Nikon 58mm f/1.4 is (hopefully) Not About the Numbers

Posted by

The Silence

We have a routine when a new lens comes out at Lensrentals: the first new copies get sent to me for optical testing. I have about 4 or 5 hours with them because they'll have to be packed up and shipped out to the customers that have been waiting for them and expect them tomorrow. So when Kenny brought me the first half-dozen Nikkor 58mm f/1.4 lenses he said the most unexpected thing. "We only have 1 preorder, you can keep 5 of them."

We got another shipment the next day. Most of them are still on the shelf, too. For whatever reason, there doesn't seem to be a lot of mouth-watering excitement about this lens.

Today's Comparison

Let me be fair. Nikon has said, very clearly and very plainly, that this lens is not about the numbers. It was designed to have a very smooth look, have excellent bokeh, minimize sagittal flare and coma for shooting lights at night, have limited vignetting, and be evenly sharp across the field of view. Those are great goals and in real photography are often more important than how well a lens resolves.

But they are largely things that will require field photographic evaluation, not lab evaluation, to determine. Which means you'll have to wait for the real photography reviewers to tell you about that. I'm not one of those; I do optical lab testing.

But, I had some lenses, and I had some machines, testing lenses in the lab is what I do, so I tested them. Just take it for what it's worth: a lab evaluation of a lens that isn't designed for lab evaluation.

All of that being said, if I was deciding if I needed to buy the Nikkor 58mm f/1.4 or something else, the logical something else would be the Nikkor 50mm f/1.4. (The 58mm f/1.2 would be the other logical comparison, but I had to go one way or the other, so I went with dueling autofocus lenses.) So that's what I'll compare today.


The Nikkor 58mm and 50mm f/1.4 G lenses

Obviously the 58mm is quite a bit larger and has a cooler-looking hood, but let's take a look at the other differences.

Nikon 58mm f/1.4 GNikon 50mm f/1.4 G
Angle of View (degrees)40.5 46
Min. Focus Dist. (feet)1.91.5
Aspheric Elements20
Diaphragm blades99
Weight (ounces)13.69.9
Filter thread72mm58mm

The optical diagrams show that the 50mm is a fairly classic double-gauss design, while the 58mm is more highly modified.


Nikon 58mm (left) and 50mm (right) f/1.4 lens diagrams


Nikons computer-generated MTF charts clearly show improved resolution and astigmatism with the new 58mm lens at both MTF 10 and MTF 30. So there should be a clear resolution difference, particularly out towards the edges. But those are theoretical, computer-generated MTF charts, and I like reality, so let's do some testing.


Nikon MTF charts for 58mm (left) and 50mm (right) lenses


Imatest Evaluation

I compared 6 copies of the Nikkor 58mm f/1.4 G with 6 copies of the Nikkor 50mm f/1.4 G from our stock. The 50mm were used copies, but all had been recently tested and were known to be in spec. (I should also note that I did not use previous results we've gotten testing Nikon 50mm f1.4 because our testing methods have improved since the last time I tested these - larger charts, better focusing techniques, etc. - so I didn't want to use results obtained with older testing.) Testing distance was 13-16 feet (depending upon which lens) using unsharpened raw images from a Nikon D3x as test camera. If you want to get into the geeky parts of our Imatest procedures you can read all about them HERE or HERE.

The table below shows the average (mean) MTF50 in LP/mm at the center, averaged across the entire surface of the lens (26 points), and averaged for the 4 corners.

Center MTF50 Avg. MTF50 Corner MTF50
50mm f/1.4 G620555395
58mm f/1.4 G645520435

OK, it's easy to see that the center resolution is tiny bit better for the 58mm lens, and that the corners are clearly better (which is what we were told would happen). But there's one weird thing that should stand out. Despite being not-quite-as-sharp in the center, and definitely not as sharp in the corners, the 50mm has higher average resolution than the 58mm. What's up with that?

Geeky What's Up With That Answer You Might Want to Skip

The answer is some interesting field curvature. When we run Imatest, we assure best possible focus (by focus bracketing-- taking images on either side of what appears to be the best focus - then keeping the image with the best focus and therefore highest values). The image I use is the best center focus (because I think that has the most real-world application) and then accepting the numbers for the other 24 points on that image. Some testers take the best value for each point from all of their images, so they may use the center value from image 3, the right corner from image 6, the midpoint from image 2, etc.

If a lens has field curvature, as many do, then taking the best value from each image gives you the best the lens could do if focused on that spot. Taking all the values with best center focus (what I do) makes the best values from the corner seem lower, but I think they're more real-world, since most of the time I'm focusing on something near the center. Neither one is right nor wrong, it's just two different ways to do it. We could do both, but that makes for some long articles so most people do one or the other.

