Olympus OM-System Zuiko Reflex 500 mm f/8
Warning: Images on this page may display examples of ring bokeh, also known as doughnut/donut bokeh. This bokeh is known to cause uncontrollable fits of wrath, overwhelming disgust, existential panic attacks, and other allergic reactions in sensitive photographers. If you suspect you may be one of these, stop reading this page now. I will not be responsible for any physical damage or psychological discomfort resulting from you not heeding this advice.
Virtually all camera brands have marketed at least one model of catadioptric lenses at one time or another. Catadioptric lenses use a combination of refracting (="transparent") and reflecting (="mirror") optical elements, and fold the optical path of the lens twice by using a large primary reflector at the rear of the lens and a smaller secondary reflector at the front. As a result, a catadioptric lens makes it possible to shorten the lens barrel to one-third the length of an equivalent refractor lens. Reflecting optical elements can sometimes be made thinner and lighter than refracting elements.
Statements that catadioptric lenses are immune to chromatic aberrations can be read in several Internet sources. Reflecting optical surfaces are indeed immune to chromatic aberrations. However, the refracting elements also present in catadioptric lenses do introduce these types of aberrations. Therefore, catadioptric lenses are not intrinsically immune to chromatic aberrations. The only lenses truly immune to chromatic aberrations are pure reflectors that use no refracting elements (e.g., large reflecting telescopes). Nonetheless, if catadioptric lenses are carefully designed, the amount of chromatic aberrations of these lenses can be quite low.
Another characteristic of catadioptric lenses is that the secondary mirror is coaxial with the primary one, and masks off the center of the latter. As a result, catadioptric lenses produce annular bokeh highlights. If the subject contains out-of-focus branches or sticks, their bokeh consists of two parallel stripes. This is called nisen boke (two-line boke) in Japanese literature. Incidentally, the bokeh spelling ubiquitous in English literature is plainly wrong, and was originally introduced by an English-speaking photographer with inadequate knowledge of Japanese. As a wrong transliteration of a Japanese word, Bokeh is by no means alone. Fuji Yama / Fujiyama is probably the most famous example. Even though I still resist the likes of who'se and who's in place of whose, you're in place of your and it's in place of its, I am aware that resistance against illiteracy is futile. Therefore, on this page I use bokeh as an English word (since English speakers made it up, they deserve to have it), except when combined with other Japanese words.
There are types of reflector lenses immune to nisen boke, including the Makowsky Katoptaron, but they are quirky and differ from refractor lenses in several aspects, and are relatively slow (f/11 or sometimes f/8). Their bokeh is disk-shaped but not centered onto the actual position of the highlight. Incidentally, the Makowsky Katoptaron is not radically different from other optical designs of reflector and catadioptric lenses. It only looks unique. Some details are available in my book on scientific photography if you are interested.
The image quality of catadioptric camera lenses is quite variable. However, aside from the limits imposed by diffraction on all conventional optics, there are no intrinsic limitations of reflector versus refractor lenses in terms of contrast and image resolution. After all, the Hubble Space Telescope and virtually all high-magnification space- and Earth-based telescopes are reflectors. The high resolution super-telephoto lenses used in military drones and intelligence satellites are likewise reflectors or catadioptrics, probably including the four-lens array used in a 1,800 megapixel camera. The broad differences in resolution and contrast observed among commercial catadioptric lenses are simply caused by different design compromises and manufacturing quality, which in turn are determined by the intended price class. Refractor super-telephoto lenses of well-known camera brands tend to be of better quality because they are far more expensive. In the 500-600 mm range of focal lengths, eBay offers many low-cost refractor lenses, including 500 mm f/8 models branded Panagor, Galaxy, Kalimar, Opteka etc. that are even worse than most catadioptric lenses and were originally designed as cheap spotting scopes.
