OM System 90 mm f/3.5 Macro IS Pro
At least among macrophotographers, this has been a long-awaited lens. It was originally featured on Olympus development time-lines a few years ago, then delayed when Olympus sold its camera division. It was supposed to be Olympus' first macro lens in the M. Zuiko Pro series, their first macro lens capable of 2x magnification, their macro lens with the longest focal length so far (early rumors were of a lens around 100-105 mm FLfocal length), and the third macro lens in their Micro 4/3 lineup. It was finally announced by OM System in early February 2023 and initially scheduled to be released at the end of the same month, then in mid-February OM System announced that deliveries would be delayed by up to two months because demand of this lens largely exceeded expectations. However, I received my pre-ordered lens only three days after the originally announced release date.
The parcel was shipped without any external indications it came from OM System. The sender is a provider of industrial services and the packing list in the sealed external pocket was carefully folded to hide any text, sensible precautions to lessen the risk of theft during transportation. This lens is packaged by OM System in their new, environmentally friendly packages of nondescript brown corrugated cardboard, with a minimum amount of plastic packaging materials. Gone are the black Olympus-style boxes of earlier OM System lenses.
The first reports of a suggested price around 2,000 €/USD were received with dismay by potential buyers. However, the price for pre-orders directly from OM System, at 1,499 €/USD inclusive of sales tax and shipment, was more reasonable. At the time of launch, EU prices of Japanese camera equipment traditionally have exceeded US prices by a good 20-30%, but this time EU and US prices are roughly the same.
Although this is still an expensive lens, and nearly three times the current price of the Olympus 60 mm macro, it is much cheaper, for example, than the Olympus 300 mm f/4 Pro (around 2,700 €/USD), and only moderately more expensive than the Olympus 40-150 mm f/2.8 Pro (around 1,250 €/USD). It is also 3.5 times less expensive than the Jenoptik (formerly CoastalOpt) 60 mm f/4 Apo (albeit the latter is not only a good macro lens, but also a lens for extended-spectrum imaging).
The OM System/Olympus M.Zuiko lens range now contains, in addition to the 90 mm, the 60 mm f/2.8 and 30 mm f/3.5 macro lenses (Figure 2). The 60 mm is weather resistant, and the 35 mm is unusual for reaching 1.25x and for its low weight, size and price, but neither is a Pro lens. I routinely use the 60 mm in the macro range, especially in the field, while I find the 30 mm more useful in the close-up range, because of its short focal length and working distance.
Olympus had a long tradition of developing macro lenses. The Zuiko system of lenses for 35 mm OM film SLRs offered 20, 32, 80 and 135 mm bellows lenses, some of them made in two successive versions, and a large photomacrography system of dozens of devices and accessories, as well as 50 and 90 mm macro lenses with built-in focusing helicoids and capable of focusing at infinity.
For about a century, Olympus also had a microscope division. Microscopes were in fact among Olympus' earliest products. The former Olympus microscope division and camera division were sold to other companies a few years ago, and their products are now marketed under the Evident and OM System brand names, respectively. Olympus still produces a broad range of medical and biomedical imaging devices.
Panasonic offers two macro lenses in their Micro 4/3 system, a 30 mm f/2.8 and a 45 mm f/2.8. Samyang and Laowa offer manual-focus macro lenses in Micro 4/3 mounts. I have no hands-on experience with any of these lenses. I often use third-party macro lenses, specialty lenses, macroscopes and microscopes with adapters on Micro 4/3 cameras.
When discussing a macro lens, it is particularly important to understand what is magnification. This is not straightforward, because the marketing departments of several lens makers, together with photography magazines and self-appointed experts, tend to be ignorant of, or to intentionally obfuscate, the concept of magnification. The latter is, quite simply, the ratio of image size to subject size. Magnification is not relative to any camera parameters, but absolute and dimensionless.
Size and weight
At a lens length of 132 mm from mounting flange to edge of filter mount, this lens looks quite long in illustrations on the web. However, a direct comparison with other macro lenses of similar focal lengths shows it is just slightly longer and slightly slimmer than average (Figure 3). In comparison, the Sony 90 mm f/2.8 G is 130 mm long (albeit it only reaches 1x). The Laowa 100 mm f/2.8 macro is 128 mm long in Nikon F mount, and reaches 2x. The legacy Nikon Micro Nikkor AF 105 mm is 132 mm long at 1x (its barrel extends when focusing).
Two of the above lenses are designed for DSLRs, and therefore to allow a high distance between the sensor and the rear optical surface. The OM System 90 mm is comparable in this respect, because it is designed to allow the use of Olympus FLMfocal length multipliers. The latter have a projecting front optical element that requires a lens with a significantly recessed rear element.
