Canon IS effectiveness: amplitude and frequency

[Frank Kolwicz]

I've read some about Image Stabilizers and almost all reviewers ever talk about is extra stops of shutter speed, but, it seems like there must also be some range of effectiveness for the amplitude and frequency of camera movements for which it works best, well, fairly or poorly.

I'm specifically interested in the EF600/4IS (not IS II).

Any info?

[Bob Atkins]

There's no published data from any of the major lens manufacturers that I'm aware of. Canon certainly don't publish data on stabilization other than their "up to 2/3/4 stops" statements.

I presume they are tuned for a bandwidth in the cycle/sec to the 10s of cycles/sec range. I think the IS system on the FTM lenses might be tuned for some lower frequencies because they are said to help stabilize video when walking with the camera (presumably the source of low frequency movement)

They don't need to be tuned to high frequencies because people's hands don't shake at high frequencies! The plot here (http://www.bobatkins.com/photography/tutorials/mlu.html) shows camera mirror induced vibrations, which do have a higher frequency than what you'd expect from hand held movement. The major vibrations while the shutter is open seem to be at around 30Hz. However IS systems aren't really designed to minimize mirror induced vibrations (though they may work to do that in some cases).

IS systems certainly don't work at very low frequencies because the system drifts. At zero frequency (i.e. with the lens on a tripod), IS with exposures of maybe longer than 0.5s (that's just a guess) causes blur though drift of the IS system. At don't know at what point the drift becomes larger than the stabilization. Very low frequency stabilization could also be a problem when panning the lens, for lenses that don't have the option to switch off stabilization along the panning axis.

My guess would be that the bandwidth is maybe 1Hz to 50Hz at the outside. Probably more like 0.5Hz to 25Hz. I've read that hand tremor peaks in the 8-12 Hz range so I'd assume that's where IS systems are tuned for best performance.

I've seen some data on Canon's stabilization system for professional video shooting. There the stabilization seems to be tuned for maximum effect in the 2Hz to 10Hz range. Those systems use VAP-IS (variable-angle prism image-stabilization) which is a little different then the method used in EOS lenses, but the principles are the same, though they may tune the IS to include motion due to walking with the camera as well as hand/shoulder holding it.

If you want some technical reading on the subject, try this:

http://www.image-engineering.de/library/conference_papers/image_stabilizing.pdf

I suppose you could measure the characteristics of the lens you are interested in, but I don't know what the point would be. There's nothing you could do about it and the way to minimize image blur is to put the lens on the biggest heaviest tripod you can find and use a remote release with mirror lockup. It's not like you can control the frequency at which your hands shake when handholding the lens, or easily tune a tripod for resonance in a particular frequency range. Even if you could, I doubt it would be of any practical help and tuning for resonance would only increase the vibrational amplitude, something you certainly don't want.

[Frank Kolwicz]

Thanks, Bob,

I'm not sure yet if that's helpful. I'll have to digest it a bit, especially since the paper cited is about hand-holding cameras with relatively short focal lengths and that is almost certain to have different characteristics compared to my tripod/window mounted 600 which I expect would minimize the lower frequency movements due to whole-body movements when hand-held and, perhaps, shift other movements to higher frequencies.

I differ about not be able to do anything about it: I can add weight to the camera/lens and shift it's location to dampen some movement and I can turn the IS off, if it isn't really helpful under some conditions (as long as I know what those conditions are). Also, I can shift the balance point of the camera/lens on the support to affect movement in some way or use a different head.

Frank

[KeithB]

I think this might be a good chance for a roll your own experiment. Put a light source (laser pointer?) a few hundred feet away and shoot twenty to thirty frames in continuous mode. (Or even video!)

Combine all the images and draw a circle around the cluster of points. The radius would represent the effectiveness of the IS.

Then start experimenting. This way, you can tune things for *your* vibration profile, not Canon's vibration table. 8^)

[Bob Atkins]

I'm surprised you're seeing much in the way of image blurring due to camera movement with a window mount and an IS lens. They're normally pretty stable (assuming a good design) unless you are using really low shutter speeds.

I presume you are shooting with the car engine OFF?

[KeithB]

I did *not* get good results at an air show with my Sigma 100-500 hand held with IS. I was panning and forgot to put it in mode two, so that *might* have been the problem. The shutter speeds were in the thousandths, so I *could * have tried without IS, but I was afraid of missing any shots to experiment too much.

[Frank Kolwicz]

I'm surprised you're seeing much in the way of image blurring due to camera movement with a window mount and an IS lens. They're normally pretty stable (assuming a good design) unless you are using really low shutter speeds.

I presume you are shooting with the car engine OFF?

Bob,

Yes, engine is off and, yes, cloudy, rainy Oregon skies require that I do often use long shutter speeds, like 1/200 sec and slower at times (effective focal length is some 1300mm, so that's pretty slow and well below what the IS is supposed to effect), but I also wonder about possible ill effects of IS at high shutter speeds. Some times both high and low speed images are fine, sometimes not. EXIF data doesn't record IS setting, so it's impossible to tell which are which when I review them, if I turn it on and off in a series of shots and it's difficult to set up test conditions to match the field work. Maybe I shouldn't use IS at all!

Also, I think I'm pretty good at handling long lenses at slow-ish shutter speeds, I've been doing it for 25 years and, with reliable equipment (ahem), get high counts of critically sharp images either with my home-made car window mount or on a tripod. My EF600/4L IS, when it is working right, with the 1.4x extender on the 7d has given me remarkably sharp images, but at times fails utterly, which is why I'm questioning the useability of IS under some conditions, but would like to have it's benefits when it will work properly.

Note: the (ahem) above refers to the fact that I've had to send the lens back to Canon because it no longer focuses properly after the most recent repair. That failure caused me the loss of a series of images of two adult Virginia Rails chivvying a mob of 10 chicks (probably two broods) across a road with one dangling a worm to attract them while the other brought up the rear. That's the loss of useable images from a once in a life-time sighting.

[Frank Kolwicz]

I think this might be a good chance for a roll your own experiment. Put a light source (laser pointer?) a few hundred feet away and shoot twenty to thirty frames in continuous mode. (Or even video!)

Combine all the images and draw a circle around the cluster of points. The radius would represent the effectiveness of the IS.

Then start experimenting. This way, you can tune things for *your* vibration profile, not Canon's vibration table. 8^)

Keith,

I'm trying to think of how this could practically be done. It seems like it would take days of effort just to gather the data for the range of parameters, never mind the analysis.

Frank