The small size and low weight of catadioptric lenses, compared to refractors of the same focal length, makes catadioptric lenses attractive. A large group of low- to medium-quality catadioptric lenses are branded Opteka, Tamron, Sigma, Samyang, Makinon, Centon, Bower, Cambron, Chinon, Falcon, Kalimar, Revue, Revuenon, Rokinon, Spiratone, Walimex, Sunagor, Soligor, Quantaray, Telesar, Lentar, Tokina, Hanimex, Vivitar etc. The Vivitar Solid Catadioptric series is interesting for its unusual optical design, but gives a poor image quality. The Tamron 500 mm f/8 is said by some photographers to be the best in this group of lenses. If this is true, then this group does not offer any lens I may want to use.
Soviet-made MTO models are longer and wider than average. Opinions on their image quality vary. The best image samples I have seen are far from excellent but could be acceptable, and this is as good as they get. The more modern Rubinar are smaller but not better. There seems to be a large variation in image quality among specimens of both brands, so basically you have to buy one, or two, or three, and hope to win the lottery. There are reports of some of these lenses being assembled with rear elements in the wrong orientation, and of course performing poorly. I do not know whether these are factory accidents/sabotage, or the result of careless reassembly by owners/third-party technicians after internal cleaning. Others report that the front/rear mirrors are sometimes deformed by screwing their retaining rings too hard at the factory. This type of damage in principle could be reversed, but optical quality is likely to be spoiled by performing maintenance without suitable instruments to realign the optical elements.
MTO lenses don't seem to possess a front internal baffle, only a rear internal baffle. Most of the other catadioptrics have both a front and a rear internal baffle, and in principle should be less susceptible to off-axis illumination. The MTO catadioptrics are also unusual in not covering the front of the secondary mirror, which is visible as a silvery coin-sized circle through the front lens element. This may allow multiple reflections of off-axis light within the front element and make the lens potentially more vulnerable to flare and low contrast.
Minolta made the only autofocus catadioptric 500 mm. The most recent among the 500 mm models by Nikon (with orange strip around the focus ring) is said to be better than average and than earlier Nikon models. The Olympus Zuiko Reflex 500 mm f/8 for the OM film SLRs is regarded as among the best of the lot in terms of image resolution. The Yashica ML 500 mm f/8 (but not the other Yashica models) is also said to be in the top tier. The Zeiss/Contax Mirotar is supposedly very good. On the second-hand market, the Zeiss Mirotar usually costs three-four times more than the Olympus, but its image quality in available image samples is very similar to the Olympus. In general, famous camera brands had better quality control than manufacturers of cheap third-party lenses, and the product performance of these brands is more reliable. Good-brand catadioptrics also sold at higher prices than third-party ones (and still do on the second-hand market). However, bad apples can occur in any brand, and good lenses can become misaligned through accidents or improper servicing.
The contrast of catadioptric lenses is almost invariably lower than in good refractors. Some of the reasons are discussed below. A possibly important variable among catadioptric designs is that some models use first-surface mirrors, others second-surface mirrors. First-surface mirrors are more delicate and their reflective coating can be degraded by exposure to air and humidity, resulting in a gradual loss of reflectivity and increase in susceptibility to flare. Some first-surface mirrors are coated with calcium fluoride or silicon dioxide films, which retard the degradation but add to the cost of the lens. First-surface mirrors are also immune to chromatic aberrations. Second-surface mirrors are much less likely to be affected by age-related degradation, especially if the rear of the mirror is sealed with paint or lacquer. However, in second-surface mirrors the front refracting surface is crossed twice by light. This makes the lens design more complex and more expensive to implement because both surfaces of this element must be produced to high optical standards (as opposed to only one in front-surface mirrors) and the introduced chromatic aberrations must be corrected somewhere else along the optical path.
In most catadioptric lenses, the diameter of the front element and primary mirror are chosen as a compromise to reduce the production costs. Large optics are expensive, and in catadioptrics these two elements are a major part of the cost. Therefore, their diameter is as small as possible, and most of these lenses display a darkening in the periphery of the image caused by vignetting by the edges of these elements. This problem is much lesser when these lenses are used on APS-C or Micro 4/3 sensors. Another limitation of catadioptrics is that image quality is significantly lower near the image corners. Since the large majority of catadioptrics are designed for use on 24 x 36 mm film, image quality in the corners is not an issue when used on APS-C or Micro 4/3 cameras.
