home
contents
copyright
donations
contact


3.7 Sharpness and Bokeh

In this section we’ll deal with two commonly cited aspects of image quality for artistic bird photos: sharpness, and bokeh.  The former, sharpness, should be a fairly intuitive concept, though there are some aspects of lens design (and lens operation) that can potentially lead to variations in sharpness, and we’ll discuss these first.  The issue of bokeh—the quality of having a smooth, pleasing, and non-distracting background—will be addressed later in this section.


Sharpness

Assessing the relative sharpness of two images, when they’re shown side-by-side, is usually pretty easy as long as they’re images of the same subject or scene.  Assessing whether a lens is
sharp, or sharp enough, based on images seen in isolation can be much more difficult.  One problem is that what really needs to be measured is not the sharpness of the lens, but the sharpenability of images produced through the lens under ideal conditions.  In other words, if we were to take some bird photos using a given lens, would we be able to sharpen the photos enough in Photoshop (or similar software) so as to produce usable images?  Of course, the answer depends on the intended use of the image: whether for a web page, or for an 8"×10" print, or for a 40-foot billboard.  The larger renderings (i.e., the billboard) will obviously require more fine details in the image file, in order to still appear sharp after enlarging.
    To illustrate the potential for differences in sharpenability between lenses, consider the two hawk photos taken below.  The left image was taken with a 1200mm lens (actually, a 600mm lens with a 2
× teleconverter attached—but that’s not really relevant at the moment), while the right image was taken with a 600mm lens with no teleconverters attached.  Both images are shown at 100% crop, which means that you’re seeing the native resolution of the camera (in other words, each pixel on your screen represents exactly one pixel in the image file; only a tiny portion of each image is shown, however, as the word crop should suggest). Because the left image was taken using a 2× TC, it should be much less sharp than the image on the right—meaning that fewer fine details of the bird should be visible, and any that are visible should be less distinct and more fuzzy.   In this case, since we haven’t applied any sharpening yet, the difference in sharpness between these two images isn’t terribly striking, though the left image may appear a little fuzzier than the image on the right.




Fig. 3.7.1: Red-shouldered Hawks (Buteo lineatus) at 100% crop. 
Differences in sharpness are difficult to see, without sharpening
in post-process.  Left: 1200mm, f/10, 1/200 sec, ISO 640.
Right: 600mm, f/9, 1/250, ISO 320.


    To demonstrate the differences in sharpenability between these two lenses, we’ve applied an identical amount of sharpening to both images, via Photoshop, with the results shown in the figure below.  Notice that the image on the right now exhibits many more fine details than the image on the left.  These are especially visible in the bird’s lore region (the area between the eye and the beak, where the feathers form a swirling pattern), in the edge around the eye, and in the texture of the beak.  Because of these differences, we’d say that the image produced through the second lens is more sharpenable, and lends some evidence to that lens being fundamentally sharper than the other.




Fig. 3.7.2: Same images as in the previous figure, but with
identical amounts of sharpening applied in Photoshop.
Now the image on the right appears sharper (as it should,
since the image on the left used a 2
× teleconverter).


    Examining the properties of images at 100% crop, as we’ve just done, is often referred to as pixel peeping, since we’re looking at the actual pixels of the image file.  Although pixel peeping can reveal differences between lenses, it’s not always useful for assessing the practical value of a lens.  If you just want to make small prints (8"×10" or thereabouts) of your images or share them via a web page, it’s probably more useful to assess image quality at a smaller crop level than 100%. 
    In the figure below are the same two hawk images as above, but at 25% crop, and with some additional sharpening and contrast adjustments made to each.  Because these are 25% crops, each pixel in the images below corresponds to multiple pixels
(16, to be exact) in the original image file—i.e., we’ve zoomed out, and are no longer seeing the individual pixels of the image file.  Instead, Photoshop is averaging small groups of neighboring pixels from the original image to produce virtual pixels to display on your screen.  As a result, it’s much more difficult at this crop level to see differences in sharpness between the two images (and, by implication, between the two lenses).




Fig. 3.7.3: The same hawks, again, but at 25% crop.
Sharpness and contrast were adjusted in both images.
For small prints or web use, the differences in sharpness
between these lenses can probably be ignored
.

