I’m sure everyone reading this had had the experience of shooting an image which was spoiled by blurring because they couldn’t hold the camera still enough. A basic rule of thumb is that in order to have a reasonable chance at a sharp image, you need to shoot with a shutter speed equal to 1/(focal length) seconds. So if you are using a 50mm lens you need to be shooting at 1/50 sec or faster, and if you are using a 300mm lens you need to shoot at a shutter speed of 1/300 sec or faster.
This generally works well for full-frame 35mm cameras, but for cameras with smaller sensors (most digicams and many digital SLRs), the equivalent full-frame 35mm focal length needs to be used. So if you are shooting with a Canon EOS 30D (which has a 1.6X “digital multiplier”), and you are shooting with a 300mm lens, you need to shoot at a shutter speed of 1/480 sec or faster (since 300mm x 1.6 = 480mm).
One way to get a fast enough shutter speed is to increase the ISO setting on the digital camera (or use faster film in a film camera), though that is only a partial solution, because as you increase ISO, you also increase noise/grain and overall image quality decreases. (Some compact cameras have automated this faulty feature; more about that later.)
There are several ways of shooting at slower shutter speeds and still getting sharp shots. One way is to use a tripod, but that’s not always convenient. A second way is to attach a large and heavy mechanical gyroscopic stabilizer to the camera. A few years ago this was just about the only option for a handheld camera, but the gyro systems (such as Ken-Lab units) weighed several pounds, cost several thousand dollars, and needed large external battery packs to run them for a few hours. Effective, but not very convenient.
Enter Internal Image Stabilization
In September of 1995, things changed. Canon introduced the first commercially available SLR lens with internal image stabilization, the Canon EF75-300mm f/4-5.6 IS USM (current version is the 70-300mm f/4-5.6 IS USM model). This lens used two gyro sensors in conjunction with a movable set of internal lens elements to optically stabilize the image on film (1995–remember–we were still shooting film!). This allowed a 300mm lens to be handheld at shutter speeds down to about 1/90 sec rather than the 1/300 sec which would have been needed without stabilization. In contrast to the bulky, heavy, power hungry external bolt-on gyro stabilizers, the system was small (contained inside the lens), added little weight and consumed little power. Nikon followed in 2000 with a similar system which they call “Vibration Reduction” (VR), in 2004 Sigma released its first “Optical Stabilization” (OS) lens and in February 2007 Tamron introduced its first “Vibration Compensation” (VC) stabilized lens.
In the fall of 2004, Konica-Minolta (now Sony) introduced a DSLR (the Maxxum 7D) with a different stabilization system. Instead of including optical stabilization in the lens (which keeps the image on the sensor stable), Minolta took a different approach and moved to sensor around to follow the image from the unstabilized lens.
They had previously used a similar system the A1 EVF camera, which had been introduced in the previous year (2003). The camera detects motion via internal motion sensors and the digital imaging sensor is moved up and down, left and right, via an electromagnetic system to compensate. Currenly the Sony Alpha A100 uses in-body stabilization and Pentax also uses a similar system in their K10D and K100D DSLRs which they call “Shake Reduction” (SR).
Some camera makers have a third system which goes under a number of names but which Fujifilm calls “Picture Stabilization.” You might think this is some sort of mechanical stabilization system that operates on the optics or the sensor or maybe even both, but you’d be wrong! It’s actually just an in-camera program which automatically selects higher ISO settings when it thinks that you are using a shutter speed and focal length at which it could be difficult to hold the camera steady. This can certainly work (at the expense of higher noise), but it’s confusing, at best, to call it “stabilization”. It’s really more of a “smart program” mode.
There’s actually one more stabilization system, but it’s only used for digital video. It uses digital image processing to shift the image from frame to frame to compensate for camera motion. Individual frames aren’t stabilized, so it’s not usable on still cameras. However it’s pretty effective at stabilizing videos.
How well does stabilization work?
Below are three typical shots taken under three different conditions. The top image shows a 100% crop from an image of a street sign taken using a Canon EF 70-300mm f/4-5.6 IS USM lens at 300mm, using an EOS 20D body set to ISO 400 and a shutter speed of 1/100 sec. In this case the image stabilization system was turned off. As you can see, it’s pretty blurry. The shutter speed rule of thumb says that you’d need a shutter speed of at least 1/500 sec to have a good chance of a sharp image.
The middle image shows what happens if you increase the ISO to 3200. The shutter speed now increased to 1/800 sec and the image isn’t degraded by motion blur. However you can see that the result of increasing the ISO is more image noise and less sharpness than in the final (lower) image.
The bottom image shows the effect of going back to ISO 400, but this time turning on the IS system of the lens on. As you can see, image noise is low and sharpness is high, demonstrating the effectiveness of image stabilization.
How much stabilization do you get?
This is actually quite a difficult question to answer, since it depends to some extent on how shaky your hands are, as well as which system you are using and which lens it is applied to.
Manufacturers claim anywhere from 2 to 4 stops of stabilization and practical tests seem to bear this out. Typical tests seem to gain about 2-3 stops of additional stability (meaning you can shoot at shutter speeds 2-3 stops slower than without stabilization).
Which system is best?
A tripod is the best way to stabilize the image, but second best are the optical and mechanical systems. Stabilization in the lens has the advantage of actually showing you the stabilized image though the viewfinder. Not only is this reassuring but it can make aiming long telephoto lenses easier. In-camera stabilization via a moving sensor can’t do this.
However in-camera stabilization works with just about any lens you mount on the camera, so you only have to pay for it once and the cost isn’t very high (maybe $100-$200). In-lens stabilization adds at least $200 (sometimes more) to the cost of every lens that it’s used in–and not all lenses (particularly wide and normal prime lenses) are available in an IS version.
In practice, both systems seem to provide similar levels of stabilization for wide, normal and short telephoto lenses used on APS-C sensor DSLRs. If and when more cameras with 
full frame sensors become avialable, image stabilized lenses will still be effective on them (as they are on current 35mm film cameras). With the additional size and mass of a full frame sensor, it may be difficult to impement a full frame sensor camera with in-body stabilization.
In-lens stabilization may also be more effective than in-body stabilization for long telephoto lenses. This is because the image shift is proportional to focal length and can become quite high at long focal engths. For example for a 0.5° deflection of a 600mm lens, the image moves by about 5.5mm, and Canon IS telephoto lenses can shift the image by this amount. Moving a whole sensor +/- 5mm both horizontally and vertically to compensate for image movement is difficult, even for APS-C sized sensors, due to both space constraints and limitations on how fast the mass of the whole sensor assembly can be moved. Optical stabilization in the lens can be designed so that a small movement of the optical steering elements causes a large deflection of the image and so rapid and effective stabilization is possible even for large image shifts.
When is image stabilization not so effective?
It’s important to remember that image stabilization allows the use of slower shutter speeds, but it doesn’t help in freezing action. So, for example, when shooting football you may be able to handhold a 300mm f/5.6 stabilized lens at 1/100 sec, but if 1/100 sec isn’t fast enough to freeze the player’s action. For that you need either a faster lens (such as a 300mm f/2.8) which would allow a shutter speed of 1/400 sec, and/or you have to resort to increasing the ISO setting if that is possible. For action work there’s no substitute for a fast lens (though a fast lens with IS is even better!).
© 2007 Adorama Camera