Someone once said that every photographer is destined to buy a catadioptric lens during his/her career, and to get rid of it shortly thereafter. I did buy one, and found my second-hand Sigma 600 mm f/8 quite unremarkable in both resolution and contrast. I could get far better results with a good 300 mm and 2x focal length multiplier, and therefore sold the 600 mm for a modest profit afterwards. So why did I risk buying a second catadioptric lens, this time the Olympus 500 mm f/8, for use on my Olympus E-M1 and E-M5 cameras? Four reasons:
As a whole, the large majority of image samples shot with catadioptric lenses and available on the Internet range from crappy to unimpressive (I am talking about image resolution and contrast, not about the donut bokeh that I discuss separately later on). In some cases, this is likely due to poor lens quality. In other cases, it is likely due to incorrect shooting technique (especially vibration) or poor shooting conditions. It is difficult to decide which is the exact cause of each individual poor image. For this reason, I decided to ignore poor image samples and to consider only those that range from acceptable to good. There are not many of the latter but, together with my earlier experience with a Sigma 600 mm f/8 and refractor telephoto lenses, enough to allow me to form an opinion about the average quality of at least some catadioptric lenses. What I have seen confirms the general opinions of Olympus and some Nikon being among the best commonly available 500 mm catadioptric lenses.
Olympus OM-System Zuiko Reflex 500 mm 1:8
The Olympus 500 mm easily mounts on a Micro 4/3 camera via an OM adapter (above). I chose a Metabones adapter because it has front and rear bayonets of chrome-plated brass, rather than the anodized aluminium of cheaper adapters. The Metabones adapter has a built-in Arca-compatible tripod shoe, very handy because the Olympus 500 mm doesn't have a built-in tripod collar. Since this lens has only manual focus and fixed aperture, it loses no functionality because of the adapter. The focal length (500 mm) must be manually entered in the camera configuration in order to allow a correct functioning of the in-camera image stabilization.
As far as I know, this is the only catadioptric lens made by Olympus for the OM film SLR system, and the last one they ever made for cameras with interchangeable lenses. The optical scheme uses five elements in two groups, and differs from typical catadioptric lenses in that the primary mirror is not drilled in its center. Instead, light passes through the non-silvered center of the primary mirror, which in this region acts as a refracting element. As a whole, the thickness of these optical elements, and especially of the two rear ones, is remarkably high. The primary and secondary reflectors of the Olympus 500mm are second-surface mirrors, and should therefore be well protected against degradation.
Light passes multiple times through most of the optical elements: three times in total through the large front element, twice through each of the two smaller front elements, and three times through the primary mirror. Only the small rear element located in front of the primary mirror is crossed just once by light. This means that, unlike simpler catadioptric lenses with front mirrors, the "unfolded" equivalent optical path of this lens is complex. Assuming that adjacent optical elements are made of different glass types, and counting the interface between cemented elements as a single optical surface, in this lens light crosses in total 13 optical surfaces, not counting the two reflections. This compares favorably with the large number of optical surfaces found in expensive super-telephoto refractor lenses. However, the actual number of manufactured optical surface in this catadioptric is only 10, and since light repeatedly crosses the same manufactured optical surface the degrees of freedom available to a lens designer are fewer than in a refractor. Thus, the optical design of this catadioptric is probably more difficult to fine-tune to the same extent as in a refractor.
At least some of the specimens of this model are multi-coated, as shown by the green reflections on the front element. I don't know whether early specimens were single-coated like other early Zuiko OM lenses, which switched to multi-coating after a few years of production. Single-coated specimens are more likely to display a lower contrast and higher sensitivity to flare than more modern ones. A small white stamp is present on the rear mount of at least some Olympus lenses made starting from the early 80s, and is present in my specimen.