    Which crop level is best for comparing two lenses depends on how often you’re likely to use each crop level when
publishing your images (i.e., when making prints, or when posting photos on a web page).  If you typically photograph large birds close-up and are using a large-focal-length lens, then you probably won’t be cropping at all, whereas if your passion is tiny birds like warblers and hummingbirds, and you’ll be viewing them at a distance or through a moderate focal-length lens, then you may indeed find yourself aggressively cropping your images in order to make the bird appear large in frame.  In the latter case, assessing the sharpness of the lens via pixel peeping makes more sense.
    In terms of assessing image sharpness digitally—i.e., on the computer screen—there are a number of important issues to consider.  First, most cameras employ an antialiasing filter which sits over the imaging sensor, and which will to some degree decrease apparent image sharpness.  For this reason, some sharpening in post-process is always recommended for cameras having such a filter, to counteract the slight blurring effect that the antialiasing filter imposes on the RAW images.  That’s one reason why we introduced the notion of sharpenability above.  Second, if you’re not shooting in RAW (and you probably should be), then it’s important to realize that your camera may be applying some sharpening behind the scenes to the JPEG file that it’s exporting.  This is more of an issue when comparing the resolution of different cameras, but it can become relevant for lens comparisons if the amount of in-camera sharpening is such that it ends of masking slight differences in lens sharpness.
    Another very important issue is that apparent sharpness tends to increase with smaller apertures (i.e., larger f-numbers)—at least, up to a point.  As we mentioned in section 3.1, the depth of field (DOF) increases with decreasing aperture.  Thus, for smaller apertures (like f/11), a wider margin in front of the bird and behind the bird will appear to be in focus than at smaller apertures (like f/4).  To the extent that autofocus doesn’t always perfectly set the focal point exactly on the bird, a larger depth of field can help to mask these slight focusing errors, since a slightly mis-focused bird that is wholly contained within a large DOF will still look mostly in focus.  This is illustrated in the figure below.




Fig. 3.7.4: Depth-of-field changes with aperture, and can
mask focusing errors.  Top: the lens is focused slightly in front
of the bird.  Because the aperture is wide, the depth of field is shallow,
and the bird will probably look fuzzy.  Bottom: a smaller aperture
produces a wider DOF, and now the bird should appear less fuzzy.

In the top part of the figure (above), the lens’ iris is opened to its widest setting (f/1.4 in this case), corresponding to the shallowest depth of field.  In this example, the lens is slightly front-focused—notice that the black dot (what we’ll informally call the
focus point) to which the red arrows converge is slightly in front of the bird.  The orange gradient surrounding this black dot represents the depth of field; the fact that the orange fades out gradually represents the fact that the in-focusness of the points around the focus point also gradually fades out at increasing distances from the focus point. 
    In the bottom part of the figure, we’ve stopped down the lens (meaning that we’ve reduced the aperture) by closing the iris a bit (to f/2.8), resulting in a wider depth of field.  Now the orange gradient (representing points that appear more-or-less in focus) extends further to either side (forward and backward) of the focus point.  Since the bird is now more completely contained within the darkest part of the orange gradient, it should appear much more in-focus than in the top part of the figure.  Thus, by decreasing the aperture, we’ve reduced the effect of the focusing error (i.e., fact that the black dot wasn’t positioned exactly on the bird).
    The reason depth-of-field increases with smaller apertures (and also with distance to subject) is explained in Box 3.7.1, below.  Feel free to skip the box if you’re not interested in understanding why DOF works the way it does.  Just note one additional thing: at some point, reducing the aperture can actually decrease sharpness rather than increase it, via diffraction.   Diffraction limits typically don’t become extreme until the aperture drops below f/16 or f/22 on most lens/camera combinations, so it’s usually not an issue for practical bird photography.