The weight of the 90 mm is 453 g. This feels lightweight for a lens of this size, probably because of the generous internal use of plastic and the rather small diameter of the optical elements. I still did not get used to the low weight of this lens, and I often do a double-take when lifting it.
The non-missing lens shade
Some reviewers reported that the lens comes without a lens shade. Mine (which is a series-produced specimen, not an early-review specimen) came with the LS-66E lens shade (above figure, right), also used on the 40-150 mm f/4 Pro. Unlike the lens shades of the 60 mm macro and of the largest Olympus/OM System telephoto zooms, which are retractable, the LS-66E is a generic lens shade of fixed length, quite wide and quite long. It was developed for the 40-150 mm f/4 Pro and can be mounted on the lens in reversed orientation for storage, but this is all. At macro range on the 90 mm, it "eats up" a substantial portion of the working distance, blocks the photographer's view of the subject, and may also block the light emitted by a camera-mounted flash. By choosing to supply this lens shade with the 90 mm, OM System entirely missed the point.
This lens shade and the 62 mm diameter of the filter mount seem to have been chosen for compatibility with other Olympus and OM System Pro lenses. However, on a macro lens it would be useful to make the front end of the lens barrel, as well as the lens shade, as narrow as possible, in order not to restrict the illumination of the subject. The front lens element in the 90 mm has a diameter of slightly less than 30 mm, potentially allowing a much narrower front part of the lens barrel, with a 43 mm diameter of the filter mount. The focus ring could have been moved a bit to the rear and shortened a little without reducing its diameter (which helps to manually focus with precision) to accommodate this.
There is nothing I can do about the diameter of the lens barrel, but I can do something about the lens shade. A much narrower lens shade can be mounted on a step-down adapter, or stack of adapters, screwed into the filter mount of the lens. The above figure shows a lens shade with a 40.5 mm attachment mounted onto stacked step-down adapters. This lens shade does not vignette at any magnification, and it is just as effective as the LS-66E.
This lens shade potentially allows a relatively narrow LED ring light to be mounted onto the step-down adapter. The stacked adapters also allow a 52 mm or 40.5 mm filter to be used instead of 62 mm, which is exaggeratedly large for this lens. A 46 mm lens cap can be mounted at the end of the lens shade.
When it comes to ring lights, I am not a fan of using them as an illumination source for shooting generic subjects, except for special subjects that require a near-axial illumination, but I do appreciate their usefulness for framing and focusing a subject in poor ambient illumination.
The missing lens collar
A lens collar with tripod shoe is a must for using a long macro lens in the lab or on a tripod. Macro lenses of focal lengths up to about 70 mm FL can be used without a lens collar, by attaching the camera body to a support. A 90 mm lens, however, is already a bit too long for the latter method. Even though the 90 mm is lightweight, it does shift the center of gravity of the camera forward. On other lenses, I solved the problem of a lacking lens collar with either of these workarounds:
The OM System 90 mm has an evident, if shallow, groove for a lens collar relatively close to the lens mount. This groove is not designed for the same lens collar used on the 40-150 mm f/2.8 Pro lens. The latter lens has steel studs designed to fit into a groove on the inner side of the lens collar, but the 90 mm lacks these studs. The lens collar of the 40-150 f/2.8 can indeed be used on the 90 mm, but it does not fit as securely as on the former lens when the collar is unlocked and the lens turned within the collar.
Hopefully, OM System will introduce a dedicated lens collar for the 90 mm, and sell it as a separate accessory. An alternative is to try a cheap third-party lens collar made in China for the 40-150 mm f/2.8, like the one in the preceding figure (branded Haoge and purchased on eBay). It is equipped with an Arca-compatible plate and a ring of stiff plastic inside the metal ring for the lens, and is sufficiently stiff, although probably not as much as the genuine collar. The four notches on the front shoulder of the lens ring facilitate mounting the collar on a lens equipped with matching studs.
There is a QDquick disconnect steel socket between the two threaded sockets of the plate, apparently compatible with the QD System sockets on camera and lens plates offered by several manufacturers, and the matching neck straps and hand straps. A neck strap can also be attached to the built-in eyelet at the front of the plate. The genuine Olympus collar lacks both features.
The Promedia SS2 system is a comparable, but not interchangeable, quick-connect strap system with a smaller socket and a few functional drawbacks, compared to the QD system. The SS2 system is currently end-of-life, and not recommended for new purchases. The QD system is also used for rifles and other firearms. Care should be taken when using QD straps designed for firearms on camera equipment, since the metal fixtures of these straps are often rough and may accidentally damage delicate camera parts, including LCD screens and painted or chrome-plated parts.