The internal baffles that prevent stray light from entering the optical path from off-axis directions seem to be well-designed, although off-axis light that enters the lens has a chance of doing its mischief even before reaching these baffles. The rear element is surrounded by a conical baffle, carrying a velvet-like coating that absorbs stray light.
Focusing moves the front group away from the rear group. The barrel extends only about 5 mm between infinity and the closest focus. The whole front half of the barrel, including the lens shade, rotates while focusing. The minimum focusing distance is 4m, at which distance the focal length is shorter than 500 mm (probably around 450 mm). On Micro 4/3, the closest focus covers a subject area of 140 by 95 mm, which is a respectable close-up capability. Focusing slightly beyond infinity is possible, and is a design feature that allows a correct infinity focus even at very high and low temperatures. The total focusing range is covered in slightly more than half a turn of the focusing ring. This requires adjustments of only a fraction of a mm for precision focusing, which may be difficult or impossible to carry out while hand-holding this lens.
In general terms, the optimal lens aperture for the Micro 4/3 format is around f/8, and therefore it is not a major problem that this lens has a fixed aperture. With a variable-aperture lens on this format, one should not exceed f/11 to avoid loss of resolution due to diffraction. The edge of the secondary reflector slightly increases the negative effects of diffraction, but not by much.
Mechanically, this lens lacks a tripod collar. It is solidly built with metal barrel, but at 590 g it is not very heavy, and with a maximum diameter of 81 mm and length of 97 mm from the lens flange (in practice, close to 120 mm including front and rear caps) easily finds a place in a camera bag. Front filter thread is 72 mm. There is no rear filter mount. The size and weight are approximately average for a catadioptric 500 mm f/8.
As a comparison, the next longer telephoto lens in the OM system is a refractor 600 mm f/6.5, which at almost 2.9 kg and 380 mm barrel length does not fit in an ordinary camera backpack. Some of the difference in weight and length between the two lenses is due to the greater focal length and speed of the 600 mm, but most of it is simply due to the different optical design. Incidentally, many specimens of this 600 mm lens on the second-hand market suffer from molds, element separation and defective focusing mechanisms, while the 500 mm is less vulnerable to age problems and usually in good condition.
The Olympus Reflex 500 mm has a built-in lens shade that slides forward on the foremost part of the barrel. This lens shade extends by 24.5 mm forward of the filter mount. A third-party lens shade available for the Samyang 500 mm mounts on the filter thread and is 95 mm long. The Vivitar Solid Cat series has lens shades as short as 20-30 mm. The Nikon 500 mm (latest model) has a lens shade less than 20 mm long. At least some specimens of the Tokina 500 mm f/8, in spite of apparently coming out of the same factory that provided very similar models to many other brands, came with a screw-in lens shade almost as long as the lens itself. This extreme variability in the length of the lens shade deserves a discussion.
The function of a lens shade is to prevent off-axis illumination from entering the lens and causing flare and/or a reduction of contrast. In practice, the lens shade should be as long and as narrow as practical, without causing vignetting. Wideangle lenses need broadly conical or rectangular lens shades. Long telephoto lenses use cylindrical lens shades. The length of a suitable lens shade increases with the focal length, and with the diameter of the front lens element. A lens mounted on a camera with a small sensor (as is the case of my Olympus 500 mm, designed for 24 x 36 mm film but used on a 12 x 18 mm Micro 4/3 sensor) can use a narrower and/or longer lens shade than an APS-C or full-frame body without vignetting.
Among refractor telephoto lenses, the Olympus 600 mm f/6.5 has a built-in sliding lens shade approximately 180 mm long. The Nikon Nikkor 500 mm f/4 refractor has a removable carbon-fiber lens shade approximately 178 mm long. Other refractor telephoto lenses have even longer lens shades. Some Nikon super-telephoto lenses have enormously long lens shades built in two separate stages. One stage can be used when wind is a problem, while both can be stacked together for best performance against stray light.
