New Items

Sony FE 12-24mm f/4 G MTF Tests

Lensrentals.com, 2017

I usually don’t do any prerelease testing. Mostly that’s because manufacturers will send one prerelease copy to a reviewer. If you tell them you won’t test less than ten copies they look at you like, well, I don’t know what they look at you like because I only communicate by email. But they say nope; you can’t have ten. And I usually (unless I’m really, really interested, which is rare) just decide to wait until it is released.

The other reason I don’t often do prerelease testing is that there’s some gamesmanship about prerelease copies, especially if you’re not a giant review site. I’d rather just review ten purchased copies after release than play games.

With that in mind, I want to practice full disclosure about these ten copies. They are prerelease copies owned by Sony; not by Lensrentals. Sometimes we get the opportunity to test them for Sony before they’re released. I give them the data, make my comments, and keep my mouth shut for a while (because there is, of course, a keep-my-mouth-shut clause in our agreement). Then I wait six weeks until Lensrentals gets copies, test ten of those, and write a blog post.

In this case, because the lens tested really well, I asked permission to go ahead and do the write-up based on the tests of their copies, which they granted. If it had sucked, I wouldn’t have asked. So there, now you don’t have to read the article because I already told you it’s good.

Oh, and because someone is going to scream ‘those are hand picked copies,’ trust me, if most manufacturers could hand-pick half as well as you think they can, there wouldn’t be any reason for us to test them, would there? I’ve been doing this for years, and so far, I’ve never seen a big optical difference between pre-release copies and off-the-shelf copies.

So, About the Lens

Lensrentals.com, 2017

The Sony FE 12-24mm f/4 is a full-frame, ultra-wide angle zoom that will be available in August for about $1700. It has Sony’s ‘Direct Drive Supersonic Wave Motor,’ which is a linear piezo drive, like the one in the Sony FE 85mm f/1.4 GM lens. It has four aspheric elements, a curved aperture ring, yada, yada. Most impressive to me is it’s a reasonably small lens, around 4.5 inches long and weighing about 1.25 pounds (565 grams for those of you who have a logical measuring system). That’s a full pound lighter than the Sigma 12-24mm f/4 Art and Canon 11-24mm f/4 L lenses that I waxed poetic about in an earlier post. And Sony’s price is just a few hundred more than the Sigma and about $1,000 less than the Canon.

Let that soak in for a second. This is a Sony lens, that is not more expensive than the competition. It’s almost July. I’m looking for snow.

Some MTF Results

As always, these are the averaged results of ten copies, each tested at 4 different rotations. Because of the baffle used in the Sony mount there are fewer points to average at 18mm and 20mm away from center, so take those edge measurements with a tiny grain of salt.

At 12mm

Olaf Optical Testing, 2017

This is really very good. I was a bit surprised after I first picked up the lenses, which are tiny compared to the Sigma 12-24mm f/4 or Canon 11-24mm f/4, that the results would be this good.

 

 At 24mm

Olaf Optical Testing, 2017

The 24mm results follow Roger’s Rule of Wide Zooms; the long end isn’t quite as good as the wide end. But that’s as it should be, we don’t buy 12-24 zooms to shoot at 24mm most of the time. And honestly, it’s still quite good even at 24mm.

Sample Variation

We actually do have a variation number finally set up and approved. But until we’ve recalculated it for 500 lenses, I won’t be presenting it. I will show you our standard variance graph that basically shows you the range of these lenses. It’s better than average. Sony seems to be making progress with their sample variation in the newer designs.

Olaf Optical Testing, 2017

Especially at 12mm, there is almost no variation in the center. As with most wide-angle zooms, if your copy isn’t quite like the others it will probably be at the long end, but this is really a good performance.

Of Course You Want Comparisons

It seemed to me the logical comparison was with the Sigma 12-24mm f/4 Art.

12mm

Olaf Optical Testing, 2017

I’m sure some fanboys will find some hair-splitting here at 12mm, but whatever differences are insignificant.

24mm

Olaf Optical Testing, 2017

At 24mm the Sigma isn’t keeping up with the Sony anymore. The Sony is clearly better both in the center and all the way to the edge.