Anyway, the 58mm lens has an interesting pattern: when the center is in best focus, the corners are also in best focus, but the middle area is a bit out of focus and hence less sharp on Imatest. The Imatest print out from one of the test shots probably will explain it.


Typical Imatest image of 58mm at f/1.4


There's sort of a 'moustache' field curvature at this shooting distance with the lens wide open. Compare it to one of the 50mm Imatest graphs at f/1.4, which shows the more typical Nikon 'just falling off in the far corners' pattern.


Imatest data for 50mm at f/1.4


I mention this because if I had ignored the field curvature and used the best numbers at each point from any image, the 58mm average would have been higher. By f/2.8, though, the field of best focus has straightened out completely and the point is moot. It also seems that the field curvature isn't so significant at infinity, at least according to our optical bench (I've attached an optical bench cut as an addendum showing how flat the focus field is at infinity, but I'm not going to compare optical bench results yet. I've only got optical bench data on a few hundred copies compared to thousands of Imatest results, so I'm not comfortable writing about it yet.)

As an aside I would guess this tradeoff was selected by design to keep corner aberrations low, which is one of the points Nikon's marketing emphasizes. It may also be a positive if this is used as a portrait lens since areas off center may be more blurred than they would otherwise. But it may also mean using those off center focus points will have a bit of a learning curve. (See, geeky stuff can have some practical use.)

Distortion for both the new 58mm and the 50mm is about the same with 1.7% barrel distortion.

Improvement Stopped Down

One thing I'm trying to do more is include numbers at various apertures for each lens. Here are the results for the 58mm as we stop down:

Ctr MTF50 Avg MTF50 Corner MTF50

Compared to the results for the 50mm below.

Ctr MTF50 Avg MTF50 Corner MTF50

Both sharpen up nicely and steadily as you stop down (not that a lot of people are going to shoot these stopped down) through f/5.6. There are some minor differences between them, but nothing dramatic.


As I said when I started, Nikon's marketing blurbs on this lens "virtually no sagittal coma or light falloff" are not things I tested in the optical resolution lab. For many people (OK, for some people), those are the things that will make them want this lens at any price.

Optically, it has excellent resolution. It's a bit better than the 50mm f/1.4 G. I'm not saying the 58mm is overpriced; it's about the same price as the 85mm or 35mm f/1.4G lenses. I am saying that this test makes the 50mm f/1.4 G look like quite a bargain.

Roger Cicala


November, 2013

Addendum: Optical Bench Results

If you aren't familiar with the differences between optical bench and computer image analysis (like Imatest) results, you can read the geeky stuff HERE.

I'm including an example printout of one of the 58mm lenses because it shows there really isn't much field curvature at infinity.The optical bench takes multiple readings, changing the focus in 5 microns increments. Looking at the data across the lens field we get a look at the field curvature at infinity. For this lens the best focus across the entire surface of the lens across the entire field was at -30 microns, indicating a flat field at infinity - at least to the limits of this test (+/- 15 degrees from center).



There are a lot of reasons why field curvature would be different at different focusing distances. In fact, it's quite common.

27 Responses to “Nikon 58mm f/1.4 is (hopefully) Not About the Numbers”

Karl Baker said:

would love to see some results for nightime landscapes.

Matthew Saville said:

Moustache field curvature at normal portraiture / photojournalism distances? EPIC FAIL NIKON.

I'm frankly much, MUCH more excited to get my hands on the sexy looking new 50mm f/1.8 G AI-eSque...

Craig said:

Roger, could you do one more test on each lens? Something along the lines of shooting a Christmas tree, with lights going from top of frame to bottom and into at least two corners. BLUR the shot, as if the subject were a few feet in front of the tree so the tree and lights are out of focus. This should give some nice, big, round light globes for each light. And it should tell us easily if those lights are nice and round at the edges as well as the center? I've got shots like this with other lenses, and the lights are round in the center and oval at the edges.

pieter kers said:

Having seen sample images at Diglloyd.com it seems the lens is made for medium portraits - not closer- and for night shots. At 1.4 the images at night show far more clarity than i can achieve with my regular 50mm 1,4G lens.
Also the flat field of sharpness at ( near ) infinity and color correction is obvious.
Question: can you measure the result of the coating in the lab? For me T* and Nanocoating are really making the difference. I would have liked Nikon to make the original 1.4G 50mm a bit more expensive and add some nano here and there..

LeFred said:

This moustache field curvature reminds me of the Sigma 8-16mm. It can create strange effects sometimes. When the curvature of your subject is opposite to the curvature of the field, you get blurry areas in your image.

derek said:

amazing test , thanks always for your hard work.
but how does this Nikon 58mm f1.4G compare against the Canon 50mm f1.2L?
I want to buy either of these above(I use both Canon, Nikon and NEX).

