What Is Image Stabilisation?

Sony a9 with FE 100-400mm f/4.5-5.6 GM OSS, at 400mm, 1/60-sec. handheld, with no IS

Untitled by David Stephens, on Flickr

Same rig, with IS on:

Untitled by David Stephens, on Flickr

I couldn't turn off IS at the lens and keep IBIS on. As someone wrote early, apparently they split duties.

Later in the week, I'll mount a Canon lens and try it with no IS, lens IS only, IBIS only and both lens and body IS engage.

Most fairly recent IS Canon lenses can sense if the lens is tripod mounted and switches off the IS, or at least Canon do not feel it necessary to say anything about it, but the earlier Canon IS systems were suggested to be turned off. It is advisable to read the instructions for each lens in this regard.

 

I am not a slave to "ultimate sharpness" which I generally think is a never ending and rather pointless pursuit, but in general I do like an suitably sharp image when required.

Actually, at least with my EF 500mm f/4L IS II USM, the IS stays on when on the tripod. With a big lens like that, even on a heavy tripod, with heavy ballhead and Sidekick, you can see the image settle down when the IS spools up. There's no need to turn it off.

I've had people look at some of my 1,000mm handheld moon shots and claim that they see a half-pixel smear. For enjoyment of the shot that doesn't matter. OTOH, I have a 50" print of the Grand Canyon in my office. Most people stand six-feet away and love it, but photographers put their good-eye six-inches away and then they approve, the sharp results from low ISO, tripod mount and remote release. Still, most of my landscape shots are handheld, but with ISO engages.

My single most expensive lens lacks in-lens stabilization, but I also don't consider it a useful feature for that particular lens(14-24mm 2.8).

BTW, here's one person's take on the value of VR. Thom Hogan is pretty well respect in the Nikon community, and his basic advice is to only use VR when you need it

All About Nikon VR | DSLRBodies | Thom Hogan

I think the Canon system is similar enough that we could likely extrapolate much of what he says to IS.

BTW, VR isn't infallible, but it can help. As an example, at ~50mm(full frame) I can often get about every other shot or every third shop acceptably sharp down to about 1/15 or maybe a bit slower. I could never do that without VR, but it also means firing a couple of shots to be safe. Also, if I'm doing something like chasing kids around inside, those kind of shutter speeds mean that I'm probably going to have sharp furniture and blurry kids :)

BTW, here's one person's take on the value of VR. Thom Hogan is pretty well respect in the Nikon community, and his basic advice is to only use VR when you need it

What is the downside, other than use in conjunction with a tripod?

You can keep the subject sharp while blurring the background by panning in IS mode 2. It's pretty obvious that a subject in motion is likely to be blurred at a slow shutter speed, VR or not. That merits repeated shots in order to capture a pause in the motion. VR is not infallible, even with stationary subjects, if you don't use a modicum of technique. If you are accustomed to whipping your camera to and from your eye, you might want change that behavior..

... As an example, at ~50mm(full frame) I can often get about every other shot or every third shop acceptably sharp down to about 1/15 or maybe a bit slower. I could never do that without VR, but it also means firing a couple of shots to be safe. Also, if I'm doing something like chasing kids around inside, those kind of shutter speeds mean that I'm probably going to have sharp furniture and blurry kids :)

I didn't want to confuse things, so I didn't show it in my examples above, when one of my 400mm, 1/60-sec., IS-off shots turned up sharp. This is not unusual. When shooting the moon, handheld, with IS on, I'll take several shots, in rabid succession and, almost invariably, there'll be one sharper than the others.

in rabid succession

There are "Other Shots" for that problem! :D

There's no need to turn it off.

Yes, the 500mm is one of the more recent IS systems, and I suspect the instructions don't suggest turning it off. I seem to remember the 70-200 f4 IS did suggest turning it off when on a tripod, and I noticed a small improvement with that lens, but all of this is dependent on whether you think the improvement you get from reducing tripod vibrations, wind etc is more beneficial than removing any IS-induced vibration. I am sure with a 500mm the former would virtually always win out anyway.

I spoke too bradly about the effect of using IS on a tripod. I have not noticed any artifacts with my Sony and Sony/Zeiss lenses. However when I used my original Nikon 70-200 AF-S IS on a Nikon D3 (or earlier), the image would drift slowly and sometimes jerk at the moment of exposure. That lens uses gyroscopic acceleration sensors, which apparently have a tendency to precess. Linear inertial accelerometers do not exhibit precession.

If you slowly move the lens in one direction, at some point IS will get to its limit, and there is a jerk as it recenters.

This also means turn IS off while panning for moving objects.

If you slowly move the lens in one direction, at some point IS will get to its limit, and there is a jerk as it recenters.

