Image stabilisation and tripods

[TonyW]

Being somewhat of a contrarian nature, I tend not to believe things that don't make sense to me. One of them was the statement that image stabilisation should be turned off when the camera is on a tripod. However, I was caught out the other day when we were out photographing the stars. Many, but not all, of my pictures showed what looked like camera movement. I hadn't turned off the IS and i am prepared to accept now that this the cause of the problem.

I still don't understand the mechanics, but one thing I read said that there is some sort of feedback loop which operates to upset the lens. If this is so, why does it not occur when the camera is hand-held?

I should explain that the lens was the Canon 24-105mm, not the latest version. I was using 20 second exposures at 24mm with the lens at maximum aperture (f/4). Other people have said that they always turn off the IS when using a tripod, especially with the larger lenses.

I also had a problem with focus. At night with dim scenes like these, auto-focus does not work. Similarly, I find it difficult to focus with live view, especially in the dark. Some people seem to have success using the infinity mark on the lens but earlier in the day I did some experiments with the auto focus on distant objects, namely clouds. It seemed to me that, at 24 mm especially, the markings did not give an accurate indication of where the focus at infinity should be.

[Manfred M]

Tony - IS is designed to counter small movements of the camera, so the underlying assumption is the camera is moving and the floating element in the lens moves in the opposite direction to counter camera shake. When one is hand-holding there will always be a tiny bit of movement so this works.

Put the camera on a rigid tripod that does not move and leave IS turned on, the camera assumes movement and there will be a tiny bit of "jitter" as the actuators try to counteract that non-existent movement. Turn IS off and this doesn't happen and the image should be sharp. Some more modern lenses and camera bodies can sense that the camera is on a tripod and shuts down the IS, so it is not a problem. The older lenses need you to physically shut down is.

I don't shoot the sky, so I'm not sure if my advice here will work for you, but for night shots I do use LiveView to manually focus, but I do so buy pushing up the screen magnification as high as needed so that I can see the subject clearly. That technique works well for the type of shooting I do.

The infinity mark on modern auto-focus lenses is not particularly accurate, but it is repeatable. If you focus (use single point focus) at an object that is effectively at "infinity" for that focal length. Remember the actual position of where the mark lines up on your lens and put it to that position. Your focus should be bang on. Unfortunately. modern lenses are not tuned on an optical bench with a hard stop any more, so we have to use these workarounds.

I hope this all makes sense.

[xpatUSA]

Put the camera on a rigid tripod that does not move and leave IS turned on, the camera assumes movement and there will be a tiny bit of "jitter" as the actuators try to counteract that non-existent movement.

Sorry, Manfred, that sentence makes no engineering sense to me.

It's akin to saying that, if there is no movement (tilting) whatsoever of a gyroscope axis, there will still be gyroscopic precession. Or, if movement is detected by accelerometers, they will output a signal in the absence of any movement (acceleration).

With your engineering background, I doubt that you meant to imply either of the above, so would you like to clarify the sentence?

[JBW]

Tony, focusing for wide field astro photography is as much an art as it is a science. here's what works for me.

Wide field, manual focus, center on the brightest star, focus magnification to highest, focus, test shot, focus again, test shot, and one or two more times.

Also remember that for wide field dslr astro-photography clear dark nights are crucial.
Brian

[royent]

Why would anybody want to use image stabilisation with a tripod ?
Roy

[george013]

Why would anybody want to use image stabilisation with a tripod ?
Roy

That's why you must not forget to switch it off.

George

[DanK]

You can find a brief explanation of the issue in this post by Canon. The system doesn't anticipate motion. It detects and responds to slight movements of it's own components, setting up a feedback loop that Canon refers to as "shake return." It's only the older lenses that have this problem, but I don't know when the design change occurred. I have often forgotten to turn IS off when using the tripod and haven't had a problem, so I assume that my lenses are new enough that they don't have much of this problem.

[Manfred M]

Why would anybody want to use image stabilisation with a tripod ?
Roy

Shooting on a surface that is moving. For instance a bridge with heavy traffic on it will have low frequency vibrations due to traffic driving over it. Shoot with the IS turned off and that movement can be transmitted through the tripod to the camera inducing movement / camera shake.

[xpatUSA]

You can find a brief explanation of the issue in this post by Canon. The system doesn't anticipate motion. It detects and responds to slight movements of it's own components, setting up a feedback loop that Canon refers to as "shake return." It's only the older lenses that have this problem, but I don't know when the design change occurred.

Now it is clear, thanks for the link:

these might be caused by the tripod being knocked, or the photographer adjusting a camera setting. The IS system then swings into action to correct that movement. The movement of the IS lens group causes its own minute vibration, which is in turn detected by the movement sensor, which triggers another correction. This ‘feedback loop’ can continue endlessly

Once again in the world of photography, a word gets missed out - "positive". A control loop's feedback is normally negative at relatively low frequencies. Most items in a loop have a lagging response with increasing frequency. If the loop gain is 1 or more when the total phase lag becomes 180, then sustained oscillation will occur.

