[Originally published in four parts under the title "How to Stress a Camera Lens" in 2008. This newly re-edited republication is Parts I and II.
Copyright 2008, 2014 by Michael C. Johnston. All Rights Reserved.]
Part I. How Not to Stress a Lens
Most lenses in history, and especially now, are capable of excellent results—as long as they aren't stressed. In a moment I'll talk about how to "stress" a lens (i.e., deliberately explore the boundaries of its performance in order to expose its weaknesses). But first, here's how to see the best performance of your lens under optimum conditions:
1. Shoot in strong white light with the camera pointed away from the light source, with no specular reflections or light sources in the picture and no impinging light shining directly on the objective (outermost lens element). Use at least a short lens hood to block light impinging from a radical angle.
2. Use the widest aperture at which aberrations are minimized, which usually involves stopping the lens down two or three stops from wide open.
3. Use a fine-grained, low-contrast pictorial film, or your digital imaging sensor at its optimum or "base" sensitivity ("ISO"). With many digital cameras, the base ISO is around 200, even if the camera can be set at lower values.
4. Minimize blur caused by subject and camera motion—this might be accomplished with the use of a tripod or other means of mechanically steadying the camera—and arrange your subject so as to minimize both foreground and background blur.
5. Don't focus too close—lens designs vary, but perhaps stick within a magnitude of 2 or 3 of (50 • ƒ) (where ƒ = the lens focal length). Thus, for a 50mm (2-inch) lens, you would want to place your main subject perhaps 50 to 200 inches (~4–16 ft.) from the camera.
6. Focus carefully. Focus errors are a very common cause of image degradation—much more common than most people realize. (I see focus errors in published work and "great master" photographs regularly.)
7. Don't enlarge the resulting picture in the print or onscreen past a modest limit.
8. (Digital only) to minimize "chromatic aberration" (i.e., purple fringing), don't ask your lens to image complex, fine-lined subjects with high-contrast edges, such as bare tree branches against a white sky, especially near the edge or corners of the frame.
9. (Zoom lens only) use an intermediate setting in the zoom range. Generally, a kit-type zoom will show its strongest performance if you zoom in roughly a third of the way in from its widest setting. (Even though the actual optimum setting will vary from lens to lens, this will usually be close enough.) Doing this will mainly ameliorate geometrical distortion, and falloff (also called "vignetting").
10. Make sure the lens is clean.
With these ten conditions met, you'll probably find that your lens is a very good performer indeed. Even lenses that are very far from state-of-the-art (older budget 50mms, "kit" zooms) will perform well. This includes many very old lenses and lenses with significant budget compromises, or even the lens permanently attached to an inexpensive consumer camera. Some of the prettiest prints I ever saw were pictures made with a tripod-mounted single-coated Miranda lens from the 1960s. They just happened to be made under conditions pretty close to those described above.
Part II. How to Stress a Lens
"Stressing" a camera lens basically involves moving incrementally away from these ideal conditions, such that the lens will be less and less likely to maintain peak performance. Here are a few of the ways this can be done.
Flare is the effect of non-image-forming light on a picture. It takes two basic forms: ghosting and veiling glare. "Ghosting" refers to reflections or any kind of spurious visual artifact; veiling glare is an overall softening or contrast-cutting effect. Flare might also have numerous other causes, including but not limited to reflections within the camera or lens and light leaks in and around the film chamber or sensor plane, or perhaps impinging from the unblocked viewfinder window; or such effects as "starbursts" around bright sources of light in the picture.
Although generally perceived as undesirable, flare properties can create expressive photographic effects. Here is a good example of both kinds of flare. (Note the illuminated strands of hair across the face that "survive" the veiling glare.) If you aren't accepting of such fortunate accidents, however, a lens that flares as badly as this one did could be quite frustrating, or limiting.
I have three basics trials for lens flare. One is to make pictures with a bright light source in it, such as the sun. Especially with that light source off-axis, you can get a handle on the lens's common ghosting artifacts. A second is to place the sun or source of light outside the picture area, but let it shine on the objective (outermost element) of the lens. This gives you an idea of how susceptible the lens is to veiling glare and might give you clues as to whether internal reflections in the lens are a potential problem. The last is what I call "The Window Test"—with the lens opened up as for an interior scene, place a large source of much brighter light, such as a large window illuminated by the sky, just outside the picture area. Look especially at the contrast of fine detail in the interior scene. Some lenses which otherwise handle the causes of flare with aplomb will flunk the Window Test (notably including the otherwise excellent Leica 35mm Summicron-R designed by Walther Mandler); others will flunk it at certain apertures and not others, something that's good to know.
