AF Fisheye Nikkor 16 mm f/2.8 D

This page discusses the Nikon AF Fisheye Nikkor 16 mm f/2.8 D, introduced in 1993 and discontinued in 2020. AI and AI-S versions of a comparable lens were available before this lens, but with different optics and with an f/3.5 lens speed. This 16 mm f/2.8 D was not preceded by a non-D model. No AF‑S successor of this lens was ever introduced, either. This lens was replaced by the AF‑S Fisheye Nikkor 8‑15mm f/3.5‑4.5E ED zoom.

This 16 mm fisheye completely covers a full frame sensor with a diagonal angle of view of 180°. The AI and AI-S models had a coverage of only 170°.

This lens is relatively frequent on eBay and not so expensive, barring the occasional fantasy prices.

This 16 mm can be used on APS‑C cameras (both DSLRs and, via a lens adapter, Z mirrorless), and on this format it gives a diagonal angle of view of 107°, equivalent to an ultra‑wideangle but not rectilinear (i.e. only radial straight lines on the subject are rendered as straight lines in an image).

This lens was made in Japan.

Features

The 16 mm is small and compact (57 mm long from mounting flange, including the lens shade, and 63 mm wide), and quite lightweight (290 g). It makes virtually no perceptible difference when added to a medium camera backpack. If used correctly, this lens produces a more muted "fisheye look" than some fisheye lenses of other brands (see e.g. stuckincustoms.com).

Its images can be de‑fished in post‑processing and subsequently cropped (see e.g. epaperpress.com) to produce the equivalent of an image produced by a rectilinear ultra‑wideangle.

 

The lens shade of the 16 mm is an integral part of the lens barrel and cannot be removed. The effectiveness of this petal lens shade is limited, considering the diagonal angle of view of 180° on full frame.

There is no AF/MF selector on this lens. AF had to be disabled on legacy SLR and DSLR cameras when using manual focus. Focus is only manual when adapted to a Nikon Z camera (except with a MonsterAdapter F to Z adapter, which I do not recommend in its present model because it is expensive and unreliable).

The aperture ring must be set to f/22 and locked with the nearby small slider when this lens is used on a Z camera via an FTZ or FTZ II adapter. This allows the camera to control the lens aperture. The camera also reads data from the lens firmware that lets the camera properly configure in‑camera image stabilization and the automatic correction of vignetting and chromatic aberration.

This lens is not weather sealed.

Optics

This lens uses a moderately complex optical scheme with 8 elements in 5 groups, devoid of aspheric elements and special glass, similar to its AI and AI‑S f/3.5 predecessors, but with better coatings and an improved f/2.8 optical design.

The optical scheme of the 16 mm published at Nikon's site on historical lens development (modified in the above figure), unlike the optical scheme shown in Figure 3, does show one filter mounted at the rear of the lens. The relevance of this is explained below.

The shape of the girdle of a few optical elements differs between the two drawings. Figure 3 probably is closer to the actual shape of the optical elements, ground to remove some weight, while Figure 4 seems to be an optical design document, rather than a manufacturing document.

The two vertical dashes in the optical scheme indicate the position of the lens aperture, which in this lens has 9 blades. The straight horizontal line crossing the optical scheme is the lens optical axis.

Focusing

Focus is not internal, unlike what stated by multiple Web sources. The whole optical assembly moves back and forth less than 2 mm when focusing, because of the short focal length. This can give the (mistaken) impression that this is an internal focus lens. The AF 16 mm, unlike its predecessors, is equipped with CRC (Close Range Correction) by using a floating front optical subassembly (not rear) that moves back and forth less than 1 mm and rotates with the focus ring when focusing (like in the AF 20 mm f/2.8 D). The rest of the optics does not rotate.

Like in most AF‑series lenses, the focus ring is rubber‑coated, while the aperture ring is plain plastic, which contrasts with the crinkle‑painted lens barrel and lens shade. The focusing scale is covered by a transparent plastic window and printed in ft and m. The depth of field marks below the distance scale are only indicative, and on a high‑Megapixel‑count camera like the Nikon Z8 you must stop down by one stop more than indicated by these marks.