There aren’t a ton of other logical comparisons to make, but I’ll throw a couple up.

Canon 11-24 f/4 L

There is a real difference between 11mm and 12mm, but still an interesting comparison.

Olaf Optical Testing, 2017

And no, I’m not commenting other than ‘those are both excellent’.

Sony Vario-Tessar T* FE 24-70mm f/4 at 24mm

This is actually a useful comparison because I think people will carry both of these in their bag and may want to choose which to use at times. This is somewhat surprising to me, as usually the short end of the longer zoom will be better than the long end of the shorter zoom. Not true this time.

Olaf Optical Testing, 2017

Summary

This was a test of the MTF of a new lens. It was not a lens review. Had it been an actual lens review you would have been instructed which lens to buy via a handy click link at the bottom of the post. It was simply an MTF test. There are lots of other things that go into selecting a wide angle lens; flare resistance, autofocus accuracy, reliability, how it handles, etc.

The MTF results tell you if the lens should be a sharp lens, and the results here indicate that it should. That makes it worth further investigation for those of you interested in this focal length. It’s one of the more impressive Sony offerings that have come out; reasonable price, excellent MTF, reasonable sample variation.

There is something that is more interesting to me in the bigger picture, though. Sony has made this lens smaller than the others in this range, which will be nice sitting on an FE camera. And they’ve done that without the customary ‘Sony surcharge’, which is nice on  your wallet. At this point, I can’t help but be very impressed.

 

Roger Cicala and Aaron Closz

Lensrentals.com

June, 2017

 

Addendum:

Several people have asked about the field curvature of this lens, so I’ve posted the field of focus at 12mm and 24mm. Both are really flat at infinity, most impressive.

 

Olaf Optical Testing, 2017

 

Olaf Optical Testing, 2017

 

Another addendum for the Nikon 12-24

Author: Roger Cicala

I’m Roger and I am the founder of Lensrentals.com. Hailed as one of the optic nerds here, I enjoy shooting collimated light through 30X microscope objectives in my spare time. When I do take real pictures I like using something different: a Medium format, or Pentax K1, or a Sony RX1R.

Posted in New Items
  • No. 🙂

  • is there is chance to test Laowa 12mm 2.8?

  • Brandon Dube

    There is no singular stack thickness for a mount — e.g. Canon cameras vary from ~1.25 to almost 3mm. The older a mount gets, the more the stack thickness has converged, generally. In the case of Sony tests, everything we do now is with the complete coverglass assembly from an A7 series camera, I think it’s the A7rII one.

  • Chik Sum

    That’s an unexpected great lens! TBH I didn’t expected such performance from a G series lens, MTF alone it deserves even a GM status! But there’s one question in my mind, I remember in one of the articles in Lensrentals I read that the filter stack in front of the sensor plays an important role in the MTF measurement, may I ask did your tests now all incorporate the correct filter stack thickness with the mount the lens was testing?

  • Zé De Boni

    Going further: A new lens is being designed. Focal distance: 3 11/32″. Exit pupil: 1 27/32″. What will be its Fnumber? To use decimal in imperial measuring standards is cheating! It may work for simplified calculation like above, but it is hard to find imperial measuring tools with decimal divisions. What can we say of a system for calculating dimensions where 1 inch = 1/12 foot, 1 foot = 1/3 yard and 1 yard = 1/ 1760 mile??

