My favorite DX portrait lens is an FX lens

My favorite Nikkor lens? I like them all for something. That something can be more rational. Or more emotional, like with the Nikkor 50 mm 1.4G. This blog entry talks about technical minutiae first. Then it goes into cons and pros.

The technical side: Designs compared

A lens with just one element cannot change focus, and even if the focus by chance is correct, it would get it right for a single wavelength only, meaning one spectral color. We want to change the focus distance, while getting all colors sharp in a flat projection on the film or sensor, without much sharpness or brightness falloff towards the edges. This is done with additional elements which create optics much more complex than a single element. In the end it still functions a lot like a magnifier glass, but corrected for good image reproduction.

Making fast lenses complicates the design as it has to process light rays from more angles. Manufacturing costs set a practical limit.

The standard Nikon 50 mm option for SLR cameras is the 1.8G version. Affordable, lightweight, sharp. It has an aspherical lens element, which allows to perform multiple corrections by using a more complex – and harder to manufacture – surface.

The 1.4G is much more expensive and just a bit faster. It is Nikon’s cheapest f/1.4 option within the G-series. Unlike the 1.8, it has as no aspherical lens element! Just an additional spherical element. How can that even work, then?

Here is the trick: A lens element which is flat on one side is easier to make, but gives up potential for optical correction. Both 50 mm designs use a couple of such elements which are flat on one side, but the 1.4 has more elements which are spherical on both sides. These additional curvatures, and the one additional lens element, allow to get acceptable optical correction.

Older, pre G-series 50 mm 1.4 designs use one lens group less than the G version. I don’t know which, if any, optical properties suffer in comparison.

This is the challenge of building a fast lens: 50 mm f/1.8 means the absolute aperture is 50 / 1.8 = 28 mm. For 1.4 the calculation is 50 / 1.4 = 35 mm. (Both results are rounded.) This is a substantially bigger aperture which has to be projected on the film or sensor with acceptable quality. Area is diameter squared, meaning at least the front lens elements are by area much bigger, by some 55%. Also the model is in lower demand and produced in smaller batches, that increases cost per each part. This seems to justify the price of the 1.4G lens from a technical perspective.

But does it deliver?

Why it is not as popular

Already having the DX 35 mm 1.8, I wanted something more different than 50 mm with the same 1.8 aperture. Of course, a 50 mm lens with 1.8 also provides a lot more background blur compared to 35 mm because 50 mm is longer. But then I wanted something even more different and ordered the 1.4 version.

Then put it on, took a couple of test shots wide open. And liked the result. Close inspection revealed several issues, but I was already sold.

A couple of weeks later I took the lens out to photograph a historic Nuremberg cemetery and discovered more optical issues. The post-processing for wide-aperture shots took more time than with other prime lenses. I understand why this lens is not often recommend. But still like the overall look.

Color fringes

If used wide open in broad daylight, or at night pointed at bright lights, color fringes are very prominent. Stopping down to f/2 helps to makes it manageable in post. Under direct sunlight, f/1.4 is unusable for me anyway because the fastest shutter of D7500 is 1/8000s, resulting in over-exposure. Again f/2 helps.

In the night, car lights shot at 1.4 get a thick blue ring. Correcting that in post involves a lot of work to get natural-looking results. Daylight shots which show blue, or green fringes also need quite some time to remove those without leaving visible artifacts. Global image adjustment doesn’t do it, local editing is required.

Sharpness

Color fringes rarely appear shooting on cloudy days, allowing to use f/1.4 under those conditions? Only if! Images taken with f/1.4, even if in perfect focus, come out visibly soft. Digital sharpening can get a lot of detail back. It appears that the lens is technically not soft, it just renders fine detail in low contrast. An aperture of f/2 gets some improvement, f/2.8 is visibly better, f/5.6 seems optimal.

At least in the center. The edge part of the image get softer again. Also some small color fringes, while quite under control at f/5.6, can be there. And this is on a DX sensor which cuts off the FX corners!

Used with an FX camera and a typical resolution of 24 megapixels, the issues shift. Even f/1.4 provides almost acceptable details now. Color fringes, counted in pixels, are thinner and thus less an issue. The absolute thickness does not change of course but the overall frame is bigger on FX, so compared to the final image, color fringes are smaller.

In other words, those results are created by the different pixel pitch of the FX sensor used in the D750. While the high pixel density of modern DX sensors magnifies all the lens issues.

As the frame on an FX camera is larger, one is much more flexible and it is possible to get more background blur because one can move closer to the subject in the foreground. Sharpness and fringing issues are visibly worse looking at the FX-frame corners though. Digital post processing does help, and is even required when used wide open, in order to get proper results if used at 1.4 aperture. Again, the lens performance peaks at f/5.6.