 Box 3.7.1: Why DOF increases with smaller apertures

The reason depth-of-field increases as the aperture is stopped down can be understood in terms of simple geometry.  Note that in the figure above, as the aperture was reduced, the cone of light converging at the eyeball became narrower.  Imagine what would happen if the eyeball was moved slightly closer to the lens: it would intrude into that cone of light, and what would be perceived wouldn’t be a single point, but a slice of the cone—i.e., a circle.  Thus, positioning the focal plane (i.e., the imaging sensor) too close to the lens would cause light rays from any point on the bird to scatter over a circular area, rather than focusing to a point.
    Now, if instead of the sensor being too close to the lens, the bird was slightly in front of the focus point, a similar thing would happen: the cone of light coming from any point on the bird would be focused to a point just behind the sensor plane, and once again we’ve got a point on the bird giving rise to a circle of illumination (rather than a point) on the sensor.  If the circle is very small—say, about the size of a photosite (pixel) on the sensor—then the bird will still seem to be in focus.  For circles that are slightly larger than this (i.e., if the bird is a bit further from the true focus point), there will be some
smearing of color around each pixel, but at small crop ratios it will still look fairly well-focused.  For yet larger deviations from the focus point, the bird will start to look progressively fuzzier in the image formed on the sensor plane.
    As noted above, however, reducing the aperture causes the light cone to become narrower, which means that the rate at which the circles grow in size, as the bird moves away from the focus point, decreases.  As a result, larger focus errors become more tolerable at smaller apertures, with the criterion of
tolerable being effectively defined by the photosite size.
    You may still have one lingering doubt, however—namely, how is it that less light (resulting from a smaller aperture) can produce more information (in terms of the level of detail visible in the image of the bird)?  The answer to this conundrum has two parts.  First, whenever you reduce the aperture, you invariably end up compensating for the reduced light admission by changing the shutter speed (or, alternatively, the ISO setting), so as to maintain a properly exposed image.  Thus, rather than
throwing away light, stopping down effectively just redistributes light into a narrower cone persisting over a longer time interval.  Finally, in terms of the apparent increase in bird information content (i.e., level of detail), this is due to these narrower cones resulting in greater concentration of light into informative pixels and less light contributing to reduced contrast via light scatter.



    There’s another good reason to
stop down a lens to a smaller aperture.  It turns out that most lenses are not their sharpest when shooting wide open (i.e., at maximum aperture), and this isn’t simply an artifact of the depth-of-field effect discussed above.  In the figure below are three images of a Brown-headed Cowbird (Molothrus ater), the two leftmost images taken at f/5.6 (the maximum aperture of the lens that was used), and the rightmost image taken at f/8.  Since the focus point was directly on the bird’s cheek or neck area in all three photos, we can largely ignore issues of DOF.  As you can see, fine feather details appear more distinct in the cheek/neck area of the image taken at f/8.  For this particular lens (the Sigma 800mm f/5.6 prime lens), I’ve found that maximum sharpness occurs at about f/11, which is two full stops down from maximum aperture (f/5.6). 




Fig. 3.7.5: Many lenses get sharper when you stop them down.
The left and middle images were shot wide open (f/5.6), while the
rightmost image was shot one stop down (f/8).  In addition to the
greater DOF, the rightmost image appears sharper, even at the
focal point, than the other two images.

    The conventional wisdom is that while most lenses need to be stopped down a bit to achieve maximal sharpness, not all lenses need to be stopped down by the same amount.  Thus, a higher-quality lens might require less stopping-down than a lower-quality lens, to achieve maximal sharpness, and that would allow you to get the sharpest images without sacrificing as much light by stopping down.  Keep this in mind as you read lens reviews.  Good reviewers will typically test a lens at various apertures.  As a general rule, cheaper lenses are cheaper for a reason, and a lack of sharpness—particularly sharpness at maximum aperture—is often one of those reasons.  Whereas my Sigma 800mm f/5.6 lens typically needs to be stopped down to f/11 (2 full stops) for maximal sharpness, my Canon 600 + 1.4× TC (effective: 840mm, f/5.6) combination is generally at its best at f/7.1—only 2/3 of a stop down from wide open.  In this case I’d chalk that up to the R&D advantages of the larger company (Canon).


Bokeh

A much more subtle differentiator between birding lenses is what is known as the bokeh.  Bokeh is a Japanese term used to describe the background of an image.  Today, many people prefer images in which the main subject stands out well from the background.  Photographers thus try to get photos in which the background is smooth and indistinct, with few details to distract the eye from the main subject.  An example of poor bokeh is given by the Belted Kingfisher (Megaceryle alcyon) photo shown below.




Fig. 3.7.6:  Good Bird, Bad Bokeh.
The distracting background patterns in this image were
caused by a large DOF and by the fact that the lens was a
catadioptric mirror lens with a
black hole in the center
(causing the doughnut shapes).  1600mm, f/12.


