September 3, 2013

Metabones Nikon F to Fuji X adapters – part III

The Metabones approach to aperture control

Click on any image or chart to see a larger version


Nikon started to introduce G-type lenses in 1996. These lenses come without an aperture ring: they retain the mechanical diaphragm coupling of other Nikkors but the aperture can only be set from the camera body. With the exception of a few early consumer zooms launched between 1996 and 2000, all AF-S lenses are G-lenses, and so are all DX lenses.

Fuji X-cameras with the latest firmware can set the aperture of XF and XC lenses from the camera even when such lenses have no aperture ring (today : XF 27mm and XC 16-50mm). This requires an electronic communication between body and lens.

With any current lens adapter installed, neither a mechanical nor an electronic communication is possible. So how can we then effectively use a lens without its own aperture ring?

A little background on the Nikon F-mount

Since the introduction of the F-mount in 1959, Nikon relies on a mechanical mechanism to control the lens aperture from the camera body. (Almost) every F-mount lens has a spring-loaded aperture lever on its rear bayonet mount, which can be pushed by a linear actuator on the camera, located inside the body mount. The image below shows the back of an AF-S 50mm f/1.4G, and a D700 body.



Initially, this system was only used to force the lens iris fully open during viewing and focusing, and to fully release it when making the shot, letting the aperture close to the value set on the lens.

Beginning with the AI-S models released from 1982 onwards, Nikon made the lens apertures move in a standardized fashion in relation to their calibrated stop-down levers. Nikon AF cameras could now control the aperture by varying the precise position of the control lever. The correlation between this lever’s position and the opening of the diaphragm needs to be adjusted correctly, as well the correlation between the aperture ring on the lens and the diaphragm. Any slight variation in this internal adjustment will vary the exposures you get with that lens, and some lenses may give slightly different exposures than others. There lies the penalty for Nikon’s long-term compatibility with manual lenses, but the system works surprisingly well considering how little movement there is in the mechanism.


It is important to note that the lever position does not correspond to an absolute value for the working aperture: rather, it represents the range from fully open to fully closed iris. F-mount lenses depend on another contraption to “communicate” the maximum f-stop of the lens to the camera: the lens speed indexing post (available on AI and later lenses).

Setting a lower-than-minimum aperture value on a lens with an aperture ring limits the range for movement of the control lever, as illustrated here by the minimum and maximum f-stop settings on an AF 50mm f/1.8. To enable camera control over the full exposure range, the aperture on the lens must be set – and locked – at its minimum value. Now you know why AI-S and AF(-D) lenses have that locking switch!

On G-lenses the spring-loaded mechanism keeps the iris fully closed unless some other device commands it differently. So without any further help, this lens will always be at its minimum f-stop setting, not very practical…

The ‘classic’ lens adapters: keep it simple


Well-established suppliers like Kipon and Novoflex offer adapters for Nikon G-lenses that come equipped with an aperture control ring. The range of control however is limited: each of the two models I have allows the ring to move between ‘fully open’ and ‘fully closed’ over only 30 degrees… The Kipon NIK G-FX does not provide any indication at all (even not which side is ‘open’ or ‘closed’). The Novoflex FUX/NIK has a symbol at each end but no intermediate markings. That means that there is very little precision and repeatability of the settings, other than at the extremes.

The recommended way to work with G-lenses on these adapters is to have the camera in aperture priority mode, first set the control ring to fully open, and then observe the exposure time increase as you ‘close’ the control ring, until the desired corresponding amount of f-stops have passed (starting fully closed and then opening up the aperture is another possibility, of course).

The same adapters can also be used with F-mount lenses that do have an aperture ring. In such case, make sure to move the adapter control ring to its fully closed position (remember how the lever mechanism works!), otherwise your lens aperture setting will have no effect over part of its range. You may even use this particular ‘collusion’ to your advantage, using the adapter’s ring to quickly move between ‘fully open’ (for composing and precise focusing) and the working aperture as set on the lens itself (when taking the shot).

The Metabones approach: 8 stops to happiness (or not?)