Since the critical factor that determines the minimum suitable length of a lens shade is not lens barrel length but lens focal length, it is immediately obvious that most catadioptric lenses have inadequate lens shades. In my experience, inadequate lens shades have a large impact on contrast and flare, especially, but not exclusively, with older and less effective lens coatings. For example, the expensive and otherwise excellent Coastalopt 60 mm f/4 Apo lens suffers from an internal flare hotspot within a certain range of focusing distances. This problem can be completely cured by using an adequate lens shade.
With this in mind, I built an initial, makeshift lens shade for the Olympus 500 mm, in the form of a synthetic A4-sized sheet of lightweight, relatively stiff black synthetic foam. Wrapped around the built-in lens shade, it extends its length by about 18 cm. For testing, the makeshift lens shade was kept wrapped around the lens barrel with a couple of rubber bands. A more durable solution is to attach strips of adhesive Velcro around the built-in lens shade and along three sides of the rubber-foam sheet. This additional lens shade can be transported either flattened or wrapped around the lens.
Surprisingly, my outdoors tests showed no sign of flare or loss of contrast attributable to stray light or internal reflections. Even with the built-in lens shade retracted, I had no problem shooting in full sunlight. Perhaps there are special conditions, like shooting in the general direction of the sun, which could be a problem and may require the extra lens shade, but I simply did not encounter them. It may be useful to remember this possibility when testing other catadioptrics, but the Olympus 500 mm simply does not seem to need an extra lens shade in ordinary conditions of use. Perhaps its efficient multicoating and careful design, plus the fact that its front element is rather deeply seated within the front of the barrel, and also shielded by a projecting cover of the secondary reflector, avoid apparent problems.
Handling and focusing on Micro 4/3
A 500 mm lens is never going to be easy to use, especially on a Micro 4/3 camera where this lens gives the same field of view of a 1,000 mm on full-frame. Image stabilization helps, but shooting hand-held with this lens simply cannot be made easy. For one thing, locating a moving subject in the viewfinder with this lens is not trivial, and takes training and some luck. A solid tripod and dampened head greatly improve the success rate. It does not matter that suitable tripods and heads look stupidly oversized for this small lens. What matters is that its focal length is very real, and requires the same tripod and head that you would use with a 5 kg Nikkor or Canon super-telephoto refractor (even more so, in fact, since the low weight and length of a catadioptric lens cannot help to stabilize the system by adding weight and inertia).
When shooting far subjects, convection air currents rising from sun-warmed objects can blur images and produce unacceptable images. I shot the example of Figure 3 in mid-day sun in the Southwest US with a Panasonic 100-300 mm at the focal length of 280 mm and 1/250 s with image stabilization. This focal length is already enough to illustrate this problem. This lens is capable of better resolution and should produce sharp details in these conditions, but convection blurs the image (especially the farthest detail). In the viewfinder, the background could be seen to "boil" and shimmer.
A 500 mm would be much worse in these conditions. Shooting early after sunrise is usually the best option in this case. This problem is not specific to catadioptric lenses, and affects refractor lenses to the same extent. It is just one of the things a photographer must take into account when shooting with long telephoto lenses.
The whitish specks visible in the crop at the right might be either insects in mid-flight or dry grass lifted up by a weak wind.
No longer having to use an optical viewfinder is great, especially with long telephoto lenses. However, to correctly focus with manual-focus telephoto lenses, I am unable to rely on image sharpness in an unaided viewfinder (either electronic or optical). My eyes are simply not accurate enough, and with a 500 mm I can easily miss focus by a few cm and spoil an image. In general, the following methods are available for manual focusing:
In conclusion, viewfinder magnification is the only usable method with the E-M1. This makes tracking a fast moving subject impossible.
An initial test (Figure 4) was intentionally carried out in unfavorable conditions, and suffers from a series of problems, including vibration, movement of the subject by wind, and low light. The exposure was 1/15 s. The subject was approximately 7 m from the camera and swaying in wind gusts. The late hour and twilight illumination resulted in low color saturation and relatively low contrast. The lens was attached, via the Arca-compatible shoe of the Metabones adapter, to a Manfrotto 468MG hydrostatic ball-head on a relatively lightweight tripod with a central column (Gitzo 1227). The shutter button was pressed directly by hand (not via a remote control), which resulted in a very obvious "jumping" of the subject in the viewfinder. As a whole, this test is representative of the poor shooting conditions of many sample pictures available on the Internet.