Long story short, 1.4G comes with issues on any camera, but using a current DX camera almost requires you to stop down a bit.

Bokeh

Used wide open, bokeh disk edges are harsh and distracting. Stopped down to f/2, they get a lot softer. We see a pattern: 1.4 is offered but quality improves visibly with f/2.

Looked at it objectively, we can conclude that the 50 mm 1.4G lens should be used with an f-stop of 2 or higher. Why then pay a premium for the f/1.4 aperture?

First, the 1.4G model has nine aperture blades, the 1.8G lens only seven, resulting in a somewhat angular-shaped disks when the 1.8G lens is stopped down. The 1.4 lens on the other hand can be stopped down while keeping an almost round shape. The difference is not drastic and not in itself a reason to go for the 1.4G lens.

And I have to admit that the open-aperture performance of the 1.8G version is really good. Still, if you compare bokeh at same aperture, the 1.4 lens is a little bit better.

When light is needed

Shooting portraits in a quite dark room is possible thanks to being a 1.4 lens. All the issues when using the lens wide open are there, but iso is low and motion blur kept in check. It is a trade-off, but a trade-off which is not available with an 1.8 lens.

Looking at other, more high-end 50 mm 1.4 lenses, it is clear that the comparatively cheap and small Nikkor version has its limits. It is an 1.4 nonetheless and its construction with relatively few lens elements means that little light is lost. It might not be the best 50 mm 1.4 lens, but at least it is bright!

However in closer range, the depth of field is too shallow with 1.4, one has to stop down in order to have enough of the frame in acceptable focus.

Looking at results

With the right settings I get a creamy appearance with a smooth transition from in-focus parts to out-of-focus parts. And the lens issues gracefully increase towards corners, leading the attention to the best part, the center. NOT using aspherical elements seemingly can be a weird advantage sometimes.

My 50 mm lens took some beating, once I ran against a wall, breaking one of the lens hood brackets. When the lens later dropped to ground by mistake, focus broke (both auto and manual). I ordered a new 50 1.4G, though by then I got more realistic about its performance. For example, colors are not spectacular. But for me, this lens provides the balance of performance and traditional photographic output which I like. And if needed it is very fast.

Looking at the gear

From the outer appearance, I think this lens looks like a good fit the D7500. Judged objectively this combination is not optimal because the crop-sensor uses less than half of the projected area of an FX lens.

On the D5600 the 50 1.4 lens makes a good impression, too, judged from the outside looks. On a borrowed fullframe camera, the D750, the lens appears almost tiny.

Your mileage may vary, but I like the look and the balance with this lens on each of my Nikons DX cameras. Of course it well with the Nikon F65 which I bought used. The lens barrel surface feel is nice, the focus ring turns smooth. Everything feels a bit more serious than the 1.8G version. For me, such things help to justify the price.

The front lens element is quite big, the rear lens, too. Seemingly there is a lot of glass in there. This also helps to enjoy the purchase, as the lens just radiates “I gather a lot of light”. And then the experience when using the aperture dial. I was used to have 2.8 or 1.8 as smallest number depending on the prime lens. Being able to go lower feels great. It might not help the photo directly if one enjoys toying around with the gear. But liking the gear for whatever reason means to use it more often. This can turn into practice and eventually yields a couple of hopefully useful photos.

Autofocus compromises

Both automatic and manual focus is slower than with the 1.8G. This should improve precision. While not always perfect when using viewfinder autofocus shooting wide open, the lens usually does get in focus. For consistently perfect results, I use the contrast-based focus in live-view used with enlarged preview, if I can. Because of the low-contrast rendering of fine detail one still would do some work in post.

Not sure about differences within the series. The new copy I got seems to nail focus better, but that could also be because I handle it more carefully.

In most cases I work with the viewfinder and find the focus precision good, and the speed good enough. I use the lens for outdoor portraits while the model is not holding still, and get nice results with the camera’s AF-C mode. In other words, yes the Eye-AF of recent cameras might get higher yield but with a bit of practice I feel confident to use this lens for portraits when it matters.

What about premium lenses?

Nikon marks its best lenses with a gold-colored outside ring. The 50 mm 1.4G has no such gold-colored mark. But the NPS terms (Nikon Professional Services) still consider it a pro lens.