Figure 8 shows a detail of the 90 mm equipped with the third-party collar. The collar is a good couple of mm too narrow to fit the groove in the lens barrel. When clamped, the collar safely holds the lens in position, but if you release the collar for any reason, you must remember to hold it pushed against either the front or the rear of the groove in the lens barrel when locking it again. So this is not a collar you can use to quickly shift between landscape and portrait orientation. It probably works as such on a 40-150 mm f/2.8, held in place by the studs on the lens barrel, but not really on the 90 mm.
The lens collars of the Olympus 300 mm Pro and 100-400 mm "non-Pro" are similar in shape, but are wider and do not fit on the 90 mm.
Anothe possibility is using the collar of the Zhongyi Mitakon Creator 85 mm f/2.8 Apo 1-5x Super Macro. The collar diameter of this lens is slightly narrower than the above, but is sufficiently thin to adapt itself to a tight fit around the barrel of the 90 mm, and holds it quite well. Also this collar leaves a gap between one of its ends and one end of the groove on the barrel of the 90 mm lens. This collar is too tight to allow rotation of the lens within the collar once the locking bolt of the latter is released.
The optical scheme (18 elements in 13 groups) is at the high end of complexity for a macro lens, unlike the average complexity of the 60 mm f/2.8 (13 elements in 10 groups) and the even simpler 30 mm f/3.5 (7 elements in 6 groups). In the 90 mm, focus is achieved by two internal optical sub-assemblies moving independently of each other, and a third subassembly provides ISimage stabilization.
Neither the frontmost nor the rearmost optical elements move when changing magnification, as far as I can see.
Like in the majority of M. Zuiko Pro lenses, the focusing ring has a push-pull clutch to switch between MF and AF. On the OM-1 and other modern bodies, the camera switches to MF and displays focus peaking (if set in the menu) when the focus ring is turned while in AF+MF mode. You can also program one of the buttons to activate the image magnification when pressed, which helps with precision manual focusing.
In the 90 mm, the focus clutch has a longer travel than usual. The reason for this is to reveal a double scale when pulled back, with magnification at the top (ranging from 0.25x to 2x) and focus distance (not working distance) at the bottom.
The focus clutch, when set to MF, overrides the menu and SCP AF/MF settings. If the clutch is set to AF, the menu or SCP settings override the focus clutch. Thus, you can for example set the focus clutch to AF and choose MF in the SCP, and in this case the actual mode is MF.
The manual focus scale is strongly skewed in favor of macro and close-up distances. The rotation of the focus ring between infinity focus and 2 m is almost imperceptible, while the rotation between 0.5 m (0.25x) and 0.224 m (2x) is almost one-quarter of a turn. I can still focus at a distance in MF with some care, with the aid of focus peaking, but it is not reliable enough for routinely using this lens in manual focus on distant subjects. AF on distant subjects, on the other hand, is both accurate and fast.
Just think like this: The focus ring in this lens is not really for focusing in MF, it is for setting the magnification. Afterwards, you focus by moving the camera with respect to the subject, which does not change the magnification. Every experienced macrophotographer knows this.
Manually focusing on a distant subject in AF+MF mode is way more precise than in MF mode, and AF+MF is the right way to manually focus on distant subjects with the 90 mm. If the focus ring is rotated slowly, focus at long distances changes at a snail's pace. It takes literally more than one slow full turn of the focus ring to focus from infinity to half a meter. A quick turn of the focus ring engages a much faster burst of focusing speed, then you can slowly fine-tune the focus to your liking. You need to learn to resist the temptation of engaging MF with the focus clutch whenever you want to manually focus, and to just turn the focus ring instead. Use the MF clutch position only to set the magnification in the macro and close-up range.
According to DPReview, the 90 mm uses a stepper motor for focusing. In general, a stepper motor is slower and noisier than the AF linear servo motors or voice coil motors found in most modern lenses. The 90 mm, however, is almost silent in spite of the stepper motor. AF in the 90 mm takes over one second to rack all the way from 1x to infinity, so it is not a speed demon but not a snail, either. There can be good reasons for OM System's choice of motor, for example focusing in this lens requires a longer actuator travel than typical for general‑purpose lenses, as well as a high positioning accuracy.