A common characteristic of camera lenses is for their performance to vary at different apertures. As mentioned above, most lenses have a one- or two-aperture "sweet spot" where aberrations are minimized and diffraction effects are least prominent. With 35mm prime (single focal length) lenses this is almost always within a stop of ƒ/8; with many double-Gauss view camera lenses it's within a stop of ƒ/22. With digital cameras that have smaller than "full frame" sensors, the optimum aperture will be wider. It's useful to know what the optimum aperture of your lens is.
Most lenses will be stressed by moving away from this optimum aperture. Diffraction effects can lower the overall resolution of a lens at small apertures, and of course aberrations of various kinds tend to be less well corrected at wider apertures.
Many of the properties of a lens at various apertures are "designed in," so to speak. They are the results of what might be called the designer's "philosophy," i.e., the tradeoffs selected where desirable parameters oppose. One aspect of this is whether the lens is optimized for best performance in the axial zone (center), allowing the performance to degrade as image height increases (i.e., toward the corners), or whether the choice is made to accept a somewhat lower but more consistently even performance from center to corner. Another design choice that relates to the speed of the lens (its maximum aperture) is that the designer might go for the highest possible performance at the optimum aperture and let performance fall off relatively more dramatically as you move further away from that aperture in the lens's range, or the designer might try for more consistent performance throughout the aperture range.My own preference is for lenses that show consistency, both across the frame and up and down the range of aperture settings. That is not everybody's choice; I remember for instance that when Lewis Baltz was doing his 35mm pictures of tract houses and industrial parks, like those in his book Park City, I heard that he had tested the Summicron (ƒ/2) and Summilux (ƒ/1.4) lenses then available, and found that although the Summicron was better wide open and was more consistent throughout it aperture settings, the Summilux reached the higher performance at optimum aperture. Baltz chose it for that reason, since he anticipated using a high-resolution film with his rangefinder camera on a tripod.
Consistency of the sort I prefer is relatively easy to detect from tests, if you have a family of MTF curves to interpret, or from the useful new "Blur Index" reports at SLRgear.com. In a lens test at SLRgear.com, if you click on the "Blur Index" panel, it brings up an interactive window that shows you a graphic of sharpness at each aperture. The lower (more towards zero or violet color) and the flatter this graphic representation is, the better. If you look quickly through the tests of a couple of dozen different lenses, you can see how it tends to go: wide open, lenses tend to be worse in the corners (that is, the test graphic is more bowl-shaped) and higher (less sharp), then it flattens and lowers as you move to ƒ/8, then it rises (while staying flat, more or less) as the aperture is reduced further (this shows the sharpness-deteriorating effect of diffraction). As an example of this, look at the Canon 28mm ƒ/1.8 (the uneveness in the concave graph of this lens wide open, with two corners higher than the other two, is probably the result of decentered lens elements in the tested sample). A consistent lens of the type I prefer will show a flat, low plane wide open that doesn't change much as it is stopped down, only rising somewhat as the aperture is stopped past ƒ/8 and diffraction kicks in. As a superlative example of this, look at the graphic for the Olympus Digital Zuiko 50mm ƒ/2 Macro. This lens's best aperture is ƒ/4, but, as you can see, it is only marginally worse at ƒ/2.
You can't tell everything about a lens from lens tests—most tests tell you nothing about flare and bokeh, for example—but a lens's consistency from center to corner and up and down the aperture range is one thing you can learn from tests. The general rule, however, is clear: most camera lenses are at their worst wide open. To most effectively get less than optimum performance from a lens—to "stress" it in the way I've been using the term—use it wide open.
Years ago I used a camera with a fixed ƒ/4 lens, the Fuji GS645s, but the camera was severely limited, not so much by the modest maximum aperture as by the fact that that aperture was unusable—the Fuji needed to be stopped down at least one stop to reach decent optical performance, and two stops to reach good performance. Although its maximum aperture was ƒ/4, its maximum effective aperture was ƒ/5.6, with ƒ/8 considerably better still, meaning that it was essentially a daylight or tripod camera—not the happiest situation for a camera that was designed to be light, portable, and hand-holdable. (Picture courtesy pbase.com Camera Database)
A lens which reaches its optimum aperture when wide open is said to be "diffraction limited." The term signifies a lens for which diffraction is the worst cause of image degradation at full aperture, with diffraction effects exceeding the deleterious effects of all the other aberrations. You can also speak of a lens being diffraction limited at a certain ƒ-stop—"diffraction limited by ƒ/8," say—but that's really just another way of saying that the specified aperture is the lens's optimum aperture.
