This non-technical article is intended to give you a brief introduction to the often-misunderstood issue of camera shake, and to image stabilization, offered in many cameras to combat this problem.

Camera shake and motion blur

Even mounted on a tripod, your camera is never perfectly still. Especially changes in its orientation (attitude, as opposed to the position) while the frame is being exposed bring a degree of motion blur to pictures. It is the movements of the direction in which the lens axis points which are much more harmful than any others.

If the amount of that blur is small enough, less than any image unsharpness due to finite lens/imager resolution and other factors, we will not see it. Otherwise we talk about visible effects of camera shake, something usually we want to avoid.

For the same variations in the lens axis direction, the amount of motion blur seen in pictures depends on (is directly proportional to) three factors:

• The lens focal length;
• Image magnification (from the sensor to the viewed print or screen size);
• Exposure time (within reasonable limits; we are talking fractions of a second).

If the focal length is expressed not in absolute terms but as 35 mm "full frame" equivalent (EFL), then the first two factors on the list merge into one: two lenses of the same EFL magnify camera shake to the same extent, regardless of the frame size.

Avoiding camera shake

To avoid the motion blur resulting from camera shake, shorter exposure times (faster shutter speeds) can be used. The generally accepted rule of thumb is that the slowest safely handholdable shutter speed is the reciprocal of the EFL. For example, with a 200 mm lens on a 35-mm camera, 1/200 s is generally considered safe.

For a Canon APS-C the corresponding actual focal length will be 135 mm, because the EFL multiplier is 1.6&rimes;. For Four Thirds, the multiplier is 2× hence a 100 mm lens will be equivalent to a 200 mm one on a full-frame camera.

Therefore we could rephrase the 1/EFL rule in terms of the actual focal length, F (in millimeters). For Canon APS-C cameras the nominal handheld exposure is 1/1.6F, for Four Thirds (Olympus and Panasonic) SLRs — 1/2F. Example: 1/100 s for a 50 mm lens on a Four Thirds camera.

This, obviously, depends on the photographer's picture-taking technique: how the camera is being held and shutter released. More experienced (or just more thoughtful) photographers can often get away with shutter speeds four times (2 EV) longer than the rule suggests (here: 1/50 s), while some people will get blurred pictures even at shutter speeds faster than our rule indicates.

Following a few simple points may provide you with one or two EV advantage when photographing from hand.

• If your camera has an optical finder, use it rather than the LCD monitor. This alone may allow you to use shutter speeds 2 to 4 times longer (1-2 EV) compared to holding the camera in your outstretched hands. (Actually, the 1/EFL rule assumes you are using the camera this way.)
• Avoid holding the camera in one hand only; use your left hand to cradle the body, palm up, with fingers supporting the lens barrel.
• Use only your fingertip to release the shutter; do not move the whole hand, and don't start the move to put your camera away even before the picture has been made.
• When standing, brace yourself with your feet wide, or lean against some support (tree, door frame, etc.). Keep your elbows against your body.
• Especially with longer lenses, use breath control, like in rifle shooting. Take a breath in, partially let it out, avoid breathing immediately before and during the shutter release. Don't hold your breath for too long.
• Before taking the picture, check the shutter speed the camera is going to use. Even the simplest compacts display this information. If the speed is slower than our rule above says, try to support the camera against something

All this may often bring more improvement than buying a camera with image stabilization.

Image stabilization

Higher shutter speeds are not always feasible. At low or moderate light levels they will require a higher ISO settings in the camera, but this is possible only up to a certain limit, and at the expense of degrading the image quality. This is why many current digital cameras come with the image stabilization feature.

Solutions used here can be divided into two groups. Both use a motion sensor to detect in real time changes in camera/lens attitude, but then the information is used differently.

• Lens-based IS: the signal drives a micromotor which then moves rapidly a dedicated group of lens elements; this group changes the direction of the light leaving the lens towards the sensor, counteracting the camera axis movements; the image created on the sensor is more stable.
• Body-based systems: instead of "moving" the light before it reaches the sensor, the signal drives a micromotor moving the sensor itself; as if "chasing" the constantly moving image. The result is similar to that above.

In lens-based system the sensor itself is also built into the lens, so you pay for the IS every time you buy a new lens. The range of stabilized lenses may be quite limiter, too.

Makers of such systems (Canon, Nikon) claim that it can work better, because each implementation can be tweaked best to the lens in which it is contained. Those who offer the body-based approach (Pentax, Sony, Olympus) say that their way works just fine, thank you.