The focus limiter works like the one on the 60 mm macro, i.e. both in AF and MF. Both AF and MF stop focusing at either current limit of the focus limiter. In MF mode, you can freely turn the focus ring between ∞ and 2x, but the focusing action just stops within the current interval of the focus limiter. AF+MF mode also respects the settings of the focus limiter. When you change the switch position of the focus limiter, the lens instantly and almost silently refocuses to the nearest end of the newly selected focus range (more with a "click" than a "whine").
The focus limiter switch has three-positions:
This means that the lens does not have a continuous, uninterrupted focus range from infinity to 2x, but two separate, partly overlapping focus ranges: 2x-0.25x and 1x-∞. The focus limiter switch must be manually operated to switch between these two ranges. The third range (1x-0.25x) is only a subset of the second.
The focus limiter switch has no position to use this lens as a generic telephoto or portrait lens (e.g. to allow AF between 0.5 m and infinity). However, most recent high-end cameras provide this function in their menu, and also allow the lower limit of the focus range to be customized. Perhaps OM System decided against providing this position on the focus limiter switch because they trust AF to be reliable enough to prevent focus hunting when using this lens for portraits or distant subjects. Likely this requires the camera to support phase AF, so you should tolerate more focus hunting with this lens on cameras that do not provide phase AF, or do not recognize this lens (primarily non-Olympus/OM System cameras or legacy models without phase AF).
The S Macro setting of the focus limiter is further discussed below, since the effects and implications of this setting go far beyond a simple restriction of the focusing range and magnification range.
The lens has an image stabilization switch that allows IS to be switched On or Off. This controls both in-lens and in-camera IS.
The IS system built into the lens is designed to be used either alone or in tandem with sensor IS (as configured in the camera menu). I can see no good reason to switch off either in-lens or in-camera IS. In the past, IS needed to be switched off when the camera was mounted on a tripod, lest shutter vibration fooled the IS system into producing an uncontrolled motion blur. This no longer seems to be the case, and I see no difference from leaving IS on also when the camera is attached to a support.
Even when all IS is switched off, the IS electromagnets are always powered during exposure, to prevent the camera sensor or IS optical groups in the lens from de-centering under the effect of gravity. IS can be configured in the camera menu to activate whenever the shutter button is half-pressed, which is always desirable for easier framing in hand-held macro imaging.
An inertial sensor built into the lens is used to detect very small shift movements when hand-held. Information from this sensor is used to complement the camera sensors controlling IS. In the macro range, the camera sensors cannot do a complete job of stabilizing the image, because the IS sensor data necessary for this job is different from general photography. The IS sensor in the lens registers the small movements that require the additional stabilization. According to OM System, this improved IS allows focus stacking handheld in the field. This is often successful when holding the camera as steady as possible, but not always.
I am not aware of macro lenses - or cameras - of other brands using a comparable, specialized IS system for macrophotography. The enhanced IS of this lens could be reason enough to own it, if you shoot a lot of macro stacks in the field.
There is a configurable L-Fn button, which is present in most, but not all, Pro lenses. By default its function is to stop AF when the camera is set to C-AF.
This section can be of technical interest, but if you are only reading this review from a user's perspective, you can skip it and go directly to the next section.
A simplified formula for the effective lens aperture is
Ae = An (R + 1)
where Ae is the effective aperture, An the nominal aperture, and R the magnification, a.k.a. reproduction ratio. However, in the macro range another factor becomes important, i.e. the pupil ratio P:
P = Pout / Pin
where Pout is the exit pupil (i.e. the diameter of the lens aperture as seen through the side of the lens facing toward the sensor) and Pin the entrance pupil (i.e. the diameter of the lens aperture as seen through the side of the lens facing toward the subject). Since P is a ratio of two linear measurements, it is dimensionless. Effective aperture formulas (including DOF formulas) that contain the magnification can be made more precise in the macro range by replacing each occurrence of R with R / P. For example:
Ae = An ((R / P) + 1)
Note that whenever R is small, ((R / P) + 1) approaches unity, and therefore the pupil ratio is unimportant when the lens is focused on a distant subject. Thus, the pupil ratio in general-purpose lenses is a relatively unimportant consequence of other decisions made during the lens design. However, the pupil ratio becomes very significant in the macro range. For example, it allows lens designers to decrease the effective aperture in the macro range by choosing a high pupil ratio, without needing to decrease the nominal aperture by the same amount. A lower nominal aperture is very costly, because it requires a larger diameter of the front optical groups, and a higher number of optical elements to correct the rapidly increasing optical aberrations (some types of aberrations scale up much faster than lens diameter).
As always, there are trade-offs, and a high pupil ratio is often associated with a short working distance in the macro range. Another way to view this is that a high pupil ratio decreases the working distance and therefore increases the divergence of the cone of light originating from a point of the subject and passing through the lens, which results in a lower effective aperture.
