there was just one more lens I could see tested on NEX-7, it would be the Sony 135mm ƒ/1.8 Zeiss on the LA-EA2 adapter. Mainly because it's hailed as one of the best lenses ever made and because I'm thinking of picking one up for my NEX-7.

Nilesh said:

Thank you so much for these tests! no question Sigma 30 f2.8 is best native for Sony! f2.8 is fast enough I say. One thing that your test alludes (since this is camera/lens test and not lens test) to, that is perhaps importance to some is, leica 50 Summilux does not work well on NEX 7 (as oppose to Nex 5n)...it smears the corners badly and even center resolution is not as good as the lens is. But that does more good to Sigma doesn't it?

David Stock said:

I'm interested in how well the Sigma 30mm does at its best aperture (probably f5.6?). When I started using this lens, it knocked me out. It's very sharp to f8 and beyond and has nice clarity, decent bokeh and low distortion. CA is minor and easily corrected. I was using Pentax Limited glass on my NEX-7 before, but this Sigma is smaller, lighter, sharper---and it has autofocus.

One advantage of APS-C sensors is that they allow more depth of field than full frame cameras with the same field of view. I realize that some people don't like that, but for my work it's a plus. If I can get the resolution I need for large prints from a smaller sensor, I'm happy.

I was looking over your list of Nikon lenses for the D800, and most of them don't really get near 1200/1000. Quite a few of them are in the same ballpark as the Sigma. Which is tested here at a sub-optimal aperture!

GH said:

The Sigma 30 is my most used lens on my NEX-7. Sharp, lightweight and inexpensive. :)

David, FWIW, aps-c sensors don't really give a depth of field advantage over 35mm sensors, because 35mm sensors perform better in low light. So, if you're shooting ISO 200 at f4 on an aps-c sensor, then you can just boost the ISO to 400 on the 35mm sensor and shoot at f5.6 to get similar depth of field. Since the larger sensor should perform about a stop better at any given ISO, you're essentially left with the same noise performance and depth of field, so there's no real advantage with the smaller sensor (assuming similar sensor technology.)

GH said:

Hi, Roger. I forgot to ask in my last post, do you have the Sigma 30's mtf numbers for f5.6? Thanks!

lensporn.net said:

I said a long time ago that the Sigma lenses would be good but people wouldn't have it! They kept complaining that the focal lengths were useless and the apertures too slow! Some people just don't seem to realise you can put any lens on your camera and go out and find things to take photos of, you adapt to the lens you are using and zoom with your feet!

Serge said:

Even the #'s on the Sony's 30 are impressive considering everyone has been complaining about it, didn't expect to see those kinda #'s

David Stock said:

GH, I understand what you're saying, but it just doesn't work that way in some situations. If you want maximum depth of field without losing sharpness to diffraction, larger formats put a squeeze on your options. For instance, a Nikon 45mm lens on full-frame will give significantly less depth of field at f8 than the 30mm Sigma on APS-C at that aperture. (Both lenses will start to lose sharpness stopped down past f8.)

Digital is unforgiving to depth of field. Lloyd Chambers says that you need to stop down two stops from what is indicated on the typical lens depth of field scale. So that means the Sigma 30mm will give you sharpness from roughly 37 feet to infinity stopped down to f8. A Nikon 45mm lens on a full-frame body will give you sharpness from roughly 56 feet to infinity at f8. Big difference, at least for some purposes....

sam said:

I'm looking for a portrait lens for my nex-5n, should I get the Sigma 30mm F2.8 or a 50mm F1.x prime?

LensRentals Employee

Roger Cicala said:

Sam it really depends on what focal length you prefer. I'm a 'long portrait' type so I'd prefer the 50, but a lot of people prefer working around 50mm full-frame equivalent which would make the 30mm Sigma a good choice.

TerryB said:

Roger, a fascinating series of tests. I was considering the 30mm Sigma before being pointed in the direction of your site. So can you help further?

Do the conclusions still hold ie would the lenses still rank the same on the lower resolution of the NEX 5? Or am I missing something?

GH said:

David Stock, diffraction starts earlier for the smaller sensor, so you're still not getting any advantage. It's always a one way street in terms of diffraction and depth of field. A larger format can always do the work of the smaller format, but the smaller format can't always match the larger format.