The DX 16-80 mm does have the golden honor, and is also part of the NPS terms, but only as a group-2 item (in this nomenclature, that means lesser). The 16-80 mm is very good, but a blog entry would be short: If quality is not already excellent, stop down a bit, but unlike standard DX zoom lenses you can expect good results even wide open. Color fringes are well under control. You get many switches on the lens, and VR works well. One can do almost everything with the 16-80, even portraits because the bokeh is nicely soft. Color rendering is extraordinary. Sadly, the focus ring has a bit of play and the exterior, while okay, does not feel pro-grade like the 50 mm 1.4G.

The 50 mm 1.4G is roughly half the price of the 16-80, but of course much more limited in application. For use on a DX camera the 50 mm 1.4 seems overpriced. Especially because for less than half of the 1.4G cost, one can get the 1.8G version which is a bit slower but sharper over the entire frame and it has less issues with color fringes. The 50 mm 1.8 is an entry-level item in the FX world, but still solid in DX terms as it of course comes with a metal mount, dust seal, distance indicator and MF switch, all missing on current DX kit lenses.

Using the 50 mm 1.4 with a modern high-resolution DX camera stretches the capability of that design, while wasting half of its image area and thus image information. Yes the center crop gets the best part but optical limits are still there. Having more area on FX means more information gets captured.

Long story short: This lens is good but using it correctly requires practice.

Personal preference

Okay nine aperture blades, like a portrait lens should have. Otherwise nothing special. A standard “silent” (actually, low-noise) wave motor like any AF-S lens. But no ED, nor aspherical lens elements. No inside, nor rear focus, instead the front element moves. No nano coating, just standard stuff. No VR of course.

With f/2, this lens is able to get nice portraits: Optical issues are there but dialed back so that they even contribute to the image giving a classic photographic look. The depth of field is where I need it: Shallow, but not to the extreme. Bokeh is soft. Image edges might be a bit blurry and darker as some vignette is still there but that also helps to get a traditional appearance. For more sharpness I stop down, for more light I stop up.

Just cranking up the aperture and expecting perfect results – not happening. It took some time to discover all the weak points and how to avoid them – or how to apply them just enough to avoid a sterile, digital look.

Using the lens

I couldn’t use photos of a social event for this blog and I picked this landscape or rather cloudscape. The narrow field makes landscape photos more difficult, but can also serve as a force of creativity. It is not a very useful lendscape choice of course.

For events, the fixed and narrow field of view can also be a limitation. However in a good way: When others use their cellphones, usually with a wide-angle lens, they get the standard social-media results. Using a 50 m lens on a DX camera all the time gives the resulting image gallery a different look. Compared to the other photos it stands out. Especially as one gets real background blur instead of computational blur.

So far, my best work (non-commercial as I am not a pro) was when I brought this lens only. No backup with a wider angle. Instead having to walk more.

In post, I sometimes don’t use lens correction (geometric distortion, vignette) for Raw files taken with this lens, as it looks more natural than a digitally applied filter.

Some aperture recommendations

I prefer 5.6 for landscape or architecture or everything where I need the in-focus part tack-sharp. Also for close-ups of flowers, as it gives a reasonable depth of field.

For single portraits, f/2 seems to be optimal in many cases. Group portraits often benefit from 2.8, as it increases the yield where everyone is fully in focus. With 50 mm, one still gets some background blur if the background is far enough.

The lens has no VR, using short exposure times helps. Even for still objects I use 1/160 or less, for objects in motion 1/500 or less. For events, I sometimes use 1/250 to keep iso low and accept some unusable results with too much motion blur.

In-door at night, I push my luck with 1/125 needing the light because f/2 is the widest practical aperture if two or more guests are in the photo.

Talking about 1.4. I try to avoid it as it can result it busy bokeh with harsh edges, f/2 is much better.

By mistake I used 1.4 for an outside portrait shooting in snow. It was quite some work to smoothen the background and even more work to address color fringes in post, and except the very few parts which were fully in focus, the images have a dream-like haze. Luck had it that the hazy appearance turned out to be atmospheric.

For a 12-month-in-a-year portrait project I used the DX 40 mm lens and the DX 16-80 mm lens for the first months, and only the 50 mm 1.4G since then. Using it correctly requires practice but then it gets me results which are to my taste.

 

Should you buy FX lenses for a DX camera?

I am an enthusiastic amateur, buying my first point and shoot in 1995, went digital in 2003 with a compact “Coolpix” and take a lot of photos since 2015. Currently I use the Fujifilm X100F and two Nikon DSLRs, both having a DX (APS-C, crop) sensor.

In this blog I want to dispel the notion that using a fullframe lens on a crop sensor is generally a smart choice.