  • Zé De Boni

    DXO works nicely for perspective correction. Photoshop and LR do it as well. The problem lies not in the software used, but on how the image is captured. It is wise to allow some extra space around the subject so that the final crop don’t cut part of it. Also, sometimes you get weird proportions and discover too late that the original shot should be horizontal, not vertical, or just the oposite. There is also an interesting side effect of in software perspective control, which is the loss of resolution of the area that is expanded, meaning that the maximum enlargement according to one’s standards is lower in direct proportion of the amount of correction.
    So far, the perspective control is more effective with a T/S or PC lens, because you can pre-visualize the final composition at capture. But there is a problem with DSLR’s, as the shifted lens is not aligned with the mirror, causing the OVF to blank (actually darken) on one side. Using lifeview solves this question, which shows one of the advantages of the EVF and by extension the mirrorless.
    I used the Nikon 24mm TS on the D3X, then switched to Sony A900 and the Rokinon 24mm TS with the same limitatons. Things became much more practical when I got the A99. My A900 is now retired. I also use Mamiya 645 lenses via the amazing T/S adapter by Mirex (they make it for Hasselblad lenses too).
    In fact all those lenses with extra coverage have some critical limitations. They hardly can compete in IQ with the regular primes of the FF format, because optically they behave as lenses of a larger format. The 24 TS, nominally ultra wide angle, is actually an extreme WA, which means it is more prone to all the defects expected, like distortion, poor edge resolution and CA. And those appear mostly on areas visible by shifting the lens, which is contradictory to the reason to buy them. Worse is the fact that there is no profiles for in camera or in post corrections of any PC/TS lens, as far as know. The images from shifted lenses are decentered, no information is recorded. Fifteen years ago I wouldn’t mind but today I can’t live without these software tools. Whenever I needed, I tried to correct my TS images enlarging the canvas to align the optical center to the canvas center then apply a manual aberration, distortion and vigneting correction. Finally I cropped off the white addiction. Insane!
    There is no substitute for the tilting function, but if the case is not of focus control but to manage perspective, then by correcting in software any good prime or zoom lens can do it with similar or even better results. From what i exposed above, a TS should be worth its higher cost only if could offer a far better, perfect IQ. I asume this from my practical experience, but recognize that this is a nice subject for an in deep research and I may change my mind if the contrary is proved. Are you reading, Roger?
    Meanwhile, I still believe that in the future the mirrorless cameras will profit from the electronic viewfinder and provide applications to visualize the perspective control, which may be manual or automatic, computing the inclination angle and the lens used, any kind of lens. But if any lens maker could offer tilting, just tilting, autofocus lenses with full profiles, I would be the first on to order a whole set.

  • Zé De Boni

    Sorry, but this looks like an illisionist’s trick. All TS/PC lenses are actually lenses with larger image circle covering a larger format than the system for which they are designed. A 24mm for FF projects a cone with about 82 degrees. The TS with the same focal length has a much wider cone, maybe 100 degrees, which is in the same range of the 16-18mm, meaning that the same optical design could be used for a MF extreme wide angle lens (should be mirrorless because of the short flange distance). When you make multiple shots by shifting aTS lens, the final image has the same wider angle of view and the main advantage is to increase the Mp count. If you shift the lens by 10mm for each side in a vertical position the final combined image size will have 36×44 mm, almost doubling the Mp count but unable to get the same 2×3 proportion (easier to make this calculation because of my wiser measuring system) unless you crop and discard a bunch of your gained pixels (and get a 29.7x44mm final image; how much is that in inches?). All the optical problems (vigneting, CA, distortion) on the edges of are the same as those of extreme wide angle lenses and theoretically may look the same as those for a 18mm lens, because proportions are mantained. Those used to large format photography know that all Super Angulon’s (f.i. 47, 65, 90, 121mm) have the same optical design, the difference being just the scale to cover the corresponding formats (6x9cm, 4×5″, 5×7″, 8×10″). Therefore you may expect proportional aberrations, distortions and vigneting. You may use the 121 with a 6×9 back and make 24 shots shifting the camera back (not the lens, so you keep the viewpoint), and your resulting stich will be similar as if you do just one 8×10 shot. You get finer grain, but the optical problems will be the same, because, theoretically, they have just grown in proportions. The same applies for our original problem, but in the digital era you will miss an importan tool: the compensating lens profiles, which improve so much IQ mostly at the edges. You may find profiles for almost any wide angle lenses of your native system, but not for TS/PC’s.
    The tragical truth is that since these profiles (combined with perspective correcting tools in software) are available, the shifting lenses became obsolete. There will always be uses for the tilting function, indeed. So, to get a good quality image with full control of the perspective you must shoot with a good lens with whatever angle of view and camera inclination required by your best viewing point, use the corresponding lens correcting profile and then use the available perspective correcting tools on your editing software.
    Quite interesting is the original idea of using the wider angle lens like those of the featured zoom leveled to get paralel vertical lines and then cropping the image. Yes, you may correct its optical deffects, but then you will discard many Mp by cropping. So if you are at a distance of a building where you may get paralel lines with the 12mm (but then having to crop the image to get rid of unwanted extra ground or other features) or full coverage with converging lines on a 16mm lens to be corrected in software, my practical experience tells that the second option will result in better IQ. That is a nice subject for lens testers all around.