Lenses with internal focusing, including the OM System 90 mm macro, change most of their operating parameters when focus changes from infinity to close range, including the pupil ratio. Therefore, to achieve a better precision, I measured the diameters of the front and rear pupils both at infinity and 2x.
By default, the diaphragm of Micro 4/3 lenses fully opens when the lens is dismounted from a camera, or the camera is powered off. However, it is often difficult to measure the diameter of the entrance and exit pupils with the diaphragm fully open, and neither the aperture nor the focus distance can be set on a modern Micro 4/3 lens detached from the camera body (except for lenses with manual, mechanically operated aperture and focus). Turning the focusing ring of a focus-by-wire lens, for example, has no effect if the camera is powered off or the lens unmounted from the camera. Even though the focus ring clutch is set to MF and the distance scale moves with the focus ring, the lens does not actually focus.
The following table shows P for the Olympus 30 mm, Olympus 60 mm and OM System 90 mm macro lenses. I measured P of each lens both at infinity and at the maximum magnification each lens is capable of.
As the table shows, there are large differences in P for each lens at either focus distance.
The 90 mm is close to unity pupil ratio at infinity, but surprisingly the pupil ratio is much smaller at 2x. It was difficult to measure the entrance pupil at 2x, because in addition to being very large it is also located at a high optical distance from the front of the lens. At infinity focus the entrance pupil is located, unremarkably, halfway down the lens barrel.
I was able to measure the outline of the diaphragm in its entirety only by stopping it down to f/22. Any more open, and one simply cannot see the edge of the pupil by looking into the front element from a distance. At low apertures, this pupil is wider than the front element of the lens.
The measured pupil ratio in this lens, plugged into the formula for effective aperture, at 2x yields an effective aperture of f/34 for a nominal f/3.5, i.e. a loss of roughly 7 stops. This would make the lens useless at 2x, because of the high loss of resolution caused by diffraction. Therefore the pupil ratio, at least as traditionally computed, does not seem to be a reliable predictor of the effective aperture at 2x. This made me question whether the effective aperture in this lens is truly as predicted by the generic formulas.
As an additional test, I measured the effect of magnification on effective aperture by recording the exposure time for a featureless illuminated subject at infinity focus and 2x. This is not an exact method because the camera only displays an approximate exposure time, but nonetheless it is a more objective method than a calculation based on pupil ratio. A lens with unity pupil ratio at 2x should display a 3-stop "loss" in effective aperture, compared to infinity. This, in turn, should result in a 23 = 8‑fold increase in exposure time. In aperture priority mode and f/11, the camera exposed at 1/6 s at infinity, and 1/2 s at 2x. This is only a 3-fold increase, which seems to indicate an approximate loss of only 1.5 stop. Obviously, there is more going on than just the effect of the pupil ratio.
Repeating this test with the Olympus 60 mm macro yields a 2.7 times increase in exposure time at 1x, i.e. a loss of a little more than a 1.5 stops. The measured P of 3.08 at 1x in this lens leads to a calculated 1.3 stops loss at 1x, which is in the same ballpark as the figure measured from exposure time, although not exactly the same.
A likely explanation of the discrepancy of these figures in both lenses is that the physical diameter of the diaphragm opening, at a given nominal f/stop, in these lenses seems to remain constant regardless of magnification. However, the lens focal length shortens when focusing at closer distances (especially in the 90 mm), and as a result the constant diameter of the diaphragm opening translates to a faster effective aperture, which is not accounted for in the calculations of pupil ratio but does emerge from the measurements of exposure time. It is far from easy to calculate the actual focal length in a lens of variable focal length, pupil ratio and effective aperture, so for the moment I will not attempt it.
The effective aperture also affects DOF. In lenses with internal focus, using a calculated effective aperture to set the distance between steps in focus stacking/bracketing can lead to overestimating the DOF and causing focus banding. Shooting a set of test stacks at the desired magnification, or allowing for a good safety margin of focus overlap between successive images, may be the best practical solutions.
Fortunately, OM System provides some data on effective aperture in the user guide of the 90 mm. Since my calculations basically hit a wall, I will rely instead on this data.
In S Macro mode, the camera reports a nominal lens speed of f/5, but f/3.5 in the other focus ranges. The user guide of the 90 mm provides the following, more detailed information on effective aperture:
The magnification range 1x-0.25x is shared by all focus ranges, but the effective aperture in this magnification range is different in the S Macro range. I initially reasoned that at 1x, for example, it would be better to shoot without S Macro, in order to take advantage of the faster nominal aperture. However, the MTF diagrams (see below) show that just the opposite is true. You do get a better image quality in the S Macro range, in spite of the slower aperture. Exactly in which sense it is better, I discuss in detail here.