The focus ring turns only about a quarter of a turn. The 16 mm focuses down to 0.25 m, i.e. a working distance from subject to the perimeter of the lens shade of roughly 15 cm. This provides a 0.1x subject magnification (360 by 240 mm subject area).

The front element of the 16 mm is convex but not bulbous, and quite thin in its center. The lens barrel is short, and the lens shade has the same diameter of the lens barrel. The focus ring is thin and close to the lens shade. Therefore, you must be careful to keep the left hand away from the lens shade, and especially from the indentations between petals of the lens shade, which correspond to the corners of the frame.

A proprietary plastic front lens cap accompanies the lens and fits by friction onto the lens shade. This lens cap may wear out internally and become too loose. This can be fixed by adding strips of adhesive‑backed velvet inside the perimeter of the lens cap, or a suitable lens cap replacement can be made by 3D printing in a slightly elastic material (preferably by allowing for a layer of adhesive‑backed velvet glued inside the periphery of the cap). A hard, tight plastic cap directly touching the lens shade may wear out the paint on the lens shade.

Naturally, there is no filter mount at the front of the lens, but a round glass filter in a proprietary plastic frame can be clipped into the rear of the lens. Four filters were supplied with the lens, of which one attached to the lens and the rest packaged in a leather pouch. The colored filters in this set are actually not useful in digital photography.

Watch out for this

The 16 mm is designed to always have a filter mounted at the rear of the lens, as part of the optical scheme of the lens, and does not perform optimally without a filter. In particular, without a rear filter this lens cannot focus at infinity (or farther away than about 10 m). When purchasing a second‑hand AF Nikkor 16 mm fisheye, make sure that it comes with a Nikon L37C "skylight" filter mounted at the rear of the lens. Without a filter, focusing exactly at infinity is impossible, especially at f/2.8 or f/4. The general performance of the lens is also slightly worse without a filter.

This filter is occasionally missing from specimens of the 16 mm advertised on eBay. This is visible only if the sellers' pictures include a rear view of the lens without the rear cap. However, some sellers do not provide such a picture. Caveat emptor.

It is possible that a few photographers unaware of the necessity of using a filter on the 16 mm tested this lens without the proper filter mounted at its rear. This may be one of the reasons why the majority of reviewers found this lens good, but a few regarded it as unsatisfactory.

The Nikon user guide of this lens does warn the users to always leave a filter mounted at the rear of the lens.

To remove the rear filter of the 16 mm, keep it pressed in and turn it about a quarter of a turn counterclockwise.

To reattach the filter, insert its bayonet into the rear filter mount of the lens, and while keeping it pressed in (the filter mount is spring-loaded) turn it clockwise until it stops. If it turns only a short distance and stops, you inserted the filter the wrong way. If so, extract the filter, turn it half a turn, reinsert it, and turn it clockwise as far as it goes. It is supposed to turn in its mount by at least a quarter of a turn in order to fit safely.

Nikon did provide alignment marks on the rear lens baffle and on the filter frame, to make it easier to properly align the filter with its mount. However, both the filter frame and its mount are black, and the alignment marks are inconspicuous, unless you know what to look for.

This filter is occasionally available on eBay, but I saw only two, and both were indecently overpriced. Rather than buying a 16 mm that lacks the filter and then trying to buy the filter separately, it is more convenient to make sure you are buying a 16 mm complete of filter.

Practical use

A reasonable holding technique makes use of only the thumb and index finger of the left hand to turn the focus ring. Do not extend the remaining fingers forward of the focus ring. Keep an eye in the viewfinder for stray fingers, as well as the tips of your shoes and the brim of your hat. "Chimping", i.e. holding the camera with your arms partly extended forward and checking the image from a distance on the LCD rear screen, can be a good way to avoid photographing parts of your clothing, especially when shooting in portrait orientation.

Photographing while crouching with one knee on the ground is a common way to include the other knee in the image.

You should not need to mount a camera with the 16 mm on a tripod, unless you are shooting at night or in a very dark place. If you do, carefully examine the viewfinder/LCD screen to make sure you are not photographing a tripod leg or two.