Greg G said:

My rental of this lens from you arrives tomorrow. I was attracted to the look of the few images posted so far on line and Saturday night I'm off to photograph my friend's band on stage at a dimmly lit local watering hole. I'll probably also spend some time with the 50/1.4D on board to compare a little (although the D isn't quite as good as the G you used here at the apertures I'll be using). I'm not so concerned about terminal sharpness; rock 'n' roll isn't about perfect. If the lens is smooth, prduces great OOF areas, handles those flare inducing stage lights better, and works on the pre-gig band pics I'll take then I'll be happy. Whether $1700.00 happy, remains to be seen. Thanks to this work up, though, I'll try to stop down a bit on the pre-gig promo shots.

ginsbu said:

Tim & Roger-

Photozone's recent test of the Fuji 27mm f/2.8 gives their MTF measures both ways: http://www.photozone.de/fuji_x/855-fuji27f28?start=1
They normally use what Roger calls the ‘best focus at that point’ method, but here they've included a chart generated according to the ‘all values from one image’ method to illustrate this lens's field curvature. That gives a nice illustration of how much field curvature can cause their resolution numbers to vary. It also serves as fair warning that the particular testing method used can make a lens appear to perform quite differently.

Personally, I'd like to see resolution numbers presented according to both methods in an in depth review—and to include an analysis of field curvature at other focus distances, including at infinity as Roger has done here. But that's a big job, as Roger points out. Since no lens review site does this that I'm aware of, I think we ought to be wary of over-reliance on any of them in evaluating lenses. At a minimum, I'd like to see testing sites be more up-front about there chosen testing methods—and Roger is perhaps the best in this regard.

Thanks for another informative test, Roger. Are you planning to publish a test of the new Olympus 12-40/2.8? I'm sure many are eager to see how it compares with the Panasonic 12-35 that you've tested previously!

KnightPhoto said:

The more I read about the Nikon 58mm f/1.4 lens the more "mouth-watering excitement" I get about it. DigLloyd has got > 10 pages and growing of reviews, aperture series, and comparisons with the NOCT etc. up and wow does it look great! The rendering and properties look very very good to me and the bokeh is marvellous!

I don't think this is the jaw-dropping resolution king that the Zeiss Otus is going to be, the Nikon is not a clinical sharpness monster. The Nikon is a complex design selected for a number of very specific properties and is so doing tolerating some focus shift and field curvature intentionally to favour it's selected design properties. The resultant images speak for themselves on direct comparison to the NOCT (and Voightlander).

LensRentals Employee

Roger Cicala said:

Tim, if I understand your question correctly we sort of do that: we take a number of images at slightly different focus points. Our protocol is choose the best center focus and use the values at all points across the frame using that shot. Others look at all of the various focuses and for each point across the image choose the best value -- the one that is best focused at that point.

Both methods give you useful information. The 'best focus at that point' shows you how well the lens can possibly perform at that point, as if you had chosen the AF point at that location to frame the picture (well, assuming AF works which it often does). The 'all values from one image' method gives you an idea of how a real picture would look if you focused on a centered subject.

Reality is if I posted all 32 images for one lens, then test a dozen lenses -- well, you get the picture.

Joachim / CH said:

A reference point can't be a frame. If the lens is decenterred, you'd get a range of results, but a point gives only one result. There's no point in making a frame a point...

Tim said:

Is it possible to have a set of measurements that does not establish the center point as a reference point for focus calibration, but that the midframe is used as a reference point? Are these lens measurements attempting to measure *maximum* sharpness achievable of mid and outer regions *after* focus *per* region, or sharpness of outer regions upon establishing center focus?

Jim said:

Trading on their name? Hmmm. This lens with the penumbra of the almighty NOCT, and then the Df?

Two words come to mind.... Harley Davidson.

Carol said:

I bet this would make a killer crop portrait lens, except of course you are paying for full-frame coverage. For full-frame, seems an odd focal length?

Joe said:

Not sure what Nikon is expecting out of this lens. At the price point, are they expecting lots of sales, or is it just to pair with specific cameras (the df)?

Is this a case where Nikon is trading more on it's name than it's ability?

Graxxor said:

Call me a conspiricy theorist, but I could swear that this lens almost looks as if it is gaming the benchmarks by being sharp in the areas where most lenses are tested for sharpness.

Stefan said:

As per Richard's Point - thank you, this was eye opening - why? my initial reaction was similar, not my lens.

YET, thinking about composition and the-rule-of-third - this lens might support this wonderfully with center-and-edge unsharpness once you select a focus point in the rule-of-third patches.