Nisen boke is visible at the tip of the leaf near the top left corner of the image. It could easily be blurred away in post-processing. The bokeh of out-of-focus flowers is a bit "unruly", with some small donuts visible. Resolution is obviously poor when observed at 1:1 pixel ratio. There is enough dynamic range to increase both contrast and color saturation to pleasant levels. The unevenness of the green background in the 1:1 crop is an artifact caused by JPG compression for web publishing.
As discussed above, this lens-camera combination is extremely sensitive to vibration. Initial tests proved this beyond any reasonable doubt, regardless of whether in-camera image stabilization was used. To reduce this problem as much as possible, I mounted the lens on a heavy-duty cine head and one of my largest tripods (Figure 5).
The resulting combination of equipment may look ridiculously disproportionate. Believe me, it isn't. A tripod and head significantly heavier than these would actually be desirable. Don't forget that the small size and weight of the camera and lens worsens the stability problems.
In practice, I can hardly be expected to carry the above tripod and head in the field, for the same reason I don't want to use a 500 mm refractor. The most practical alternatives for field use are a significantly smaller tripod with its central column replaced by a flat plate (to reduce sagging and vibration) and/or a bean bag. I am simply forced to acknowledge that 500 mm is somewhat beyond the range of focal lengths I can expect to use in the field without losing some image resolution to vibration, and that the upper limit of this range is probably between 300 and 400 mm.
In the end, for this test I focused the subject at maximum preview magnification, which made it easier (but still difficult) to focus on the chosen detail, which is the nose/beak of the figurine at the right in the picture. I used an exposure time of several seconds at ISO 200 to let vibration caused by the camera shutter to disappear during the first couple of seconds, and shot the test pictures with a Wifi remote control (by mobile phone) from another room to reduce vibration to a minimum. I live in a wood house, and while standing two meter away from the camera I could clearly see, with maximum viewfinder magnification, that my small muscle adjustments required to keep my balance were causing the floor to vibrate and to transmit this vibration to the camera. The unfortunate truth is that only by staying in a different room I could make a test that reveals the actual resolution of a lens of this focal length.
I would rate this image resolution between good and excellent, considering the difficulties discussed above. I can get a higher resolution with some of my best lenses of much shorter focal lengths, but this is not a useful comparison in terms of the amount of detail I can record on a subject I cannot approach physically.
A more useful comparison is with an image shot with my longest native Micro 4/3 lens I own at this moment, the Panasonic 100-300 mm at 300 mm and f/8 (Figure 7, top and middle). To compare the two lenses, I took the test image of Figure 7 and reduced its resolution by bicubic interpolation to achieve approximately the same field of view across the 1:1 pixel crop. The purpose of this exercise is to answer the question: which of the two lenses provides more visible detail? The answer is not easy. Color and saturation are better with the Panasonic 100-300, but this image looks a bit "oversharpened", perhaps as a result of axial chromatic aberration and/or other aberrations. The Olympus 500 beats the Panasonic 100-300 in terms of lack of haloes around the black nostrils and eyebrows. There is also a more "relaxed" and "natural" character in the Olympus 500 mm image. Pixel-peeping comparisons are made difficult by the fact that the two images differ by a few mm in the exact plane of focus. As a whole, I prefer the image from the Olympus 500 mm.
A further possible test is taking the image shot with the Panasonic 100-300 and increasing its size by interpolation, to make its 1:1 pixel crop similar to the Olympus 500 mm pixel crop. When this is done (not shown), the Olympus 500 mm image is naturally better, but this is only to be expected because the original 1:1 pixel crop of the Olympus 500 image contains more pixels to start with. In this sense, the Olympus 500 mm is better than the Panasonic 100-300 mm in recording more detail, and the extra magnification provided by the Olympus 500 mm over the Panasonic 100-300 is real, not empty magnification. However, the difference is not very substantial. On the other hand, a 40% difference in focal length between two lenses is not enough to record substantially different amounts of detail, even assuming both lenses are of the best possible quality.