It is interesting to note that effective aperture increases more slowly, with increasing magnification, than expected for a lens of unity pupil ratio, and even more so for a lens with pupil ratio of 0.23 as computed in the preceding section. This tells me that my calculation is way off. On the contrary, the 90 mm actually behaves like a lens with a pupil ratio above unity.
At 2x, the working distance is 72 mm.
Focal length multipliers
The lens can be used with the MC-14 and MC-20 FLMfocal length multipliers, and with either of these, it reaches 2.8x or 4x, respectively. You cannot stack two of these FLMs atop each other. Their rear optical surface flush with the rear bayonet prevents this.
The mount of the 90 mm carries the two additional electrical contacts for an FLM (above figure) also found on the 40-150 mm f/2 Pro, 300 mm f/4 Pro, 150-400 mm f/4.5 Pro, and (I believe) 100-400 mm f/5.0-6.3. Like in these lenses, the additional contacts on the 90 mm are round and chrome-plated, unlike the gold-plated contacts used for other functions. The MC-14 and MC-20 have only one additional contact on their front mount, in different positions in the two FLMs. This suggests that these specialized pins do not transmit data, but are only shorted to ground. We also know that these FLMs have no upgradeable firmware, and behave like "dumb" adapters with straight electric connections between their rear and front mounts. The camera knows that an FLM is mounted, and which FLM, only if the lens tells so to the camera.
In the 90 mm, the rearmost optical element is so recessed that it makes me wonder whether its design allows for a new FLM yet to be announced, or a new type of accessory mounted at the rear of the lens. As I argued in an earlier occasion, a new FLM (or another device mounted between camera and lens) could use both contacts to uniquely identify itself to the lens. Beyond that third device, additional pins or a different communication protocol will need to be used, and will likely require new lenses.
The MC-14 increases the lens aperture (both nominal and effective) by one stop. The MC-20 increases it by 2 stops. Therefore, with the MC-14 the nominal lens speed becomes f/4.5, and with the MC-20 f/7. Working distance remains the same as without an FLM. Hand-held photography at 4x, even with the help of this lens' specialized IS and AF, is quite a challenge.
As a whole, working with an FLM with this lens is feasible without major problems when the lens is used as an ordinary telephoto lens, or a close-up lens. In the close-up range, an FLM can be useful to achieve a longer working distance. If the latter is not necessary, up to 2x it is generally better not to use an FLM because the lens alone gives a better image quality. The MC-20 turns the 90 mm into a nominal 180 mm f/7. Image quality with an FLM decreases in the macro range, because of the increased effective aperture.
From Figure 15 we also learn that the lens is made in Vietnam. It was probably a wise decision by Olympus to leave China years ago, to avoid becoming yet another hostage of Chairman Xi when (rather than if) China finally blunders its way into serious international conflict and gets trapped in crippling sanctions, arms races and/or war.
Focus stacking and bracketing
For the first time, in-camera focus stacking and focus bracketing becomes possible with this lens - with careful hand-holding and reasonable illumination - up to 4x. This is a paradigm shift, especially for photographers that do focus stacking in the field. With this lens, for most applications it is no longer necessary to carry a motorized rail and controller. It might even be possible to work above 4x with an add-on lens attached at the front of the 90 mm (so far I have not tried).
In-camera focus stacking and focus bracketing with this lens are much faster than a motorized external focus rail, and the new IS does allow hand-held stacking. If you only have the 90 mm and MC-20 in your bag, and perhaps a mini tripod, you can expect to take home a usable focus stack in the 2x-4x range, at the expense of a moderate loss of image sharpness.
The following cameras are known to be fully compatible with the 90 mm when their firmware is updated to the present version (as of February, 2023):
Note that most cameras purchased before the second half of 2022 need to update their firmware, in order to perform optimally with this lens. Without updated firmware, the camera probably will not allow focus stacking and focus bracketing with this lens. I will not be surprised if OM System will release future firmware updates that contain optimizations to improve the performance of any of the above cameras with the 90 mm.
Older versions of the E-M5 and E-M10, as well as the E-M1 "Mark I", are not on this list. Most likely you will not be able to use the 90 mm for focus stacking and focus bracketing on these older cameras (I heard it can work with a hacked firmware, of which I know nothing). I don't know what else you could miss, if anything, by using the 90 mm on these cameras.