I also suggest that you use the electronic level of the camera to keep the camera truly horizontal, in order to minimize the "fisheye look". Of course, there are circumstances where tilting the camera up or down cannot be avoided, and/or the "fisheye look" is desirable.

If used correctly, this lens produces a more muted "fisheye look" than some fisheye lenses of other brands (see e.g. stuckincustoms.com). These images can be de‑fished in post‑processing and subsequently cropped (see e.g. epaperpress.com), and their optical quality is good across a large central area. De‑fishing without cropping unavoidably does display a loss of resolution near the edges and corners. Some de‑fishing software automatically crops enough of the original image to lessen this problem.

Performance

Infinity focus

The first batch of test images I shot with the 16 mm turned out to be quite blurry when observed as 1:1 pixel crops. At first I thought that the lens (either my specimen, or the lens model in general) perhaps is not that good.

I decided to give the 16 mm the benefit of doubt, and shot a second set of test images.

In my first attempt, I had focused manually with focus peaking enabled and focus peaking sensitivity set to 2 (i.e. normal, the default value). This time, I used precision‑focusing visually by magnifying the center of the frame in live view.

By default, the Z8 magnifies the selected area of live view in four successive steps of increasing magnification. I only used the second magnification step, which turned out to be sufficient.

The results turned out very much better, and more than good enough to publish. The lesson learned is that focus peaking is not precise enough on the Z8 with this lens, if one requires pixel‑peeping the images. It does help a little to set focus peak sensitivity to 1 (low) on the Z8, but live view magnification is far more precise when using a lens of very low focal length.

The center of the test images is a building more than 40 m away. This should be practically at infinity for a lens of such a short focal length. It turns out, instead, that true infinity focus is not located at the ∞ mark of the distance scale of the lens. If the lens is focused with the ∞ mark in correspondence of the focus index, the lens is actually focused past infinity and everything is out of focus. In practice, true infinity focus is reached when the center of the oversized ∞ symbol on the focus scale is aligned with the round dot on the DOF scale (Figure 7). This dot is the infrared focus compensation index.

I don't know whether all 16 mm lenses have their infinity focus calibrated in this way. If your 16 mm is similar to mine in this respect, you must remember this if you are going to use parfocal focusing with this lens. An alternative possibility is that the Nikon FTZ and FTZ II lens adapters I am using with this lens are intentionally designed to be a fraction of a mm too short, so that they allow focusing at infinity with a variety of legacy lenses calibrated at different infinity focus. With a lens of very short focal length, a fraction of a mm translates to a significant change in actual focus.

Tests

Figure 8 shows the test subject, imaged with the 16 mm. The edge of a window frame is partially visible in the upper right and lower right corners. For this test, I shot hand-held images at intervals of one stop. There are small changes in framing from image to image, visible in the 1:1 crops of the edges, but nothing that invalidates the test.

I manually focused as discussed above before starting the test, and did not alter the focus afterward.

Center resolution

Ordinarily, I would stop a resolution test at f/8 or f/11, but with this lens it is interesting to continue stopping down, as discussed below.

Interestingly, at f/11 to f/22, there is no visible loss of center resolution. There is, however a marked amount of "purple haloes" at f/22 along high-contrast edges, visible only on the dark side of such edges.

Of course I am quite convinced that no lens can pretend that diffraction does not exist, and continue to provide the highest resolution that a high‑megapixel sensor can provide even when stopped down to "impossible" apertures like f/16 and f/22. However, this lens just appears to do the impossible.

I do not fully understand the observed behavior, but I believe that it is related to the projection type implemented in this particular fisheye. In particular, if the central region of the image is projected onto the sensor at a lower magnification factor than done by a rectilinear wideangle of the same focal length, diffraction blur is also rendered at a lower absolute size than expected with a rectilinear lens.

If my idea is correct, it should be possible to see a higher diffraction blur in portions of the image rendered with a higher magnification than the image center. For this purpose, we can look at 1:1 crops of the peripheral region. The following figures show 1:1 crops of the extreme left edge of the frame, at the vertical center of the frame. For the sake of brevity, I skipped some of the lens stops.

Why do I show test results at the center of a horizontal edge, and not in the corner where problem should be greatest? The reason is that the corners are typically cropped away during or after de‑fishing, while the center of the edges are often preserved in the de‑fished image.