The Tamron 300 mm f/2.8, stopped down to f/5.6 and used with a matched 1.4x focal length extender, does record more detail than the Olympus 500 mm, but at the price of five times the weight and size. Aside for the Olympus 75-300 mm, which is comparable in resolution to the Panasonic 100-300, there are no other 300 mm native Micro 4/3 lenses worth considering. The Olympus 4/3 300 mm f/2.8 is almost certainly at least as good as the Tamron 300 mm f/2.8 and provides a usable autofocus with the E-M1, but it is also too big, heavy and expensive for my use. Olympus is expected to release in 2015 a Micro 4/3 300 mm f/4 in the Pro series. If this lens turns out to be of better quality than the other, already excellent Olympus Pro lenses (which should be expected of a prime lens vs. zooms), I might need to revise my positive opinion of the Olympus 500 mm. This is not strange, considering that the Olympus 300 mm will probably cost at least five times more. In the mean time, the Olympus 500 mm remains my best portable long telephoto lens for the Micro 4/3 format in terms of the detail it records.
A few tests shot with the lens handheld at 1/2000 s and 400 ISO in full sunlight show a remarkable amount of detail. In fact, the 1:1 crop in Figure 8 is definitely better than my careful indoors tests. Precision focusing while hand-holding is a hit-or-miss affair, but when I happen to precisely nail the focus (more by accident than by skill), these pictures are about as detailed as any I get with my best lenses. Therefore, the evidence is forcing me to concede that the Olympus is capable of excellent image resolution, and that any results worse than the one in Figure 9 are due to my failure to focus correctly or eliminate vibration.
Udi Tirosh published an interesting way to focus a longer (and therefore very sensitive to defocusing) catadioptric lens, by using a guitar tuning key to rotate the focus ring by applying friction to it. I have never looked closely at guitar tuning keys, but apparently they contain a demultiplying worm-gear and in this way allow movements of high precision.
The Dreaded Donuts of Doom
All catadioptric lenses, except oddities like the Makowsky Katoptaron, have coaxial primary and secondary reflectors. As a consequence, the entrance pupil has the shape of a doughnut. The shape of the entrance pupil or diaphragm (whichever is smaller) of a lens determines the shape of out-of-focus highlights. Therefore, out-of-focus highlight points are rendered as doughnuts by catadioptric lenses, as mentioned above. There is simply no way around it, and the Olympus 500 mm is no exception.
Actually, this is not completely true. Since catadioptric lenses have no variable aperture, one can mask a part of the front element to lower the lens speed. A mask with a circular hole about 20-25 mm wide, located close to the outer edge of the front element in order to stay clear of the secondary reflector, transforms a 500 mm f/8 into an f/20-f/25 lens. This aperture is too small to be usable (diffraction spoils all fine detail), but the bokeh of out-of-focus highlights becomes circular. The lens also becomes optically equivalent to a Katoptaron, so if you want to investigate the imaging characteristics of this type of lens design you don't need to buy one. Using a larger hole in the mask is possible, but part of it is eclipsed by the central stop, and the bokeh shape becomes rather odd. If you really want to be finicky and your subject is static, you can shoot one image with the mask and one without, and combine in post-processing the unfocused background with circular bokeh, shot with the mask, together with the sharp in-focus subject shot without mask.
The same post by Udi Tirosh mentioned above also shows how to build a makeshift variable aperture mask that does as described in the preceding paragraph. Test images are also shown. As expected, the f/22 test image shows a degradation of resolution by diffraction, but images up to f/16 are sharp even though the aperture is sausage-shaped instead of circular. The photographer used an APS-C format camera, and therefore f/16 is still usable, while a Micro 4/3 camera should not be used past f/11.