Steven said:

Did you notice that certain batches of this lens would get an oval shape when the aperture is set to f/22? I suspect that these "poor quality" batches are not sharp when wide open.

Casey said:

Hi Roger. First of all, I want to commend you on this fantastic article. I was wondering if there was a section on your site where you published these lens scores? I'd like to take a look at the scores for a few other lenses that you're renting out.

Jordan retro 11 space jam said:

Thank you for this post, I am sure I will come way feeling better as I hope you are today.

TP said:

At f/4.0, "the Sony 16 mm pancake does, of course, still suck." The max resolution is equal to the sigma 19. Does it "suck" because of the average result?

LensRentals Employee

Roger Cicala said:

TP,
Yes, because of the average result. We don't have space to put all the various numbers (at least not yet) but the average is low because the corners and edges are simply horrible. With that good center number, and the corners/edges counting for only 25% of the average value, you can infer how bad they must be to drag the average down that far.
Roger

BF said:

Is there and aperture for the Sony 16 mm pancake where the corners and edges improve? If better than horrible at say f8 then may be a good alternative for outdoor sunlight use. Thanks for all your testing and publishing the results in the detail you do.

LensRentals Employee

Roger Cicala said:

BF - they're definitely better by f/5.6, maybe a tiny bit better than that at f/8. It's certainly decent at that point, although the corners are still a bit soft.

franz said:

hi, i've been thinking about getting a nex 7. i was doing some research and read a lot of complaints about images having very soft edges, some were saying it was a problem with the sensor. i konw that sharpness in the center will always be better than sharpness on the edges but from what i read it seems that on a nex 7 the edges are a lot worse than what is expected.

is there really a problem with the sensor? or is the problem with the lenses?

thank you

LensRentals Employee

Roger Cicala said:

Franz,

There definitely are some issues with the edge and corner microlenses on NEX7s when shot with certain wide angle lenses. I don't know if "problem with the sensor" is the right description because Sony certainly didn't design it so it would work with rangefinder lenses. But it is an issue for some folks.

Lee said:

I am very surprised by the poor maximum aperture performance of the Summilux, I'd really have expected more for the $3 billion price tag.

wansai said:

I got my hands on a zeis 24 f1.8 recently on a Nex7 and it consistently outperforms the kit lense in my shots by quite a large margin (wide open). The corners are a tad soft but not nearly as bad as the 18-55 and the 16. Otherwise the zeis is pretty consistently sharp; though not overly sharp.

I know some people base a good lense entirely on how super sharp it is, but I'm perfectly happy with the clarity of the zeis. Also, it renders very pleasantly, colours pop, and everything i point it at comes out so smooth and buttery. And just as importantly, pictures shot with the zeis have a 3D quality that none of the kit lenses can produce (generally flat images in comparison).

It's a quality lense that's not well represented by this data alone.

Samuel H said:

Any chance of throwing in some results with the metabones "speed booster" adapter?

Teresa K said:

Roger: Could you tell me if there is an adapter that would work with the NEX 6 or 7 for my old Zeiss glass made for a Contax RTS film camera?? My favorite is the Planar 1.4/85 ... I love that lens. I also have a Distagon 1.4/3.5 and aTeleTessar 3.5/200. Interested in lightweight now because of neck problem. Or are there other digital cameras that have some way to adapt. I discovered your site and itrust your opinion. Going to Yellowstone at the end of July. Will contact LensRental to get a recommendation for the trip.

thanks for your good site!

LensRentals Employee

Roger Cicala said:

Teresa,

It's possible in theory but I don't know of a way to do it without something clumsy (two adapters, but that just has a huge chance of introducing some tilt or infinity focus issue). There might be a direct adapter way that I'm not aware of, though.

Roger

Fred said:

Teresa K,

AFAIK the Contax RTS has a Contax/Yashica mount. There are some simple adapters (without optics) advertised on Ebay. Search for "sony nex contax converter".

I use a 70's era Canon FD 100mm f2.8 SCC at full aperture with a NEX 5R and a simple adapter. Sharpness is very good across the frame, with vibrant colors. You should get good results too. The only issue is my adapter is too short by about 0.2 mm, which results in the focus ring showing a slightly shorter distance than the actual.

Sofia Koutsouveli Lucifairy said:

The Sigma 30mm f/2.8 really shines! I wonder how it compares to the Sony 35mm f/1.8 OSS though.

Leave a Reply