 

This also means turn IS off while panning for moving objects.

Not with Sony or Canon.

If you slowly move the lens in one direction, at some point IS will get to its limit, and there is a jerk as it recenters.

 

This also means turn IS off while panning for moving objects.

Random motions due to hand shake keep the IS mechanism active and centered. I have noticed instability as you describe, but only in systems at least 10 years old and only on a tripod. Many IS lenses, have a mode switch with two settings. One setting has milder action which results in less jerky action in the viewfinder and permits smooth panning. The other mode is more aggressive, and has equal stabilization in all directions. Sometimes seeing stabilization in the viewfinder is optional. I can sometimes see IS action in my Sony A9 as jerky motion when I pan quickly between objects which aren't moving with the camera. That's an extreme condition visibly only because od raw EVF no-blackout properties. I never see any artifacts in normal operation.

If you move constantly in one direction, the system can't keep up forever.

It works well for vibrations around the desired aim.

I suspect that a high-pass filter, allowing it to ignore slower motion, helps.

But if you quickly move only in one direction, and for long enough, something has to happen.

If the logic figures out that you are panning a moving object, it might just give up trying to correct for it.

But it is hard to tell, looking though the viewfinder, moving that fast.

Random motions due to hand shake keep the IS mechanism active and centered. I have noticed instability as you describe, but only in systems at least 10 years old and only on a tripod. Many IS lenses, have a mode switch with two settings. One setting has milder action which results in less jerky action in the viewfinder and permits smooth panning. The other mode is more aggressive, and has equal stabilization in all directions. Sometimes seeing stabilization in the viewfinder is optional. I can sometimes see IS action in my Sony A9 as jerky motion when I pan quickly between objects which aren't moving with the camera. That's an extreme condition visibly only because od raw EVF no-blackout properties. I never see any artifacts in normal operation.

I haven't noticed that with my a9. Is it happening to you with an EF lenses or FE?

When I'm shooting with my FE 100-400mm, with electronic shutter, tracking a fast bird, I'll notice a fluttering in the EVF. I've assumed that was the actual images going by at 20-fps, with the in-between view being the EVF between shots. I've never thought to track a bird without shooting. I'll do it when I'm out this evening, to see if I can notice anything.

If you move constantly in one direction, the system can't keep up forever.

 

It works well for vibrations around the desired aim.

 

I suspect that a high-pass filter, allowing it to ignore slower motion, helps.

 

But if you quickly move only in one direction, and for long enough, something has to happen.

 

If the logic figures out that you are panning a moving object, it might just give up trying to correct for it.

 

But it is hard to tell, looking though the viewfinder, moving that fast.

I quite often pan nearly 180-degrees. I've never noticed an issue. That's usually not level panning, but sometimes it is.

I quite often pan nearly 180-degrees. I've never noticed an issue. That's usually not level panning, but sometimes it is.

If IS works, the image should not move when you move the camera.

That is the whole idea.

If IS works, the image should not move when you move the camera.

 

That is the whole idea.

I don't think that's the idea.

I'm not sure what you mean when you say, "the image should not move when you move the camera." When you make a 180-degree pan, of course the image moves. When I shoot a handheld shot of the moon with IS engaged, of course, the moon moves within the VF, because I can't hold the camera perfectly still. IS slows down any camera movement, such that instead of shooting at 1/500-sec., you can shoot at 1/125-sec. Make a short, handheld video and the subject will still move in the frame, but the shake will be greatly reduced by the IS, such that handheld videos are now watchable, if not as good as when taken on a good video head and tripod.

When I took the following handheld shot of the move, I took several shots. In each frame, the moon was in a slightly different position in the frame. I take over 100,000 handheld shots per year, so I have a lot of practice at it. Anyway, I can tell you, that IS doesn't stop the subject, even a stationary subject, from moving in the frame:

Waxing Gibbous Moon (Explored) by David Stephens, on Flickr

IS is based on motion detection in the camera and/or lens, and not of the subject. Accelerometers measure motion in several directions, causing optical elements (or the sensor) to move in compensation.

It is possible that simple devices like smart phones and P&S cameras use "digital" image stabilization, based on subjects in the image. AFIK, this technology is not used in DSLRs nor MILC cameras. I suspect that digital stabilization does not have the flexibility, resolution nor responsiveness demanded by serious photographers. Among other things, it is based on position, rather than acceleration, which is much more sensitive to small changes and in effect, anticipates the effect of camera shake.

If sensitivity is limited to acceleration, then the motion detectors are not subject to drift when stationary. Mass accelerometers are pretty close to ideal in this respect. Early IS used gyroscopes or oscillators, which are subject to precession, hence drift in static situations.