The designers at Metabones have equipped both the ‘regular’ and the Speed Booster version of their Nikon-G-to-Fuji-X adapters with a much more sophisticated aperture control ring, that according to their product page “covers a 8-stop range with half-stop markings, clearly indicating actual number of f-stops”. Unfortunately, that is not exactly the case, as we will see in a moment…

The control ring has 8 positions labeled “F” through “8”, with 7 ‘half’ positions in-between. Each position clicks in firmly, which feels right to a still photographer (those focusing on video might prefer a ‘clickless’ approach, for smoother aperture changes during a sequence). So, we get 13 intermediate settings between “fully open” and “fully closed” in a clear and repeatable way.

Update (December 2013): Metabones currently ships its adapters for Nikon G-lenses by default with a clickless diaphragm ring, ideal for video applications. A few extra hardware parts are coming in the box and two instruction videos were put on YouTube for those who want to convert to a clickable ring, or back.

The settings are transferred from the aperture ring to the control lever on the lens via an ingenious system, attempting to simulate the linear movement of the camera body actuator described before.

Metabones points out that the aperture control ring should be put at the “8” position before mounting a lens onto the adapter, to ensure a proper coupling with the aperture lever. We will always make sure to do so! (Actually, the lens will not fully click into the mount if the ring is left at the “F” setting)


Observation 1: the aperture control steps may not be uniform


When we move the control ring, we can already observe that the rotating actuator inside the adapter’s lens mount shows a visibly larger linear displacement from the “F” to the middle “4” position, compared to the displacement from “4” to “F”. It is therefore logical to expect that the steps on the control ring might not translate to uniform and equally spread f-stop changes.

Observation 2: the aperture control ring may not allow the iris to fully open

On many (but not all) lenses, the aperture control ring cannot be moved completely towards the “F” position: it gets stuck anywhere between the “F” and the first half-step. This is probably a mechanical alignment or tolerance problem, with potentially both lens and adapter contributing. I noticed this behavior first on my Speed Booster, but then also on the G-adapter that came in later. It feels like this effect diminishes as the adapter gets more used.

This raises the concern that the Metabones mechanism may prevent a mounted lens from fully reaching its full iris open setting, i.e. from working at its full maximum aperture.

Observation 3: the aperture control ring may not allow the iris to fully close

My very first tests indicated that with many lenses – but again, not with all - there was little or no change to the effective aperture between the “7” and the “8” settings. This suggests that the iris for some reason could not completely close.

A quick visual test, mounting a non-G lens with its aperture ring at the minimum aperture (an AF 50mm f/1.8 set at f/22 in our case) first on an F-adapter (no interference with the aperture control lever) and then on a Speed Booster (with its ring set at “8”), confirmed that suspicion:


In practice, this might be a lesser concern than the one above, as we don’t use that often the smallest apertures on smaller-sensor digital cameras, because of IQ degradation from diffraction.

Let’s do some testing now!

Before discussing a number of test procedures and outcomes, there are a few things that should be made very clear:

  • I am not a qualified, experienced and patient testing expert. As a matter of fact, I am not so much interested in absolute and hard numbers, but much more in developing a feeling about what is reasonable to expect from these and other products: that helps me to decide whether they are practical and meaningful for my intended applications;
  • My background and professional experience as an engineer make me very much aware of the limitations, pitfalls and inconsistencies in my testing approaches. Therefore I try to mostly make comparisons between two alternative setups or situations, rather than trying to measure things;
  • One of the obvious flaws of my work is that everything is based on a sample of one: one copy of each camera, one copy of each lens adapter, one copy of each lens, etc. Not a very solid base to draw definitive conclusions from, especially since a number of the lenses used show some age.

The aperture control tests were done by shooting a 15x20cm Novoflex ZEBRA grey card, placed in a 60x60x60cm light tent, and filling as much of the frame as possible. The scene was illuminated from the top by a fluorescent bulb continuous light source mounted inside a medium-size Chimera softbox. Over the whole test period the light intensity was repeatedly verified to remain at the same level, using a handheld incident light meter.