There are cases where a lens you want is available only for fullframe (in Nikon terms, FX) so that is your only choice of course. Another reason might that be you just want to get the FX variant. For example an AF-P 70-300 lens is available both as DX as well as FX version. The latter is a tiny bit faster on the long end, and it has switches on the lens which the DX version omits. Does it worth it the additional cost and weight? Only you can decide.

Are there other upsides? Some reasons can be seen quite often.

  1. A lens performs best in the center. With DX, you use the center of a fullframe lens, which is the best part to use.

    True – but you still crop off parts of the projected image and thus, information. Even if the cropped border is not as good as the center, having access to it is better than a forced crop. If the lens is used wide open, even the center of the lens does not resolve the DX sensor pixel pitch. No lens does, but especially fast lenses will not fully resolve a modern DX sensor. If stopped down a bit, lens performance improves overall including in in the corners, making it even more desirable to get those image parts instead of cropping them off. In short, cropping a fullframe lens always means to lose image information.

    And if you compare performance stopped-down, DX lenses are optimized for performance over the DX frame. So, no general advantage with FX glass on DX bodies here.

  2. FX lenses offer higher quality

    That is usually true. If you talk about build quality. Regarding optical performance, a recent DX lens – even cheaper ones – usually performs very well when compared to older FX lenses used on DX. No wonder, those FX lenses were designed for then-common FX-sensor pixel pitch. Today’s DX sensor pixel pitch is much higher, with the DX lens design taking that into account.

  3. Buy FX lenses now because if you switch to an FX camera later, you already have lenses.

    True again. But if you want to go FX, why not buy the body now? If you buy FX lenses for a possible later switch, you forgo the DX advantage of a lightweight and more affordable system. When you finally switch, your fullframe lenses might be outdated already. And you might never switch to FX anyway …

Consider the lens performance wide open

The fullframe Nikkor AF-S 50 mm 1.4G lens is one of the betst portrait options for DX. But if used wide open on a modern DX camera, the center crop magnifies the sharpness issues and the problems with chromatic aberration.

On the other hand with a fullfame camera like D750, the lens performance is quite good at f/1.4. The bokeh is a bit harsh but otherwise, you can crank it up. Because the optical issues are rather small relative to the final frame.

To get good resolution on DX, one has to stop down a bit. You are still faster than with a zoom lens set to 50 mm, but a zoom usually has optical stabilization in its favor. Used correctly, the 50 mm 1.4 lens is an option for DX but usage wide open on DX is more difficult than on FX.

The Fujifilm world recently rejoiced because of the Fujinon 50 mm 1.0 lens. That lens has just marginally more background blur than 85 mm 1.8 … If you are crazy about maximum background blur, using a smaller sensor is an unnecessary hurdle. Using a fullframe camera instead, and a couple of f/1.8 lenses, you could even save money!

Iso performance drops with further cropping

If you have an already noisy image, magnifying it also amplifies the noise. There might be cases where you just have to crop to get the frame you want because an affordable lens is only so long – but generally you want a tele lens with enough optical reach to avoid crop in post.

Take the famous 70-200 2.8 lens, used on DX. How does it compare to the kit-type DX zoom, AF-P 70-300? On the short end of 70 mm, the DX lens only gives us f/4.5. With 2.8, the FX pro lens is some stop faster.

On the long end of the 70-200 we still get f/2.8, while the DX zoom only allows 5.3 for 200 mm. The FX lens is nearly two stops faster now.

On the long end of the 70-300, we only get f/6.3. In order to get the same field of view, we have to crop/magnify an image taken by the 70-200 mm lens at full length in post. While still taken at f/2.8, through that crop the noise performance suffers by some stop. That means, in terms of noise the FX lens is only about 1 effective stop faster if cropped to 300 mm equivalent.

If you compare the price and weight of these lenses, I would like to know how many DX users really need the pro-grade build quality of the 70-200 2.8. Paying the full price of that FX lens, carrying the weight, to use only half of the image it projects – why not get an FX camera instead and if money or weight is a concern, opt for the 70-200 f/4? With the larger sensor you can stop down this much and still have iso performance comparable to 2.8 on DX. Because the DX crop increases noise by roughly 1 stop, as iso is typically measured for the whole image, not per pixel.

But …

There might be situations where the 70-200 2.8 is just the best choice for a DX camera. However those situations seem quite specific to me. 200 mm f/2.8 has a depth of field so thin, that the focus has to be extremely precise to get the sharpness where you need it. The depth of field might still be too shallow, so that you stop down. Then the DX 70-300 might be the better overall option. It is slower but has more optical reach.

