Shutter speed -vs- effective image stabilisation

Question

I have a Canon 5d full frame with a mixture of Canon, Sigma and Tamron lenses. Is there a stage when the image stabilisation system becomes ineffective because the shutter speed used is faster than the response time of the IS system. I have noticed that when shooting birds in flight, I get better results shooting at 1/2000 sec with the IS system turned off that shooting at 1/1000 sec with the IS system in use (IS used in either of the 2 settings).

Answer 1

Image Stabilizations only compensates for camera movements.

IS has no effect on subject movements. None. Zero. Nada.

If your subject is moving, only a shorter exposure time ("faster shutter speed") will reduce the amount of blur caused by the subject's movement.

Some related questions here at Photography SE for further reading:
What is more important, f-stop or IS
What reduces blur from camera movement more: large aperture or image stabilisation?
Is Canon 24-70 f/2.8L II that much better than Canon 24-70 f/4L IS?

Is there a stage at which there is a lag with the image stabilisation system that is overcome by shutter speed? There must be a time element within any IS system that impacts the image detrimentally as in rapid high speed shooting (7 or 10 fps) where the IS cannot keep up with the repeated shutter speed of 1/1000 sec?

IS in lenses such as the EF 600mm f/4L IS II keeps up just fine at 14 fps when used with the EOS 1D X Mark II. It's just that as the exposure time is reduced, the amount of blur caused by camera movement at the same rate of change is reduced by the shorter exposure time.

Keep in mind that regardless of the exposure time, as long as it is shorter than the time it takes each shutter curtain to transit from one side of the sensor to the other, the amount of time needed to expose one side of the sensor to the other does not change. What changes is how long each particular spot on the sensor is exposed as the second shutter curtain chases the first curtain across the sensor.

Notice that the shutter curtains cross the sensor at the same rate for the four exposure times shown: 1/1000, 1/2000, 1/4000, and 1/8000. The only difference between 1/1000 and 1/2000 in your 5D is that the very narrow slit between the first and second curtains is half as wide at 1/2000 as it is at 1/1000 even though the time it takes each curtain to cross the sensor is about 1/400 second in both cases.

When the EOS 1D X Mark II is shooting at 14 fps the interval between each frame is approximately 71 milliseconds. At sync speed or faster, each frame is exposed in about 2.5-3 milliseconds while any specific spot on the sensor is exposed for 1 millisecond at 1/1000 and for 0.5 milliseconds at 1/2000. That leaves about 68 milliseconds between each exposure when the mirror is moving down, the AF system and metering system are doing their thing, the mirror is moving up, the camera is waiting for confirmation that the mirror is all the way up, and the shutter is told to fire.

To understand just how little of the total time per frame is spent with the shutter actually open (partially at any given point with Tv shorter than flash sync speed), check out this video from which the above GIF is extracted. It's cued to a point right before the shutter button is fully pressed. Waiting for the mirror up confirmation seems to take forever. It's much longer than the actual exposure!

I have noticed that when shooting birds in flight, I get better results shooting at 1/2000 sec with the IS system turned off that shooting at 1/1000 sec with the IS system in use (IS used in either of the 2 settings).

The difference you are seeing between using a Tv of 1/1000 and using a Tv of 1/2000 is that the birds are only moving half as much in the 1/2000 second their image is projected onto a particular spot on the sensor as they are in 1/1000 second. It has nothing to do with IS.

In terms of lens alignment, IS always has the potential to be detrimental - even at slower shutter times. With lens based IS, the movements of the IS element/group introduce mild misalignment of the lens. This is viewed as acceptable if the blur introduced by the less than 'perfect' optical alignment of the lens¹ induced by an IS movement is less than the blur that would otherwise be introduced by the motion of the lens/camera. Eventually, as exposure time is shortened, one does reach a point where the penalty of slight lens misalignment outweighs the benefit of compensating for blur caused by camera motion. That does not, however, explain your observations with regard to birds in flight.

As with almost everything photography related, there isn't any "free lunch." For how both IBIS and lens based IS can affect image quality, please see this answer to Does IBIS reduce image resolution? How does it compare to lens based IS?

_{¹ There's no such thing as a 'perfectly' aligned compound lens, even among non-IS prime lenses. There are always manufacturing tolerances to be considered.}

Answer 2

John, I think you are missing the point of using IS. IS essentially works to eliminate the camera shake while taking the photo. However, in your case, you need a higher shutter speed to freeze the moment. So, the higher the shutter speed, the crisper and clearer would the image be of a flying bird.

Answer 3

I've wondered the same thing and from my research came to the conclusion that, "Yes there is a point at which Image stabilization is detrimental"

The Problems With Image Stabilization

Image stabilization can also have some weird effects if you use it in the wrong situations. Once your shutter speed is above around 1/500th of a second, image stabilization won’t really improve your images. Your muscles don’t twitch 500 times a second! Instead, it can actually have a detrimental effect on the sharpness of the image because of the moving elements in the lens. Although it’s mainly anecdotal, most professional photographers turn image stabilization off unless they absolutely need it for this reason.

I had read somewhere, but can't find the reference now, that image stabilization samples and adjusts the optics at around a thousand times per second. Most shutter curtains transit the sensor at around 1/250 of a second (even if you're shooting faster).

Based upon that, shooting below 1/250 would integrate stabilization shifts over the entire sensor image. Shooting faster than the curtain speed results in a moving slit that could have IS/VR shifts on only parts of the image, perhaps multiple times.

The net result is I don't know for sure and it's still a subject of discussion.

-------- Edit -------

I just found this:

Image Stabilization Testing at The Imaging Resource

Their testing is primarily aimed at blur improvement over slower shutter speeds, but note the lower left where the shutter speed exceeds twice the focal length. Non-stabilized becomes better.

The difficulty in testing is that it's highly subject to the person holding the camera. Tripods are not an option.