I suspect that the true reason behind Nikon's and Canon's approach was the existing investment in lens-based systems from the film era. In a few years we may see them adapt the body-IS technology.

There is no doubt, that being able to use image stabilization with any lens is a very attractive option for many camera users. Canon and Nikon offer IS only in some lenses, mostly with longer focal lengths (where this is most useful, anyway), and mostly at a premium. In other systems, every lens gets the benefit of image stabilization, regardless of when and by whom it was made.

Some camera makers use the term "image stabilization" to describe program modes in which the sensor sensitivity (ISO rating) is set higher, to allow for faster shutters. This is a misleading abuse of the term.

There are also solutions, mostly seen on low-end models, based on doing a quick sequence of shorter exposures (someties just motion picture frames) and merging them together, shifted to compensate for the differences. This "digital IS" remains out of scope of this discussion.

How much does IS help?

The manufacturers usually come up with some numbers describing the benefits of their IS systems. This is usually done in terms of how much longer shutter speeds become handholdable with IS than without, often expressed in terms of EV (exposure values). One EV means doubling (or halving) the shutter speed, so, for example, 3 EV means a factor of 8×.

Unfortunately, with no information whatsoever how these numbers were derived, they are worth no more than anecdotal evidence or ones generated off the ceiling in the marketing department.

The only reasonable approach to all such claims is to just ignore them. They are certainly not comparable between manufacturers.

We have to remember that camera shake is largely a stochastic process, one drawn by chance. You may get an unblurred frame at 1/15 s without IS, and the next one, shot under identical conditions at 1/30 s with IS may show motion blur; bad luck. In the transitional area of shutter speeds where IS really maters, it does not give you a guarantee of success; it only raises the probability.

This statistical improvement may depend on a number of factors. My experience strongly indicates that at higher shutter speeds IS is more effective than at lower ones. What this means is that shooting with unsteady hands and/or with longer lenses will benefit from IS more than shooting steady and/or with shorter ones.

A beginner using a moderately long lens may, for example, have a 50/50 chance of getting a no-blur picture at 1/125 s; with IS this may move down to 1/30 s, a gain of 2 EV. With the same lens, a more experienced user may get away with 1/30 s without, and with IS — 1/15 s, which is just 1 EV difference. Don't get it wrong: he still gets sharper pictures (or is capable of using slower speeds successfully), it is just that he benefits less from image stabilization.

A reasonable procedure to determine the advantage of image stabilization would involve taking a large number of pictures in two series: one with, and one without IS. For each focal length both series would have to cover a wide enough range of shutter speeds, to determine at which speed the success (no-blur picture) rate reaches some agreed upon value, for example, 50%. The difference between these speed values (expressed in EV) with and without IS would be a good measure of the advantage gained. (I am assuming, obviously, that both series are taken under otherwise identical conditions, or with identical statistical mix of such conditions.)

The disadvantage to such method is that it requires hundreds of frames to be shot, but this cannot be avoided. You cannot estimate a statistical property using just a few readings.

I have developed such a procedure and used it to test image stabilization on two Olympus SLRs: E-510 and E-3. The IS benefits ranged from about 1 EV at wide angle (EFL of 24 or 28 mm) to more than 2 EV (at EFL of 300 mm). Compare that to manufacturer's claims of "up to 5 EV" improvement.

This is not just Olympus, the practice is common in the camera industry.

Technically such claims may be true, because up to really means "we know of at least one such case". This is the same as promising "up to $30 million a year" earnings at Wal-Mart (sorry, the link is dead) \: it may happen, but don't expect it.

For a reasonable range of focal lengths (EFL 28 to 300 mm) and an average operator, I would expect the IS to bring in a difference ranging from 1 to 3 EV, regardless of the solution type, brand, or model, but depending on the factors discussed above.

How not to evaluate IS benefits

I have seen this error committed so many times, that it I feel it deserves a separate section in this article.

A reviewer says: "With IS we were getting sharp images at 100 mm EFL and 1/10 s; at this focal length the normal handholdable exposure is 1/100 s, therefore the IS brings an improvement of 10× or 3.3 EV".

Wrong. This means nothing. Do a small experiment: turn the IS off; you'll be probably getting good pictures at 1/40 s or so. Does that mean that a disabled IS gives an improvement of 2.5× (1.3 EV)? Just because your camera has an "IS" sticker on it?

The only right way to do it is to perform a with/without comparison, keeping all other things equal, and then you have to catch the region when the change occurs (i.e., the pictures start getting blurred); something similar, if not necessarily identical, to my procedure described elsewhere. Anything else is old wives' tales.