The 90 mm can use the Olympus STF-8 macro flash. It can also use either the Olympus MC-20 or MC-14 FLMs.
Note that you cannot stack the MC-20 atop the MC-14, or vice versa.
PC Magazine reported a resolution of "2,800 lines at f/3.5-11". Exactly what this means is unclear, and my cursory search did not find a separate page where they explain how they rate lenses. They did not specify whether they really mean lines, or the far more common line pairs. I assume they mean (1) lines, (2) on the sensor side, and (3) across the sensor's width (17.3 mm). These are a lot of assumptions, but if true, they translate to 81 line-pairs/mm on the sensor. This is roughly the same as the Sony 90 mm f/2.8 G Macro lens in my tests.
PC Magazine did not report the magnification at which they tested this lens, either, but state that resolution remains the same from f/3.5 to f/11. This is unlikely at 1x or 2x, because of diffraction. It must therefore follow that PC magazine tested the OM System 90 mm well outside the macro range, most likely with the same resolution target they use for non-macro lenses.
LensTip.com reported a center image resolution of 63 lp/mm at MTF 50 fully open, which increases to 77 lp/mm at f/5.6 and decreases afterwards. With the MC-20, center resolution decreases to 34 lp/mm fully open and peaks at 42 lp/mm at a combined nominal f/11.
I shot the following tests with a Thorlabs NBS-1963A resolution target ranging up to 228 lp/mm. This is a 5 x 5 cm target directly labeled in lp/mm, and the images show only its central portion.
My own results at 1x, 2x, and at 4x (total) with the MC-20 2x focal length multiplier, tell a different story. At 1x, the 90 mm resolves 128 lp/mm on the sensor from f/3.5 to f/8. Sometimes, when the target lines align favorably with sensor pixels, 144 lp/mm can be barely resolved, albeit in a minority of pictures. This happens to be very close to the sensor's theoretical Nyquist limit of 149 lp/mm. The Nyquist limit is a best-case parameter, so an actual image sometimes resolves a line pattern at the Nyquist limit, other times renders it as a featureless blur.
At 2x, the resolution target maximum line density of 228 lp/mm (i.e. 114 lp/mm on the sensor) is resolved up to f/11, and higher line densities on the target would probably have been resolved, if available.
In a previous, theoretical discussion of Micro 4/3, I computed that a 20 Mpixel Micro 4/3 camera can resolve up to approximately 115 lp/mm at MTF 50%, or 128 lp/mm at MTF 40%, while the sensor's Nyquist limit is 149 lp/mm. It is comforting to see that the actual resolution of the 90 mm at 1x (128 lp/mm) is not too different from the theoretical maximum resolution of the OM-1 sensor.
In the test at 1x, some color aliasing becomes visible at 128 to 161 lp/mm, produced by the line pitch of the target line pattern being close to the pitch of the sensor's Bayer filter pattern. In the compromise between maximum image resolution and suppression of color aliasing, OM System apparently decided for the former when choosing the strength of the electronic anti-aliasing filter of the OM-1. Paired with a lens capable of matching (or possibly exceeding) the maximum sensor resolution, like the 90 mm, and a test pattern of the critical size, the imaging system cannot avoid producing color aliasing.
With my present resolution target reaching 228 lp/mm, I am unable to test an on-sensor resolution higher than 114 lp/mm at 2x (which the 90 mm seems to abundantly exceed). I would not be surprised if the 90 mm closely approaches the same on-sensor resolution as at 1x, at least at the lower apertures.
At total 4x with the MC-20 focal length multiplier, 228 lp/mm on the target (i.e. 57 lp/mm on the sensor) are abundantly resolved up to f/16, but not at f/22. f/10 is the best aperture, closely followed by f/11. Note that the camera with MC20 reports an aperture range between f/10 (i.e. f/5 without multiplier) and f/22. It does not allow further stopping down, which would be useless. Although images at this magnification are still usable for most purposes, the lens is a little out of its comfort zone. The Laowa 25 mm 2.5-5x f/2.8 may optically be a better choice in the upper part of its magnification range, albeit with a lower working distance, and requires a motorized rail to perform automated stacking.
The above figure is a 1:1 crop of an IC chip shot with 90 mm and MC-20 at 4x. The finest detail is on the order of 2-3 pixels in diameter, or 6.7-10.0 μm on the sensor (i.e. 1.7-2.5μm on the subject). This combination of lens and FLM no longer resolves detail of the order of a single pixel, like the lens vithout FLM does at lower magnification, but the image quality is nonetheless still respectable.