The background in the image above contains a lot of distracting elements.  In this case, the bird was perched in front of a bush with many bare twigs showing.  Because the lens was a fixed-aperture lens, I had no choice but to shoot at f/12, which resulted in a large depth-of-field that wasn’t shallow enough to render the bush totally out-of-focus.  Another problem evident in this photo is the lens’ tendency to make doughnut-shaped patterns in the background, which some people find distracting.  In this case, both problems (the large depth-of-field and the doughnut patterns) were caused by my use of a mirror lens—actually, a Maksutov-style astronomical telescope.  Mirror lenses are often criticized for the poor bokeh they impart to their images.
    Even expensive lenses can produce distracting background artifacts, however.  In the Louisiana Waterthrush (Seiurus motacilla) image below, you can see that the background also contains many round or polygonal shapes which can be somewhat distracting.  This image was chosen for inclusion in a nature calendar, so apparently the bokeh was good enough for the calendar’s editor.  But the fact that the lens, an $8000 (US) prime lens made by Canon, produced background artifacts shows that no lens is perfect in this regard.




Fig. 3.7.7: Even expensive lenses can cause background artifacts.
Background circles / octagons are caused by the lens’ iris, which
controls the aperture.  This lens (Canon 600mm f/4L IS) just
happens to have an 8-bladed iris; hence the octagons.


In this case, you can see that many of the background shapes are roughly octagonal.  This is a result of the shape of the iris (diaphragm) used to adjust the aperture in this lens.  Since this lens has an 8-bladed iris, any background shapes that appear in an image are likely to have an octagonal aspect.  Lenses with more blades in their iris will tend to produce rounder background shapes.  Also, some lenses feature an iris with rounded blades, which can result in a smoother bokeh.
    In practice, poor bokeh can almost always be fixed in post-process (i.e., in Photoshop), if you’re willing to spend the time doing so.  In the figure below, you can see that I’ve replaced the background with a green gradient.  Personally, I think I prefer the unmodified image (on the left), but this at least demonstrates that image backgrounds can be replaced wholesale, if you don’t like what’s there.




Fig. 3.7.8: Black-and-white Warbler (Mniotilta varia).
Left: the original image.  Right: after replacing the background.

Another example is given below.  In this next figure, I’ve replaced the background behind this Palm Warbler (Dendroica palmarum) with an irregular green pattern, rather than a simple gradient or solid color.  In this case, the background was made by taking an out-of-focus photo of a sunlit forest edge, and then merging it with the original bird image (a painstaking process, unfortunately).  Although the resulting image looks a bit unreal, it’s certainly more striking than the original (not shown).




Fig. 3.7.9: Palm warbler (Dendroica palmarum).
Composite image: an out-of-focus background was
merged into the original bird photo, which had a
relatively poor bokeh.


In the next image (below), I’ve again replaced the background with an out-of-focus image of a forest/sky scene.  In this case, though the background was artificially added to the image, it at least illustrates something important about bokehs in general—namely, that a good bokeh isn’t necessarily entirely devoid of detail.  In my opinion, a good bokeh has significantly less detail than the foreground, so that the viewer’s eye is drawn first to the bird, but still has some interesting patterns for the eye to contemplate after it has finished contemplating the foreground.




Fig. 3.7.10: Prothonotary warbler (Protonotaria citrea).
Another composite image.

Unlike the two previous images, the next image contains the natural background captured in the original photo.  Again, you can see that the background has much less detail than the foreground, but does have some crude, indistinct patterning to give the eye something to contemplate after it’s done absorbing all of the detail in the bird.




Fig. 3.7.11: Another Prothonotary Warbler.
This image retains its original background.

Finally, the image below again shows a natural background (ie., not Photoshopped).  The light areas of the background are simple and avoid distracting the eye from the subject (the bird).  The branches and pine needles aren’t rendered totally out of focus, as the traditional recipe for good bokeh would call for, but in this case I think they add just the right amount of detail to show some of the character of the bird’s natural habitat.



Fig. 3.7.12: Gray Catbird (Dumetella carolinensis).
This image retains its original background.


    In summary, while sharpness and bokeh are certainly important aspects of lens quality, if your lens isn’t the sharpest or the smoothest-bokeh-producing lens available, it’s not the end of the world.  In terms of sharpness, you may just need to stop down your lens a bit more than you would with a more expensive lens, and compensate for the lost light with a slower shutter speed or higher ISO setting.  If, during post-processing, you find that sharpening the image still doesn’t produce an acceptably sharp image of the bird, you may just have to settle for a smaller crop ratio, in which the bird appears a bit smaller in the final image.  In terms of bokeh, if your lens doesn’t unfailingly deliver that
smooth, buttery background that lens connoiseurs talk affectionately about, then you might just need to work harder to find birds in less-cluttered environs (so the backgrounds will tend to be smoother even through your particular lens) or to get more creative in post-processing.  Smoothing and the replacing of backgrounds in Photoshop is discussed in great detail in Part III of this book.