The Fujifilm X-E1 camera was set to aperture priority mode, and to center spot metering (to avoid as much as possible side effects from e.g. lens vignetting). That allowed to record an exposure time with a ⅓ stop tolerance. A custom white balance (around 4100°K) and the same fixed ISO setting were kept constant during each test series.

Lenses were carefully mounted on the lens adapter with the latter’s control ring – if present – in the “8” position.

Test series 1: lens with aperture ring on F-adapter, aperture control from the lens

Just to set a baseline, a few non-G lenses were put onto the F-adapter, and the aperture was varied over the whole f-stop range the lens offers. As expected, we find a nice linear relationship in diagram 1 (remember the ⅓ stop tolerance on the measured values).

f-stop_FKDiagram 1: F-adapter, aperture setting on lens

Test series 2: lens with aperture ring on Speed Booster, aperture control from the lens

Now we mount a number of non-G lenses on the Speed Booster. The control ring on the adapter is left at the “8” position, (theoretically) allowing full unrestricted movement of the aperture control lever on the lens. Again, the aperture was varied over the whole f-stop range that each lens offers.

Diagram 2 indicates that we still obtain a linear curve as expected, but also shows some strange behavior at the end: five of the tested lenses no longer show any change in the exposure times near the very end of the aperture scale (almost fully closed). This most likely is caused by the iris being hindered in some way by the Metabones mechanism, and therefore not able to close completely, as already visually observed before.

f-stop_SBKDiagram 2: Speed Booster, aperture setting on lens

Test series 3: aperture control using the ring on the Speed Booster

Next, 16 lenses were mounted on the Speed Booster adapter, and for each lens the variation of exposure time as a function of the control ring position was recorded. Any lens having its own aperture ring was kept at its minimum aperture setting: the control ring on the adapter should thus be the only determining factor in controlling the lens iris.

Zoom lenses were tested at three different focal length settings (shortest, intermediate, longest; the graphs only show the curve for the intermediate position).

There is no longer an absolute scale to read the actual working f-stop from: the changes in exposure time as a function of the adapter ring’s position are therefore expressed as f-stops down from maximum aperture.

Diagram 3: Speed Booster, aperture control from adapter

Diagram 3 gives a lot of information (remember: you can click on any chart to see a larger version): 

  • Most lenses follow a similar curve (certainly when taking the ⅓ stop tolerance into account) that is however far from linear;
  • Four lenses (topmost curves) behave differently: the Nikkor 50mm f/2.0 HC (pre-AI, ca. 1970), a Vivitar Series 1 135mm f/2.3 (pre-AI, ca. 1973), a Super Carenar 200mm f/3.5 (AI, ca.1980) and a Tokina 17mm f/3.5 RMC (AI, ca. 1980?); these are all old copies, and with one exception not Nikon but third-party products; we will from now on exclude these from our “calculations” as they clearly seem not typical for the more recent lens population;
  • Several lenses don’t show much effect when moving from the “F” setting to the first half-step position: this could be a combined effect from a) not being able to move the aperture control ring on the adapter fully to the “F” setting, and b) some non-linear behavior of the transfer mechanism;
  • There still is the lack of change in exposure time when the aperture control ring moves towards the very end of its scale (the “8”position), similar to what we saw before;
  • Although some lenses offer a 7-stop aperture range (e.g. f/1.4 through f/16, of f/2.0 through f/22) we only get a 6 to 6 ½ stops out of the Speed Booster.

Test series 4: aperture control using the ring on the G-adapter

I was at first surprised to notice the non-linear behavior as well as the anomalies at the extreme ends of the aperture range, and I suspected an issue with my Speed Booster copy. Then, I found a similar report from another early user on one of the dedicated Fuji X forums. As soon as I received my Metabones G-adapter a few weeks later, I tested a subset of 11 lenses (excluding the four with non-typical behavior, and for some reason overlooking the AF-D 20mm f/2.8) on that adapter.

The results are shown in diagram 4: a very similar outcome. So the non-linearity and the anomalies at the extremes of the range are probably not just a one-time hiccup!

lens_response_MGKDiagram 4: G-adapter, aperture control from adapter

Test series 5: how much better are other G-adapters?