Curvature of field
Another "designed in" property of lenses at specific apertures is curvature of field—that is, when a lens's plane of focus is not a plane. This is a difficult property for most photographers to test, and it's generally even more difficult to recognize its visual effects in pictures. Take a look at this example (from Danny Wolpert, the same photographer as the earlier example I used—he seems to enjoy exploring lens effects. Thanks and props to Danny). It's an excellent example of the visual effects of curvature of field. Notice how the bokeh or degree of out-of-focus blur is varied depending on where in the frame it is? In the center of the frame, the yellow flowers are quite blurred only a half dozen paving stones past the gate. But the tree leaves and plants in the upper right-hand corner are considerably less blurred than you would expect them to be given that they appear to be considerably farther away from camera position. This is a visible effect of curvature of field. In this picture, the effect is not consistently what I would expect to see, leading me to to suspect that this particular picture has been cropped by the photographer (although I might be wrong about that). (I was also fascinated to see such extreme curvature of field from this particular lens, which was in some respects quite highly corrected for its day.)
Although detecting curvature of field is often considerably complicated—even masked—by complex pictorial subjects, it's possible to train yourself to see it (although speaking from personal experience I would recommend that you not do so! Sometimes it's best to let sleeping dogs lie).
Lenses can be optimized for different focus distances. (This is one of many reasons why standardized tests often reach not-quite right conclusions from correct data.) Many camera lenses are (sensibly enough) optimized for an intermediate focus distance. To stress it, go to the extremes—infinity focus or the lens's close-focus point.
Some older lenses are optimized for infinity. Typical lenses, including but not limited to these, will usually be at their weakest at their closest focus. I've even used a few "Macro" lenses that are not actually optimized for the closeup range—the superb Olympus OM Zuiko 50mm ƒ/2 Macro (note: the film version, not the same lens I mentioned earlier) is one of these—thankfully, since that made it better as a general-purpose normal lens.
Bokeh just means out-of-focus blur, as distinct from the blur caused by a moving subject or a moving camera (i.e., camera shake) or both. The further an image object is from the plane of focus, the less clearly imaged it will be, and the properties of the blur vary from lens to lens and can have varying aesthetic impact.
Bokeh-aji ("the taste of blur," literally) is indeed a matter of taste, but, very broadly speaking, photographers tend to want the in-focus subjects to also be the focus of the viewer's attention, and therefore prefer out-of-focus blur that is unobtrusive and doesn't call attention to itself. Just as a single lens has a whole "suite" of imaging properties depending on its settings and how it's being used, so no single lens has bokeh properties of only one simple description. Bokeh can differ depending on which side of the plane of focus it's on (front or back), and generally bokeh is worse:
- the closer the focus;
- the wider the aperture;
- the farther from the plane of best focus the image object is; and
- the higher the contrast of the imaged object.
So, to stress a lens for bokeh, use it at wider apertures and closer focus distances with high-contrast imaged objects on both sides of the plane of best focus and as far as possible from it. (Note that this fairly describes photographic situations typically confronted in poorly-lit interiors, where, indeed, lots of lenses come up short.)
Generally, high degrees of enlargement expose the morphology of film grain, the technique of the darkroom worker or digital printer, and make the optical shortcomings of the lens's imaging properties more apparent. Obviously, every digital photography enthusiast knows this, as "pixel-peeping" (Michael Reichmann's term) at 100% to 400% onscreen is so easy to do and so tempting to do. Thus, to stress a lens, over-enlarge its image.
Digital photographers are only too aware of the phenomenon of purple fringing, often not-quite-correctly referred to as "chromatic aberration" (purple fringing is an artifact of the lens and the sensor together, not just one or the other). How to stress a lens for this defect is well known: tree branches against a bright sky. The defect can show up in many other situations too, but that's the paradigm.
(A related phenomenon is "color replacement," where a dark cable or branch against a bright sky will appear to be a different color throughout than the same object looks against a less contrasted background. Dark blue-gray instead of black, for instance.)