  • l_d_allan

    Thanks for the article with test of the FE85 f/1.8 … and on Friday. Have a great 4th weekend!

  • Don Farra

    Thank you.

  • you don’t think the 14-24 is similar to the Sony??
    my apologies.

  • Zé De Boni

    I see that you like challenges. I defy you beat my speed to calculate these simple practical problems that photographers always face. You use US (or imperial if you wish) standards (with base 2 fractions, not decimals) and I use the metrical system. None of us may use a computer, calculator or paper and pen.
    Problem #1: One wall of your great exhibition on that famous museum is 50′ (15m) long (wide if you prefer so). You want to hang 15 framed prints on a straight line leaving exactly the same space between them (boring, but that is to ease calculations), and allowing twice as much space on the edges, which gives a total of 18 space units. Each frame measures 20 inches wide (this is a section of the show with vintage prints of when you used small enlargements).
    Question: What will be the exact size of the afore mentioned space unity?
    Problem#2: You will custom make developing tanks with one gallon (4.0 l) of processing chemicals. You need them to be, internally, 6″ (16cm) wide and 6″ deep (16cm) to hold 4×5″ hangers with ease.
    Question: How long (internally) will be that tank?
    Problem #3: … Sorry, your time is over!. You haven’t finished problem 2 yet!
    I got these ideas on a conversation with an American Mid-western Physics teacher (and nature photographer too), who tried to show his students how foolish it is to keep these stupid measuring standards (sic). To his comment I added: “Well, at least it is rational. I mean, as rational as to use Roman numbers”. Well, that was back in MCMXCVII, V years before my first digital camera. The digital revolution is here, but obsolete minds persist.

  • Brandon Dube

    How about a 1 meter f/2.2 portrait lens? https://medium.com/@brandondube/photos-from-the-george-eastman-house-vault-3df3f1e45780

    39-inch, if you want to use imperial units.

  • Brandon Dube

    “Onion ring bokeh” is the result of using CNC optical manufacturing techniques, it is a product of tool marks left in the process. A sufficiently well tuned CNC platform will not leave them. Those tool marks would not cause the MTF as a function of field to be jagged — that is a product solely of the optical design. Departure from the nominal would only serve to smooth things out.

  • SpecialMan

    I defy you or anyone to tell me a more magnificent lens than the 13-inch Dallmeyer Rapid-Rectilinear or the Voigtlander 11.25″ Petzval. You cannot because today’s lenses are so embarrassingly tiny they have to be measured in millimeters—like tumors.
    How disappointed Sander, Emerson, Watkins, Atget and the other greats of photography would be if they saw how the majestic brass and wood cameras of their day have been reduced in our time to vinyl-covered plastic toys.

  • SpecialMan

    I meant the shape of the curve on the charts. If you look, especially at the 50 lp/mm lines for the Sigma, they’re all over the place. As far as the shape of the lenses, I read somewhere that the reason you get onion-ring bokeh is because the lens blanks are molded with stepped sides and it’s impossible to completely grind off the steps. I have no idea if it’s true.
    I’m just a guy who uses cameras instead of booze and broads to give himself a reason to live.

  • SpecialMan

    And would you mind adding the 135mm lens I bought at a camera store in Hollywood in about 1982. The store, Schafer’s, is long gone and I don’t really remember the brand but the lens had a silver barrel and it was a Pentax screwmount. I still have the lens somewhere, it’s no trouble to send it to you….