This is a request to Roger: May I ask you to reprint that Imatest chart with proper focus where the above chart Shows worst? I'm keen on seeing the curvature then.

Fazal Majid said:

Perhaps the lack of interest is because the über-premium normal prime aficionados on the Nikon system are waiting for the Zeiss Otus 55mm f/1.4 Distagon...

Mark Turner said:

Thanks for a great review Roger.

The most important thing is never use "focus and recompose" with a lens with known field curvature. That's a sure way to always front or back focus.

Field curvature doesn't mean that the "moat" around the center is necessarily poor resolution, it may just be focusing on a different distance than the center. In a 3D world, this may not be a problem; if you focus on an eye with an upper right focal point with a 50mm lens and you are 10' away to get a half-body portrait (or head-n-shoulders on a crop camera), then what's in the center of your frame, and does it matter if the center focuses a little near or far? My college notes say field curvature gets traded with astigmatism, that you cannot totally solve both (the old Canon Lensworks chapter 10 alludes to this in a confusing way).

Joachim / CH said:

Any reason you put an old test on your blog? November 2 0 1 2 is not that brandnew :)

Anyway, thanks for a helpful analysis, like always. Personally - and without good logical reasons - I suspect, 3rd party lens developers could come close to that Nikkor if they'd be told "no problem to exceed the target price a bit". I'd love to see a Sigma with that FL and that aperture - but only after they offer a nice 85/1.4. For some reasons, I don't see the Nikkor 85/1.4 worth the big money they are asking for. Okay, different subject.

After all this blurb about pure photography and that lens as well I'm a bit disappointed about their results.

LensRentals Employee

Roger Cicala said:

Hi Zig,

You are correct - I'm probably the only person that understood what I meant to say. How's this:

On the optical bench we set best center focus, then the bench makes each measurement in 20 steps of 5 microns each, 10 behind and 10 in front of our best center focus.

We then look at the readouts across the entire lens at each step.

If one 'step' is the sharpest for every tested position on the lens, then the field is very flat, which is the case here.


Siegfried said:

Hi, Roger!

I'm stuck at the last sentence which is 'With the optical bench we focus very 5 microns at every position and can determine the best focus at each Here the best focus was at -30 microns across the entire surface of the lens, indicating a reasonably flat field at infinity [...]' - not that I want to be the grammar nazi (yes, there's either a full stop or a semicolon missing; and I don't think you wanted to promote numeral 5 to an adjective or a participle like you recently did with some noun which got promoted to a verb there in the fabulous geekcon post), but I'm not sure I completely, fully and entirely understand the process described. Do you move the lens forwards and backwards in 5 um steps and for each z-position you tilt it here and there (where here and there stand for X- and Y-axes) to get the resolution spec for the given angular position?


Pardon me omitting that essential 'thanks for another sexy post, great as usual' part - I take it you might have got a little bit tired of those compliments. And I wouldn't be checking your blog on a daily basis (when I do not do that I do have a serious excuse like dismantled router which went into RAM upgrade... which finally turned out into a new router and a discovery that an air gun can not only desolder chips off a PCB but also blow them away) if you didn't blog greatly.

KeithB said:

What is up with the funky lens shade? I thought that "petal" shades were for wide angle zooms to not block the corners at wide angles. Is that really a problem at 58 mm?

Richard said:

On the other hand, you point out a weakness in my technique—the way I shoot I don't have to crop, but I wind up using a less good part of the lens so I'm throwing away part of what I paid for, i.e. that center sharpness. I'm going to try it your way for a while and see if I can make it work.

LensRentals Employee

Roger Cicala said:

Richard, you read it exactly right. I think wide open at that distance you'd find the image center a bit out of focus.

And a very good point: I tend to shoot centered and crop to effect because I'm all into AF accuracy and center points are generally more accurate. But probably more people do it your way.


Richard said:

Thanks as always.....An interesting insight into how you photograph people; I by contrast NEVER use the center of the lens for people. Whether it's vert or horz, I always place one of the upper left or right focus points on the pupil of the subject's eye--in other words I'm using the outer third of the lens for what is the most important feature. And I shoot like this for most things—rule of thirds and all that.

Sounds like this lens would not work for me. Or am I misunderstanding your analysis?

(And as a side note on use cases, I shoot a lot of food with my 1.4D and close up I'm always using f/5.6 and f/8 so I can get enough depth of field).

James Scholz said:

Thanks, Roger. In my own simple test of one copy of each at 15 feet the 58 looked better wide open, but by 5.6 there was focus shift or field curvature moving sharpness closer to the camera on the new lens.
I look forward to more info on this lens, and the soon to be released Zeiss.

Leave a Reply