Late in the process I realized that I actually had never verified what range of apertures you can get from the “simple ring” G-adapters. Therefore I quickly ran two 50mm f/1.4 lenses through the five relevant lens adapters. I did not include the Speed Booster, as its behavior closely matches that of the Metabones G-adapter. Note that on the F-adapters the AF-S 50mm f/1.4G always stays at its f/16 setting, as there is no mechanism present to move the lens aperture control lever.

aperture_controlKDiagram 5: available f-stop range on various adapters

The results in diagram 5 teach us several things:

  • Only the Kipon F-adapter, the Metabones F-adapter and the Novoflex G-adapter fully support the 7-stop range offered by these two lenses;
  • Both the Metabones G-adapter and the Speed Booster reduce the available f-stop range to 6 stops;
  • The Kipon G-adapter even loses an additional ½ stop!

We can only conclude that properly managing Nikon’s aperture control lever setup is rather tricky – but also that it can be supported correctly, as the Novoflex result indicates.

Test series 6: a bit of good news – at last!

gain_chart_v2KDiagram 6: f-stop gain with Speed Booster

By now we have collected enough measurements to have a look at one of the fundamental questions: does the Speed Booster indeed deliver the claimed 1-stop gain in aperture?

Four non-G lenses (one 35mm, two 50mm and one 85mm Nikon) were tested first on the Metabones F-adapter and then on the Speed Booster, across their full aperture range.

Diagram 6 is clear: the Speed Booster boosts!

We can see the 1-stop advantage of the Speed Booster. There are a few minor remarks:

  • The AF-D 50mm f/1.4 is one of those lenses that do not allow the adapter control ring to fully reach its “F” setting: that causes the ca. ⅓ stop drop in performance at f/1.4;
  • All four lenses suffer from not being able to reach their “iris fully closed setting” when mounted on the Speed Booster, hence the flattening at the end of the curves.

Aperture control with Metabones lens adapters: the conclusions

If we average the measurements across all lenses and smooth the resulting curve, we get very closely matching relations between adapter aperture ring setting and f-stops down from fully open for both the the Speed Booster and the G-adapter:

adapter_ring_average_2_setsKDiagram 7: Metabones adapter ring control curve

From this, we can deduct the series of ‘Seven Magic Numbers’:

F 2 2.5 3 3.5 5 8

These are the control ring positions with each time a one f-stop difference between them.

If you use a G-type lens, the adapter aperture ring is your only way to have control over the shooting aperture. Keeping the Seven Magic Numbers in mind will allow you to quickly reach a specific f-stop value, depending on the maximum opening of the lens used:

adapter_settingKDiagram 8: aperture setting cheat sheet

You will have to accept the limitation of a six f-stop control range. This situation is not perfect, but definitely workable, and a whole lot more precise and repeatable that using a ’simple’ G-adapter.

If you use a lens with its own aperture ring, I recommend to leave the ring on the adapter on the “8” position and control the working aperture from the lens. That will give you a linear, well defined and clearly visible indication of the f-stop. You might still take advantage of the ring on the adapter to quickly move between “fully open” (for viewing and focusing) and the shooting aperture as set on the lens (ring back to the “8” position).

Leaving (and locking) the lens aperture at its minimum setting will also work, but you will end up with a non-linear and not-so-readable aperture indication.

If you set the adapter ring to anything but “8” and set the lens aperture to anything but the smallest f-stop, two mechanical mechanisms will compete to control the aperture lever on the lens. That makes it very hard to keep track of which one will prevail, so not a recommended way of working!


The Metabones aperture control ring approach comes with the likely loss of the smallest aperture (which will exposes very closely to the one f-stop immediately before), and possibly a ⅓ stop loss at the fully open end of the scale (again depending on the lens used). If that is not acceptable, use a non-G type lens and an F-adapter.

Too bad Metabones doesn’t offer a Nikon-F-to-Fuji-X Speed Booster variant without control ring, for those – like me – that primarily use non-G lenses…

NEXT (and FINALLY!): Speed Booster image quality

Click on any image to see a larger version

No comments: