On Ways To Check Lenses
DAVID RUETHER
[The following can also be used for checking lenses on
digital cameras...]SOME BACKGROUND ON LENS CHECKING--
For the numerically inclined, the obvious way to
establish the optical performance level of a lens may
be to photograph lens test charts and then to read the
finest resolved lines at different apertures off the film,
converting the results to a solid-sounding set of numbers.
For the casual lens user, shooting a roll of film and
looking at the 4x6 prints may be sufficient to establish
that a lens is good enough. For those not convinced that
either method is very meaningful for showing the detailed
information about how lenses perform for taking pictures,
for separating the barely adequate optics from the truly
excellent, and for quickly and easily spotting poor lens
designs and poor samples of good lens designs, I offer
some methods of lens checking that have served me well
for well over a quarter century (Yikes! ;-).
Before getting into my methods, let me give some of the
reasons for my misgivings about using some other methods
(while noting that ANY method that satisfies a user is
fine for that user - there are few absolutes here).
Using lens test charts has many pitfalls for establishing
lens imaging performance: the commonly used distance of
25X the focal length may place many otherwise fine lenses
at a distance that is unlike what would be used in normal
photography, and at which the lens performance is not
representative (and may not be very good); it is difficult
to test lenses using test charts at all the relevant
distances (especially near infinity) that would show how
a lens performs (lens performance varies at least some
with distance, and may vary a large amount, depending on
the type and design of the lens); it is VERY difficult to
focus with sufficient accuracy on a flat target to make
the test meaningful without using sequence focusing
techniques at each aperture (and a lot of film and
eyestrain); it is difficult to align the film and target
planes sufficiently well to make the off center target
information reliable without special alignment tools or
great care - and an otherwise sharp lens may have slight
field curvature which could make the edge test results
look poorer than the lens actually is in normal use; resolution tests may not tell much about contrast, which
in some images may be more important than resolution (both contribute to the look of sharpness in a lens) - also,
in lens design, it is hard to improve both contrast and
resolution together, and much easier to trade one for the
other (it is quite possible for a lens to be very high in
resolving ability, but look poor in picture taking - and
the other way around); it is difficult to duplicate
resolution test results, even when using the same setup,
equipment, and materials - which leads to questions about
the reliability of solid-seeming test result numbers; a
meaningfully thorough chart testing of even one fixed focal
length lens (let alone a zoom, or two lenses tested together
for comparison) is an exhausting ordeal (not a suitable
process for anyone but the most dedicated lens tester) and
the resulting sea of numbers may be overwhelming, and may
be less than easy to translate into a good understanding
of how one lens performs compared with another.
On the other side, taking a few photographs to judge the
optical quality of a lens also has pitfalls: what looks
sharp enough on a print (even an 8x10) may not look sharp
on a negative examined with a good magnifier (or on a larger
print or projected slide); cross-lighting on the subject
photographed can mask faults that would be evident in a
larger collection of photographs taken under a variety
of lighting conditions; optical faults can easily escape
detection (until after the return period [or warranty
period] has expired - when they have a habit of suddenly
making themselves very obvious); many people seem completely
unaware of image quality anywhere but in the center of the
image, but a photograph is EVERYTHING inside the frame edge,
and when the photo viewing (and taking) sophistication level
improves, the "sharp" lens (with the soft edges and corners)
may not look so sharp anymore; one person's "looks sharp to
me" comment may not serve another's desire to know how good
a lens is, since there is no reference standard of quality.
REASONABLE ASSUMPTIONS ABOUT LENS CHECKING--
Of the two methods of lens checking in common use (one
attempting to quantify performance using test charts and
numbers, the other qualifying performance using normal
subjects and descriptions), I prefer a variation on the
latter, but with the addition of some references to
better quantify the quality descriptions (using subjective
lens evaluation numbers). Here are some reasons for this
choice (and some assumptions that I think are reasonably
made): familiar normal subjects at various subject distances
can be used repeatedly for different lens checks; most
normal subjects are at least somewhat three dimensional,
which probably insures that some parts of the subject will
be in focus even with a slight focus error (which makes
focus errors more detectable than they would be with flat
subjects, and helps reduce the focusing accuracy standards
to practical levels); a subject near infinity presents no
subject-film plane alignment difficulties (unless the lens
shows field curvature - and, even then, opposite sides of
the frame should match in sharpness [and the four corners
should also match] if the lens is well-aligned); distant
subjects may be moved around to various parts of the film
frame without introducing focus errors (which near subjects
would do if they were moved in the frame); and the same
subject moved around in the frame presents a familiar (and
comparable) rendition on the film.
It is reasonable to assume that the center of the lens image
is no worse than any other part of the image, so comparisons
can be made between the center and any other part; it is
reasonable to assume that the center performance of virtually
all good lenses is excellent by f5.6, so that performance at
other apertures, and in other parts of the frame may be
compared with the reference center performance at f5.6 (when
this is not correct, frames exposed at smaller apertures or
experience gained checking other lenses should indicate
that); it is reasonable to assume that a well-aligned lens
will have equal performance at opposite frame edges, and
also at all four corners (assuming that experience checking
a few lenses [or using the method for body checking given
below] has established that the camera body used for lens
checking is well-aligned [the lens mount is parallel with
the film plane and the viewed image, and the film and
viewing screen have the same focus]).
STANDARDS FOR LENS IMAGE QUALITY--
If the requirement for a lens to be considered sharp were
for the lens to show excellent sharpness at all points in
the frame at all apertures and focusing distances, there
would be VERY few lenses that would be acceptable. It is
necessary to determine your own standards for the lenses
you use. For me, if a lens is not reasonably sharp in the
corners and sharp at the edges at a particular aperture and
in the focus range appropriate for that aperture (relating
lens performance to how the lens will generally be used),
it is not sharp at that aperture, no matter how astoundingly high the resolution is in the center. Whether a lens can
resolve 56 or 100 lines in the center at f5.6 at 10' is
irrelevant information for me - focus is rarely accurate
enough to make the difference noticeable (unless the
subject has enough depth to have SOMETHING correctly
focused, a large enough photograph is made [and examined
closely enough] to see the difference, and the camera is
held steady enough [and the subject does not move] to
realize the higher resolution on film). What does matter
to me is good wide-aperture performance (which can aid
focusing, and allow lower-light photography with good quality), and good off-axis performance (which is so poor in so many lenses that it is a very relevant issue concerning
lens image quality, and it is of primary interest to me).
My standards follow (good resolution and contrast center-to-corner, uniform performance at opposite sides
and all four corners of the frame [good optical alignment],
even illumination, and reasonable freedom from linear
distortion and from flare and ghosting are characteristics
understood to be present for good lens performance):
- Super-wides (6-15mm) - must perform well at all distances
by f11.
- Very-wides (18mm-24mm) - must perform well at all distances
by f5.6.
- Wide-angles (28-45mm) - must perform well at all distances
by f5.6, and reasonably well at most distances wide-open.
- Normals (50-60mm) - must perform well at all distances
by f2.8, and reasonably well at most distances wide-open.
- Short-teles (85-135mm) - must perform well wide-open,
though not necessarily near the closest focus distance.
- Medium-to-long-teles (180mm, and up) - must perform well
at all distances wide-open.
- Zooms - must perform nearly as well at particular
focal-lengths (especially tele-only zooms) as similar
focal-length non-zoom lenses at the same apertures (the
far corner sharpness and distortion characteristics may
be slightly worse and still be acceptable).
MATERIALS NEEDED FOR LENS CHECKING--
Almost any film is suitable for lens checking. Resolution
is a function of both film and lens resolution - increasing
or decreasing either lens resolution or film resolution does
not result in an equal change in on-film resolution (which
is why using extremely slow, very sharp film does not result
in the often expected vast increase in image sharpness).
Useful information in lens checking can be had by using
a film like Tri-X or one of the better 200-speed newer
color negative films, which can allow simplification of the
checking process by eliminating the need for a tripod if the
shutter speeds are high enough. (What, no tripod in lens checking?!?! ;-) A tripod does aid in accurate focusing of
longer lenses, though, and I do recommend using one for
lenses over about 200mm. Shooting the tests in bright
sunlight also helps. Using very slow films will slightly
increase the differences in the performance characteristics
that you are looking for, but the problems introduced
(having to use a really reliable tripod [rare], with
mirror pre-release, etc.) discourages me from this choice,
especially when an easier one is available. Note that a
few lenses (mostly very-wides and super-wides) do show
differences in edge and corner performance on color and
B & W film (most look better on color film when there is
a difference).
EQUIPMENT NEEDED FOR LENS CHECKING--
A good 10X magnifier is sufficient (less magnification makes
everything look good; more magnification makes film reading
too difficult, can result in "empty magnification", and can
make it more difficult to distinguish between image detail
and the film grain or sensor noise; 10X is a good choice for
easily seeing the differences in sharpness that you are
looking for) - we are not looking for 100 lines-per-mm, just
equality and sufficiency of detail in the parts of the frame
of interest in the lens check. Cheap is O.K. - I like a
folding field biological magnifier (less than about $6). If
you use a magnifier with a skirt, the magnifier often works
better turned around backward (the focus distance in the
normal position is rarely correct, unless it can be focused). You will also need a broad, even light source,
such as a window on an overcast day, a frosted globe lamp,
or a slide-sorter. A sharp-tipped china marker is useful for
writing on the processed film what each frame represents.
A *well-damped* tripod may be useful (not necessarily a
rigid or sturdy one, but one that quickly stops vibrating
after being shocked). A well-aligned camera body is
essential for lens checking (see below for more on this),
although problems with this are rare.
CHECKING THE CAMERA BODY FOR GOOD ALIGNMENT--
Before attempting to check lenses, it is necessary to check the camera body for good alignment. There is more on this
here. Here are two ways to check camera focus alignment. The
first will tell you about parallel alignment of the viewing
screen and mirror with the film and bayonet (if there is a
problem, it is usually a mirror alignment problem, rather
than a bayonet mount-film plane alignment problem, though
a slightly tipped viewfinder screen is not unknown). The
second will tell you more about focus accuracy. A camera
that fails the first test will probably fail the second,
but a camera that passes the first may fail the second.
-1) Put on the body a fixed focal length lens (preferably
a good short-tele, which is the easiest lens to use for
checking viewfinder alignment if the viewfinder is not very
good (and a short-tele is less likely to be misaligned than
zooms, wide angles, or even normal lenses) and aim it at
distant detail (like a tree-edged horizon, or distant
buildings), focus the same distant subject carefully at the
four edges of the frame several times (observe the focus marks on the barrel each time you focus). If the focus is
almost always the same in the center of all four edges of
the frame, the lens is probably forming an image of the
subject that is parallel to the viewing screen. It is likely
that the lens mounting bayonet is parallel to the film plane
(it is unusual for it not to be), but a film check is a good
idea to confirm it. Shoot the same distant subject (at about
f2.8) in the center of each of the four frame edges without
changing focus, then check the four frame edges on the film
(using a good 10X magnifier) to see if they are equally
sharp. If there is a difference in focus around the edges
of the frame, try another lens. Lenses can form images
that are not parallel to the film plane but be otherwise
sharp - but that is not acceptable to me.
-2) Put a good 50mm (f1.4-2) lens on the camera and shoot
a newspaper at about 45 degrees at about 2' at f2. Focus carefully on one letter in a paragraph that you can identify
easily on the processed film. Shoot the target maybe 6 or 8
times, refocusing each time. Look at the film with a good
10X magnifier. If you hit the focus right-on more than 1/2
the time, and the other focus points are randomly in front
of and behind the correct point, the viewing screen focus
almost certainly agrees with the film focus (especially if
an infinity-focus target also shows correct marked focus
on the lens barrel scale and is sharp on film).
If you want to know more about your camera viewfinder, you
can check framing accuracy (film area shown, centering, and
rotation) and linear distortion by photographing a distant
lake shore, road center line, window edge, etc. at the very
top of the frame, turning the camera 90 degrees with each
photo until all four sides of the frame have been used.
(If this is done hand-held, I suggest repeating each photo
two or three times to average your framing errors.) On the
film (preferably unmounted, if slide film) you will see how
far the photographed lines are from the edge of the film
frame, whether the lines are equally spaced from the edges
of the film frame, and whether there is a consistent
parallelism between the lines and the frame edges (if
there is a consistent angle instead, the viewfinder mask
is rotated slightly - a not uncommon fault in viewfinders,
which makes leveling the camera difficult). When
photographing the straight lines, you may notice that
they bend inward in the viewfinder - this is more likely
due to pincushion distortion in the viewfinder (very
common), than to linear distortion in the lens (unless
it is a zoom, in which case the two distortions will add
in the VF), but you can check the straightness of the lines
on the film to see which it is.
CHECKING FOR VISIBLE LENS FLAWS--
Check the lens to see if the lens is sufficiently free of
mechanical focusing wobble (or unsmoothness) for you; is
free of oil on the diaphragm leaves (operate the aperture
with the lens off the camera, looking in both the front and
rear at the blades to see if they are dry, and open and
close properly); focuses to infinity on the viewfinder
screen (beyond is OK, especially if the lens is AF, a zoom,
a mirror lens, or has Ed glass); and is free of obvious
flaws in the glass (by placing a bare light bulb behind and
in front of the lens with the aperture open, and looking
into the lens from the side opposite the bulb, but at an
angle so that you can see the illuminated glass but not the
blast of light (this will show some dust inside, and maybe
a barely visible hairline scratch or two, but should not
show major amounts of haze, fungus tendrils, etched
fingerprints, obvious scratches, element delaminations,
or other horrors - unless the lens was very cheap). Minor
faults like dust, a bubble or two, even a light layer of
haze, a small amount of fungus, mars in the coating, or a
few hairline scratches in the glass have so little practical
effect on the lens image (especially if they are near the
edge of the glass) that they can be generally forgotten
(unless you are like me, and are a "mint-nut"). The reason
this is true is easy to figure out: the defect occupies
such a small percentage of the lens area that it cannot
affect more than a tiny percentage of the light passing
through the lens. Even if the defect were great enough to
cause light scatter at a level of, say, 5 stops below the
average exposure, it would not register on most films. A
deep and extended scratch, a fingerprint, considerable haze
or fungus, or a lot of hairline scratches all over a lens
surface are defects that would affect image quality under
some circumstances, but not necessarily all. Also, visible
lens defects mostly affect the contrast and brilliance of
a lens, not its resolution.PROCEDURES FOR LENS CHECKING--
Well, enough of all this blather - and on to some ways you
can quickly and easily determine if a lens is good, and
well-aligned. If you are lucky enough to have a camera body
with an excellent viewfinder (100% coverage, no linear
distortion, with high contrast and excellent sharpness
everywhere in the frame), you can make good preliminary
checks for lens alignment (lack of image "tilt" relative
to the film) using the viewfinder. A viewfinder this good
appears only on the Nikon F, F2, F3, F4, F5, and F6, alas,
and all others that I have seen have shortcomings that make
visual checks of lens alignment with the VF useless - but
if you happen to have one of these cameras, you can
carefully use it to focus on a very distant, contrasty
subject, placing the subject at different locations on the
screen (the four edges, the four corners, and the center)
and repeatedly manually focusing the subject and checking
the focus points on the lens focus scale to see if you get
the same results for all of the points. You can also check
for lens linear distortion on these bodies.It is practical to check out lens alignment and sharpness
on film using a minimum of film. I check it by first locking
all camera controls and features in manual mode. I take a
vertical photo (with the shutter release end of the camera
up) of a detailed, distant scene (cityscapes and distant
lines of bare trees work well, and a mix of these is
better yet), with the detail at the top of the frame. Then,
carefully not touching the focus ring, I turn the camera
over, and repeat the photo (both should be shot wide open).
If you feel unsure of your focusing accuracy, refocus and repeat what you did a few times. Then shoot the same target
the same way at f5.6 and maybe also at f11. Then shoot
diagonals with the detailed subject material running from
one corner to the opposite corner, and then reversed in
camera tilt, at the same stops as before (compensating with
appropriate changes to the shutter speed, of course!)Take notes as you shoot! You will not remember what you
shot afterwards!
Develop the film (leave it uncut and/or mounted), and before
checking it, write on the film frames in unneeded areas how
you shot them (FL, aperture, etc.) so that you can use the
film as a reference later. Examine the film with a good 10X
magnifier. In the pairs you first shot, you have the same
subject in two frames, but just across the frame lines,
where it is easy to compare (in digital, you can flip one
of the frames, making it even easier to compare the image
detail). Check first the centers of the edges of the frame
pairs (where exactly the same subject appears). It should
be equally sharp and detailed in both photos. Next, check
the opposite corners along the adjacent frame lines (which
also have the same subject material), which should also be
equal in sharpness). In the diagonal frames, you can see if
there are any anomalies, and how the image center sharpness
compares with the edges and corners. If the lens passes this
check, it is probably well-aligned and not affected by any major optical manufacturing flaws. Use the f5.6 and f11
frames as references to see how well the lens can perform,
and how well it performs wide-open compared with optimum
apertures (this is not for checking if the lens is up to
par for its type, model, and best samples - but the wide
open results may tell you if you are satisfied with a
particular lens used wide-open). Few lenses approach perfection (darn!), but I do expect good performance for
the purposes I want to use them for! For MUCH more on this,
see my "SUBJECTIVE Lens Evaluations (Mostly Nikkors)".And, don't get too caught up in lens checking - it is a
DISEASE, I tell you, a DISEASE!!!!! Once acquired, it is
VERY hard to get rid of it!