Peripheral resolution

Since I was hand-holding the camera during this test, the framing of the image changed slightly from shot to shot. While this can be compensated for when preparing crops of the central image region, it cannot be compensated when the extreme edge of the frame must be included in the crop.

The image resolution in the edge crops follows a pattern very different than in the center crops. The edge crop at f/2.8 is blurrier than the corresponding center crop. The f/5.6 crop is very good, just slightly blurrier than in the center. f/11 is visibly blurrier than at the same aperture in the center. f/22 is markedly blurrier than f/11, and also a little underexposed.

This behavior would seem to agree with my idea discussed above. Image details, like the sizes of the windows of the buildings, at the image edges seem proportionally larger than in the center. This should magnify the diffraction blur by a correspondingly larger amount. However, optical aberrations likely also play a role, and typically are higher in the image periphery.

Discussion of resolution tests

The 16 mm proves to be a good lens. All apertures give remarkably good results in the image center, except for a slight blurriness at f/2.8, visible only when pixel peeping, a slightly colder color rendering at f/11 and higher, a hint of purple haloes at f/16, and stronger purple haloes at f/22.

In fact, center resolution is so high at high apertures that this lens appears to defeat diffraction blurring in a wide center area. Likely, this "impossible" result is due to a lower relative magnification (or a longer effective focal length) in the center, compared to the periphery. This magnifies diffraction blur by a lower extent in the center region.

The behavior at the side edges of the frame is different: a little worse than in the center at f/2.8 and f/4, remarkably good at f/5.6, then just a little worse at f/8, worse still at f/11, a larger worsening at f/16, and finally a major worsening at f/22.

As a whole, f/5.6 is the aperture of choice when uniformly high quality is necessary in both center and edges. f/8 and f/11 can be used when center sharpness and a high DOFdepth of field are important, but a little worse edges are tolerable.

Vignetting

There is no visible vignetting at any aperture. If the optics are producing any vignetting, it is completely corrected by the camera's firmware.

An example of de-fishing

For this test I used the automatic de‑fishing available in Adobe Lightroom Classic v15.1. This software comes with a large number of lens profiles available, including one for the 16 mm.

Perhaps because I am using this lens on an FTZ II adapter on a Nikon Z8 camera, Lightroom initially does not recognize this lens by reading the JPG metadata. In the Lens Corrections panel, Lens Profile section, I need to manually tick the checkbox Enable Profile Corrections and select Nikon in the Make drop‑down list. Once this is done, Lightroom recognizes this lens and applies the built-in lens profile corrections to the current image.

The de‑fished image is automatically cropped by Lightroom. By comparing this image with previous tests of my Nikon ultra‑wideangles, I found that the angle of view is wider than the one produced by the AF Nikkor 14 mm f/2.8 D. I am unable to compute the equivalent focal length from the de‑fished image, but my visual estimate is that it is approximately equivalent to a rectilinear 11 or 12 mm.

The de‑fished image has a size of 8256 x 5504 pixels, exactly the same as native Z8 images.

A visual comparison of the original versus de‑fished image shows that the central region of the de‑fished image has been scaled down in both x and y dimensions. On the other hand, the top of the tower is only slightly narrower than the original, but is quite a bit more distant from the top edge of the frame than the original. The same is true of the parked cars near the bottom of the image, which are stretched to twice their actual length but keep their original width. Thus, there are different amounts of compression and stretching, perpendicular to each other, in the peripheral regions of the image. As a whole, radial stretching in the periphery seems to always be larger than compression in its perpendicular direction, while compression is equal in x and y directions in the central regions.

Naturally, I was curious to examine 1:1 crops of the de‑fished image.

The observed stretching indeed does reduce image resolution near the sides of the frame, albeit to a lesser amount than if there was no compression in the center. The center was already very sharp before de‑fishing and, naturally, compression did not affect the visible sharpness. It is worth noting that, within the 900 pixels wide crop of the left edge in Figure 13, the picture becomes a little sharper already 750-800 pixels from the edge.