What about the fullframe 200-500 5.6? There is no DX alternative, that FX lens is the only choice for this length. And 24-70 2.8? The DX 16-80 is not as fast on the long end, but allows much short settings giving you ultra wide angle on DX, whereas 24 mm on DX is only moderately wide.

What about the 14-24 2.8 for DX? It is a bit wider than the 16-80 but the 1.7x zoom of 14-24 costs a lot more than the 5x zoom of 16-80 which even comes with VR.

Is there any reason to get an FX lens for a DX camera?

The DX lens lineup leaves something to desire. FX lenses can fill those gaps, for a premium. And you might get less from those lenses than you hoped.

I have a couple of fullframe lenses including the 50 mm 1.4. Stopped down to f/2 it is my secret weapon for portraits. 50 mm f/2 has a depth of field still shallow enough for portraits.

What about wide-angle prime lenses which are still wide angle if put on DX? Compared to a wide zoom lens, you get faster glass with those FX lenses, but forgo optical stabilization. Using a DX zoom lens with VR costs less and weights less, yet normally performs as well or better. Exceptions would include wide-angle action shots where you rely on fast shutter speeds. But if you buy those wide-angle FX lenses why would you use them with a camera cropping off a lot of that wide-angle image output you paid dearly?

There are of course cases like certain types of wildlife where you rely on super-telephoto. Even with the longest lenses available, you might only use cropped images. Getting a crop-sensor camera for that job seems more efficient. That is still not an argument per se to use FX lenses on DX glass, that is an argument to use the longest lens – though in this world, those are always made for fullframe.

Conclusion

An extreme FX lens becomes less extreme if mounted on DX: You crop image area, and if you use the lens wide open it might be not as sharp as you assumed from reading FX-based reviews.

If you determine that for some reason or no reason you want to get an FX lens, of course, get it! At least, I got a couple … but it is not a magic upgrade for your camera.

The true fullframe advantage over crop

In low light, the larger sensor shows less noise, so it is iso performance, right? Wrong.

The iso performance of a sensor is measured for the full sensor area. Other things being equal, a sensor with higher resolution has more noise per pixel but since each pixel – hence, its contribution to noise – is smaller, not more noise overall. For an iso rating of a sensor, the single pixel does not count, the whole image counts.

Other things being equal again, iso 100 on a fullframe sensor looks better than iso 100 on a crop sensor. So far it looks reasonable: Larger sensors = less noise. But if you use a lens on each camera yielding the same field of view and depth of field, you have to use longer focal lengths on the larger sensor and thus stop down further on the larger sensor to keep the same depth of field. In low light, when iso is above the camera’s base iso, you have to use a higher iso value to compensate. It still looks as good as the lower iso value on the crop sensor. But you don’t have an iso advantage in terms of lower noise on the larger sensor in this situation.

Let’s go deeper. If the fullframe user uses the lower f-stop, too, but also the smaller focal length like on the crop sensor, and crops the image in post to get the same field of view, the remaining image pixels get enlarged and so the noise gets amplified as well. Still no noise advantage!

The only option to get less noise with the larger sensor is if you can live with a shallower depth of field.

Again, a fullframe sensor only gets you less noise if you can live with a shallow depth of field, or if you are at very low iso values already. In the latter case, today’s crop sensors are quite good though, more quality is usually not needed for an amateur.

Is sensor size useless then? Of course not

Other things being equal, optical issues of a lens have a certain extent. The larger the sensor, the smaller the the lens issue compared to the final image. Meaning you get more undistorted image information just through the larger sensor area.

On top of that, with the larger area you use longer focal lengths and stop down further to get some depth of field. Stopped down a bit (but not too much) the lens performance will improve, getting you even higher resolution. That is the real advantage of fullframe versus crop! You get your lenses more often to the peak performance, and the larger area captures more information even if you shoot wide open.

Another reason for fullframe: At least in the Nikon world you get the best lenses here. The DX range consists mostly of consumer-grade products, and while you can mount FX lenses on a DX camera, you cannot fully use their performance with a crop sensor. I am told it is different in the Fujifilm world where a lot of APS-C lenses are made to professional standards.

Do you need fullframe? I personally look for a reason to finally give in. But honestly? 1.5x crop is just some stop in terms of light and thus iso performance. Meaning yes it can be noticeable if you look at individual pixels – but since when has a good photograph to be be pixel perfect? If you think it does, why being satisfied with “full”frame, which in film terms is amateurish small frame? Why not a larger sensor? Such cameras are available, look at Pentax and Fujifilm …

Conclusion

I am a crop-sensor user but of course admit that systems with larger sensors can offer better technical quality. But the common idea that this advantage would be less noise in low light only applies to situations where not a lot of depth of field is required.