This illustration, unmanipulated, shows a detail of about 1/8th of the frame from the upper corner of a picture made with the Konica-Minolta 7D and the otherwise excellent Sigma 30mm ƒ/1.4 lens. Note that purple fringing is the result of interaction between a lens and a sensor, and should not be considered a definitive characteristic of the lens alone; on other cameras this lens might not yield the identical result. However, since this was my camera at the time, I opted not to keep this lens. As this defect can be corrected in software, some users might not be as troubled by it.
Hey! No fair
Moving the camera or misfocusing it are surefire ways to degrade a lens's performance, but they can't really be said to stress the lens in the sense that I've been using that term here. A lens can't do anything about being misused, and misusing it doesn't tell you anything meaningful about its inherent properties. But I hope it goes without saying that a picture blurred by camera shake isn't going to reveal the lens's best performance.
An important thing to remember here is that—as with misfocusing—the effects of camera shake can be very minimal and hard to detect on casual viewing, while still very substantially degrading lens performance. The better you know your lens, the easier it will be to see when you haven't quite let it shine.
Another property "built in" to the lens when it's designed is distortion (simply put, whether it renders straight lines in a subject as straight—especially apparent when they're parallel to, and close to, the picture's edges). With primes, the distortion is what it is—nothing you can do about it. With zooms, however, it's another story. Generally, moderate-range zooms will progress from barrel distortion (sometimes pretty severe) at the wide end, to pincushion distortion (thought usually not very severe) at the tele end. Somewhere in the middle there is a "most neutral" setting for distortion. Usually.
Avoiding the extremes is usually good enough. It's been my observation that a fair number of people tend to use zooms as essentially a dual-setting prime lens, throwing the zoom ring from one end to the other and seldom using the in-between settings. Optically that's perhaps not the best practice, as zooms are typically at their worst at the extremes of their range. A better practice is, when you can, to go all the way to the wide or tele end and then back off from that setting just a little. Obviously that's not always practical, since you may actually need to go wider than the widest setting or longer than the longest, in which case you will need to use all the range you have available. But a lot of times it is practical, and it's not a bad habit to get into.
Zeiss 35mm ƒ/2 Biogon
The goal of learning all this is to get to know your lens. The story may be apocryphal, the speaker supposedly a Leica lens designer from back in days of yore: "The only way to test a lens is to use it for a year." That is, "lens tests," no matter how factual or scientific, are a shortcut.
Many amateur photographers conceive of their lens kits in terms of wanting to "cover" all the focal lengths or "be able to handle" any actual or imagined situation. They also like toys. Therefore they tend to overbuy lenses and have too many in their arsenals. One common result: under-familiarity with their own lenses, especially since amateurs may not shoot very much or very often. My feeling as a teacher is that one of the best things a hobbyist photographer can do to improve their seeing and their pictures is to limit the focal lengths available to them and use fewer lenses rather than more. Cartier-Bresson got by with one lens (although he often carried three, he only very seldomly used anything but the 50mm) and Sebastiao Salgado used three when he was making his early, formative work. You need more?
My preference and my habit has always been to use one or two lenses at a time. Two of the advantages of this approach is that it might allow you to invest in a relatively better lens(es), and you can get to know the lenses you have very well—two things which tend to feed off of and reinforce each other.
Copyright 2008, 2014 by Michael C. Johnston. All Rights Reserved.
Original contents copyright 2014 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site.
(To see all the comments, click on the "Comments" link below.)
Featured Comments from:
Moose: "An interesting and useful article. As Stephen Gillette so aptly demonstrates, it only applies to some uses of lenses. To most of mine, but not all.
"It seems to me that there is a piece missing, though, perhaps as a result of updating an older piece. Many contemporary lenses rely on post capture correction of various aberrations in the camera's JPEG engine or in the RAW converter. This post processing is an integral part of the lens design process. Knowing, for example, that linear distortion will be corrected elsewhere, the lens designer(s) are free to relax that design constraint at least somewhat and more fully correct other aspect(s) of the optical design.
"However, not all RAW converters are equal. For example, Olympus Viewer 3 and ACR, using data provided by the lens, quite noticeably under correct barrel distortion of the Pannasonic 12–32mm (The very compact kit lens for the GM1 compact Micro 4/3 camera) at 12mm and close focus. On the other hand, DxO Pro, using their own profiles, correct it extremely well. PTLens is about as good, used on a file converted from RAW without distortion correction. For these lenses, one may need to learn not only the lens, but the software used to convert it from RAW."
Mike replies: Well said, and thanks for the very pertinent addendum.