Spherical aberration makes the bokeh of a lens in front of the focus plane different from the bokeh at the rear of the focus plane. Some catadioptrics do display this type of aberration. On either side of the focus plane, the donut highlights can look different. Often, they are more star-like on one side and more ring-like on the other. The two types of donuts also respond differently to blurring in post-processing, which is the simplest way to get rid of donut bokeh if someone objects to it. In Photoshop or an equivalent editor, select out the well-focused subject area, which you don't want to blur. Now only the out-of-focus areas with donuts should be selected. Gaussian blur of medium intensity and rather large radius usually works on small donuts. Increase the radius until the donut holes fill up, but do not exaggerate. Large donuts can be stubborn against Gaussian blur. In this case, after applying a reasonable amount of Gaussian blur, select only the areas with still-visible donuts and apply a stronger blurring filter (e.g., average). The result is a creamy bokeh without marked highlights, which someone might find artificial. This is of course nonsense, since all types of bokeh are artificial. A side effect of this post-processing is that both foreground and background become blurrier than in the original pictures. If someone asks about it, explain to them that this is simply the way these lenses work, meaning with this super-fast, super-heavy and super-expensive super-telephoto lenses. Alternatively, tell them that you have been testing a lens prototype and are not allowed to say anything about it. Let them pore over the metadata as much as they want: this is a manual lens and the metadata gives no clue about the lens type.
It is perhaps possible to design a post-processing bokeh filter that identifies donuts and nisen boke, and fills their core with the same luminosity of their edges, and in this way simulates the bokeh of refractor lenses. I am not aware of such a commercial filter. All the commercial bokeh plugins I have seen require a manual selection of the subject or background. Some can use a depth map, where areas to remain in focus are black, areas that will be highly blurred are white, and intermediate areas where blurriness is supposed to vary are shades of gray. The shades of the depth map then control the amount of blurring applied to each point of the image by the algorithm.
Now it is time to test the performance of the Olympus 500 mm with regard to donut bokeh. Figure 9 shows a series of pictures starting from focus at infinity, then gradually focused on the yellow flowers at the center of the image, and finally closer and closer.
The most important thing to note is that, with this lens, the bokeh in front of the focus plane is identical to the bokeh at the rear of the focus plane. This means that the Olympus 500 mm is very well corrected against spherical aberration. This should also make it easier to eliminate the donuts in post-processing, since both foreground and background can be processed in the same way. The regular, circular shape of this bokeh allows a simpler manual elimination of the donuts, if there are few of them in an image, by filling their centers with samples cloned from their periphery.
This test also shows, as expected, that this lens cannot "magically" avoid the donut bokeh. Finally, there is no darkening of the corners attributable to vignetting. This result is expected on a Micro 4/3 sensor, but nonetheless a nice confirmation that vignetting is negligible with this combination of lens and camera.
Is it worth it or not?
The main question that remains to be answered is whether it is worth using one of the best catadioptric lenses, like the Olympus 500 mm, on a Micro 4/3 camera. In brief, the advantages are:
And the disadvantages:
As a concluding reflection on catadioptric lenses, their usually low image quality may be an intentional design choice, rather than an accident of sloppy manufacturing. Since these lenses are very difficult to use properly, their designers were aware that most amateur photographers would have been unable in any case to get the most out of these lenses, and therefore would not be able, in practice, to see a difference between a cheap dog and a more expensive but significantly better lens. Famous camera brands, on the other hand, were able to use their market leverage to sell more expensive catadioptrics, and at the same time also had a reputation to defend. Therefore, they were more likely than third-party manufacturers to design better-than-average catadioptrics.
The Olympus OM Zuiko Reflex 500 mm f/8 catadioptric lens produces a far better image quality than the cheap catadioptrics that have been flooding the market for the past five decades. It is, in practice, just as good as a 500 mm refractor lens from one of the major camera brands, but price, weight and volume are several times lower. However, it is extremely difficult to focus with the precision required to bring out the most from this lens. I expected the built-in lens shade to be too short, but my tests did not confirm this. The Olympus 500 mm records finer detail than the Panasonic 100-300 mm zoom at 300 mm. In other words, the magnification provided by the extra focal length is real, not empty magnification.