  • nighley
  • Which Canon? The 16-35 Mk III should have the same look as this, they’re on the same software.

  • Well, I try to avoid those because Fanboys, but I imagine there are some people who are shooting the Nikon on adapters, so I’ll try to put them up as an addendum.

  • nighley

    How is the Canon at 16mm and 24mm compared to this lens? All I found is the old blog post which had the old MTF charts which look a bit different, it looks like the sag lines look better but has more astigmatism, am I reading the charts right?

  • Sorry to ask, but is there any comparison to the 14-24mm Nikon (which many of us are familiar with in everyday use)???

  • Oskar Ojala

    Have you seen any modern 85 mm prime lens that wasn’t ok in terms of MTF numbers? 😉

  • Pako

    I know, Roger, the 16-35 is a good lens!
    but, do I need 3 lenses going from 12mm to 70mm when I can do it with 2 lenses?

  • Pako, if you want wider than 16 then I’d say so. But I wouldn’t sell it otherwise; the 16-35 f/4 ZA is a good lens.

  • Zé De Boni

    Sorry, but this looks like an illisionist’s trick. All TS/PC lenses are actually lenses with larger image circle covering a larger format than the system for which they are designed. A 24mm for FF projects a cone with about 82 degrees. The TS with the same focal length has a much wider cone, maybe 100 degrees, which is in the same range of the 16-18mm, meaning that the same optical design could be used for a MF extreme wide angle lens (should be mirrorless because of the short flange distance). When you make multiple shots by shifting aTS lens, the final image has the same wider angle of view and the main advantage is to increase the Mp count. If you shift the lens by 10mm for each side in a vertical position the final combined image size will have 36×44 mm, almost doubling the Mp count but unable to get the same 2×3 proportion (easier to make this calculation because of my wiser measuring system) unless you crop and discard a bunch of your gained pixels (and get a 29.7x44mm final image; how much is that in inches?). All the optical problems (vigneting, CA, distortion) on the edges of are the same as those of extreme wide angle lenses and theoretically may look the same as those for a 18mm lens, because proportions are mantained. Those used to large format photography know that all Super Angulon’s (f.i. 47, 65, 90, 121mm) have the same optical design, the difference being just the scale to cover the corresponding formats (6x9cm, 4×5″, 5×7″, 8×10″). Therefore you may expect proportional aberrations, distortions and vigneting. You may use the 121 with a 6×9 back and make 24 shots shifting the camera back (not the lens, so you keep the viewpoint), and your resulting stich will be similar as if you do just one 8×10 shot. You get finer grain, but the optical problems will be the same, because, theoretically, they have just grown in proportions. The same applies for our original problem, but in the digital era you will miss an importan tool: the compensating lens profiles, which improve so much IQ mostly at the edges. You may find profiles for almost any wide angle lenses of your native system, but not for TS/PC’s.
    The tragical truth is that since these profiles (combined with perspective correcting tools in software) are available, the shifting lenses became obsolete. There will always be uses for the tilting function, indeed. So, to get a good quality image with full control of the perspective you must shoot with a good lens with whatever angle of view and camera inclination required by your best viewing point, use the corresponding lens correcting profile and then use the available perspective correcting tools on your editing software.
    Quite interesting is the original idea of using the wider angle lens like those of the featured zoom leveled to get paralel vertical lines and then cropping the image. Yes, you may correct its optical deffects, but then you will discard many Mp by cropping. So if you are at a distance of a building where you may get paralel lines with the 12mm (but then having to crop the image to get rid of unwanted extra ground or unwanted features) or full coverage with converging lines on a 16mm lens to be corrected in software, my practical experience tells that the second option will result in better IQ. That is a nice subject for lens testers all around.

  • Pako

    So, Is time to sell my 16-35 and buy the 12-24?

  • I heard about a NiSi 150mm Filter Holder For SONY FE 12-24 f/4 but have no firsthand experience.

  • davicito

    Just wandering, as cameraman, is there any solution to put an ND filter on this threadless lens?

  • appliance5000

    You get a field of view of a 16.5mm but you don’t get the spacial distortion.

Follow on Feedly