Even in the lack of finer line patterns, the resolution target still provides a qualitative estimate of image sharpness at 2x and 4x, if one looks at the sharpness of the white-to-black borders of the lines. Herein may lie a solution. The edges of the thicker target lines seems to be straight and sharp enough to be suitable as MTF targets. Initial results (below) indicate that this is a viable method for using this target to measure image resolution in imaging systems that, above 1x, exceed the lp/mm resolution of the test target. This could be a way to sidestep the need to use a resolution target with even higher line density, which I am not planning to purchase because such a target is built with electron beam lithography and, together with suitable mechanical fixtures and illumination, would not cost much less than the lens I intend to test.
I took sample images of a sharp edge of the resolution pattern inclined at 5°and used them to plot the MTF as described here.
Note that the MTF graphs show the actual resolution in lp/mm on the sensor. You do not need to factor magnification into the calculation, like it is necessary when visually evaluating the resolution on images of the test target.
Figure 24 shows the MTF at 0.5x and f/3.5 (left). To read the MTF at a given percentage (e.g. 40% modulation), locate the 0.4 label on the Modulation Factor axis, then follow the corresponding horizontal line until it intersects the curve. The MTF at 40% is in this case approximately 103 lp/mm on the sensor.
Since the 1x magnification is available in both focusing ranges, I measured it in both cases. I was surprised to see that it is substantialy different in the two focusing ranges, and significantly better at f/5 in the S Macro range, compared to f/3.5 in the other ranges. This is exactly the opposite of what I expected. Therefore, it is better to use the 1x magnification in S Macro mode, in spite of the penalty dictated by the higher nominal aperture. The 40% MTF is approximately 70 and 93 lp/mm on the sensor, respectively. The increase in resolution is dramatic at high percent MTF. At 80 % MTF, for example, resolution more than triples. The resolution advantage is less noticeable at lower MTF, but is nonetheless still there, with an increase of roughly 30-50 lp/mm.
At 2x and f/5, the 40% MTF is around 65 lp/mm. Note that this is higher than at 1x and f/3.5 outside the S Macro range (but not at 1x and f/5 within the S Macro range). At 4x and f/10 with the MC-20, the 40% MTF is around 54 lp/mm, roughly the same as at 1x and f/3.5 outside S Macro. Therefore, it seems that switching the lens to the S Macro range does something that significantly improves its image quality. It is more than just a different range of focusing distances and nominal apertures.
Switching between S Macro and another setting of the focus limiter with the lens focused at 1x and the focus clutch set to MF causes a very brief noise inside the lens, so something mechanical is moving there, although focus does not change. As far as I can see, the image in the viewfinder remains exactly focused. Something else does change in the viewfinder: when switching to S Macro, for a fraction of a second the live view becomes darker, then returns at the same luminosity. When switching back to the 0.25-0.5m setting, the live view becomes brigther for an instant. This means that the effective aperture is changing. It is possible that no optical group is actually moving, and that the change is of a different nature, perhaps only involving the diameter and/or position of the aperture. Incidentally, if the position of the aperture changes within the lens, this should change the pupil ratio of the lens, which in turn changes the effective aperture at a given magnification even when the physical diaphragm diameter and nominal aperture do not change.
In this lens, between 0.5x and 1x the user is left with a "choice" of using the optimal S Macro setting of the focus limiter, or the poorer optical performance of the 0.25-0.5m setting. Giving the user this choice without explaining the consequences in the user guide is, quite simply, poor ergonomics, poor technical communication, and a clear design blunder. Likewise, breaking the focus range into two partly overlapping ranges (1x-∞ and 2x-0.5x) and not offering a single, uninterrupted range of 2x-∞ is a poor ergonomic choice.
The OM System M.Zuiko 90 mm f/3.5 IS Pro is the first macro lens with AF and IS reaching 2x. It does not have a single, uninterrupted focus range from infinity to 2x, but two partly overlapping ranges. Using the S Macro range gives a better image quality in spite of the higher nominal aperture. Image resolution is high up to 2x, and is still usable with focal length multipliers, but visibly less sharp. It is equipped with a so-far unique inertial sensor for IS and in-lens IS working in tandem with in-camera IS that enables hand-held macro and focus stacking up to 2x.
While the lens performs admirably when the focus limiter S Macro setting is correctly used between 0.5x and 1x, it is highly questionable to leave it to the user to decide whether to manually engage S Macro, or to obtain a worse IQ in this magnification range. Whatever optimization is performed by engaging this control, it should be done/undone automatically every time the magnification crosses the 0.5x threshold. There should also be a single, uninterrputed focusing range from infinity to 2x.