As a whole, the image is as good as I would expect from an 11 or 12 mm ultra‑wideangle of reasonable price and weight. A remarkably expensive one might do better, but not one that costs 329 € on eBay (and sometimes less). To obtain a picture with better edges and corners, the simplest solution would be cropping the de‑fished image. To match the angle of view of my rectilinear 14 mm, cropping the image from 45.7 Mpixel to approximately 36 Mpixel does the trick. A sharp 36 Mpixel image would still be usable for almost any purpose.

In conclusion, on a long trip, or where I expect to do a lot of walking, I will probably not hesitate to carry the 16 mm in my backpack instead of my AF Nikkor 14 mm f/2.8 D, which is rectilinear but much larger and heavier. The 16 mm is also much less likely to attract attention, for example in candid and street photography. De‑fishing is relatively time‑consuming (unless I can be convinced to learn batch scripting on a platform like Hugin), so the 16 mm is not a perfect replacement in all scenarios for the 14 mm ultra‑wideangle.

Alternatives to the 16 mm

The older AI and AI‑S Nikon Fisheye 16 mm f/3.5 are optically worse than the AF 16 mm f/2.8 D, their coatings are more vulnerable to flare, and they have no corrective floating group to improve image quality at close focus.

I can see no natural alternatives to the AF 16 mm f/2.8 D among Nikkor F full‑frame fisheye lenses. There may be alternatives among third‑party fisheye lenses in Nikon F mount, but I have no direct experience with any of them. In the past, I did find the Samyang 8 mm f/3.5 fisheye very satisfactory on Nikon DX DSLRs. This lens is designed for use on APS‑C sensors or smaller, and does not cover full frame (FX) sensors. Note, however, that Samyang fisheyes use a different projection type than most Nikon fisheyes, including the 16 mm. This produces different visual results and requires a different de‑fishing process. Some software, like Adobe Lightroom Classic, that have a profile for the Nikkor 16 mm lacks profiles for Samyang lenses.

The AF‑S Nikkor 8‑15mm f3.5‑4.5 zoom fisheye is not excessively large, and could be an alternative to the 16 mm fisheye for photographers who use both FX and DX cameras. The possibility of using it as a circular‑image fisheye at 8 mm on full frame, in my opinion, is rarely useful. In addition, this lens, when used as a circular fisheye, produces an image edge that fades to black gradually at the border of the image circle. When using it as a full frame fisheye, care must be taken to zoom in by a sufficient amount to avoid dark image corners.

A slight zooming on DX cameras makes this lens cover the whole DX sensor. For photographers using exclusively DX cameras, a third‑party APS‑C fisheye like the Samyang 8 mm f/3.5 is far cheaper.

Conclusions

On a high‑megapixel camera like the Z8, this lens is remarkably sharp in a broad central image region between f/4 and f/16.

At the image edge, there is a little blurring at f/2.8‑f/4, with the best results at f/5.6. Image quality afterwards degrades gradually from f/8 to f/22, but can be called soft only at f/22.

This lens is easily available on the second‑hand market, and reasonably priced. It is small, compact, and relatively lightweight.

This lens cannot autofocus on the FTZ and FTZ II F‑to‑Z adapters.

The depth‑of‑field marks on the lens barrel are not correct for a high‑megapixel camera like the Z8. You must stop down by one more stop than indicated by these marks.

The infinity mark on the focusing scale of my lens does not correspond to actual infinity focus, and instead focuses the lens past infinity. The dot marking the NIR infinity focus is a good approximation of the actual infinity focus. I don't know whether all specimens of the 16 mm are like this.

This lens is not weather sealed.

De-fishing with Adobe Lightroom Classic produces the equivalent field of view of a rectilinear 11 or 12 mm ultra-wideangle. Cropping the 45.7 Mpixel de-fished image to approximately 36 Mpixel produces an image with sharper periphery and roughly equivalent in field of view to the AF Nikkor 14 mm f/2.8 D.

Manual focus with focus peaking is not sufficiently precise on the Z8. For best focus, magnifying the portion of the image of highest interest in the viewfinder or LCD screen is more reliable.

This lens is designed to always carry a dedicated filter attached at its rear. Some specimens on the second‑hand market, however, lack this filter and cannot focus at infinity.