Our old friend Nyquist (he of MTF plots) has a stability plot named after him, see figure 4.8 here:

http://www.ece.rutgers.edu/~gajic/psfiles/nyquist.pdf

All fascinating stuff back when I was a Controls Design Engineer for a well-known industrial gas turbine company in Lincoln, England.

[Manfred M]

Ted / Dan - both my explanation and Canon's explanation in the article you posted are simplifications of what is really happening. Both are correct to a certain extent, but real life is a lot more complex and as I do not have the engineering analysis of what happens when a camera is mounted on a tripod, I can only speculate as to what is happening. The reason I say that Canon's explanation is not 100% accurate is that it does not match my "real world" experience or Tony's, which is why he posed the question.

Put the camera on a tripod on a solid surface, use a remote control (with mirror lockup mode) so as not to introduce any camera shake due to the photographer walking around and take a long exposure. Do this with image stabilization turned off and turned on. If the images are identical, then the Canon explanation makes sense, if not, then it does not. My experience is that leaving the stabilization turned on results in a softer image than with it disabled, which seems to be Tony's experience too.

Ted - your response is similar to what I got from the controls systems engineers when we tried to integrate the control system in a real world operating equipment and the devices did not operate as required and sometimes not at all. Sometimes it was a mechanical design issue (my problem), sometimes it as a control system issue (the control engineer's problem) and sometimes it was a component issue (operating in spec, but at the limits of the spec; that tended to be an issue both of us were involved in resolving, but it generally was easier and less expensive to change the control system than the mechanical element).

Integrating a control system with components that have mass (actuators and the floating elements), impact of previous state (hysteresis), errors in the feedback signal (reported action does not match what happened in the "real world", under damping in the system, etc. are all possible reasons for "jitter" in the stabilization system. I'm sure that the camera and lens manufacturers know, but they are not sharing with us.

[rpcrowe]

Older Canon lens IS systems were "dumb" and could not distinguish if a lens was solidly mounted on a tripod or some other stand and the "SHAKE RETURN" phenomenon might occur. Newer Canon lenses are "smart" in that they can tell when the camera/lens is solidly mounted and the IS will turn itself off preventing the "SHAKE RETURN" from occurring. I don't know if this is true for lenses other than Canon...

IS versions (at least since the 17-55mm f/2.8 IS and 70-200mm f/4L IS lenses) are of the smart variety. Canon's more recent recommendations are somewhat ambiguous recommending that you turn off the IS when tripod mounted to "conserve battery power".

There are times when I turn off the IS, especially when the exposure is rather lengthy, But, for shorter exposures, I tend to leave my IS turned on. When I use my lightweight hiking and travel tripod, the combination of "smart IS", compensating for minor shaking, along with an Arca Compatible Q.R. clamp and L bracket has allowed me to get very sharp shots at slower (but not seconds or minutes long) shutter speeds in very windy conditions on the canyon rims in the state of Utah.



The ARCA clamp and L bracket keep the center of gravity directly over the tripod head at the apex of the tripod. Not needing to hang the camera cantilevered over on its side for vertical compositions also allows the use of a lighter tripod and head.

I will however shut down the IS when I am making time exposures. When I am shooting these long exposures, I will lock up the mirror and either use a remote shutter release or fire the camera using the self timer.

BTW: A corollary question to "Should I leave Image Stabilization on when tripod mounted?" is often "Should I leave the IS on when using a monopod?" In that case, the answer is an emphatic YES!

Some recent Canon lenses are so "smart" that they detect when they are being panned and will shut off the IS for the panning axis. Photo equipment is getting smarter and smarter so that dumber and dumber photographers have better chances of bringing home technically decent images. Like the Canon speedlight/camera interface automatically shutting off HSS (if selected) when shutter speed is reduced to below the maximum sync speed and then turning on HSS when the speed exceeds the maximum sync speed. Why still have an HSS selection button?

[xpatUSA]

Ted / Dan - both my explanation and Canon's explanation in the article you posted are simplifications of what is really happening. Both are correct to a certain extent, but real life is a lot more complex ...

Ted - your response is similar to what I got from the controls systems engineers when we tried to integrate the control system in a real world operating equipment and the devices did not operate as required and sometimes not at all. Sometimes it was a mechanical design issue (my problem), sometimes it as a control system issue (the control engineer's problem) and sometimes it was a component issue (operating in spec, but at the limits of the spec; that tended to be an issue both of us were involved in resolving, but it generally was easier and less expensive to change the control system than the mechanical element).

Indeed so, Manfred. We had a case on a 4MW generator set where the whole generator rotor was oscillating (phase angle-wise) even though it's speed was a constant 50Hz. A mechanical engineer figured out the cause and I figured out the fix, being to increase the phase lead term in my speed governor. Changed a 1uF capacitor to 2.2uF, IIRC.

[TonyW]

Thanks for all the replies. Sorry for not responding earlier but I was busy yesterday.

Dan's reference to Canon's article was interesting but to me still does not explain the engineering aspects of the IS system. I suppose it is quite complicated and without a full study of the system it is not possible to understand. One interesting comment that the article makes is that the IS system itself generates some motion. Also, i would guess that any vibration on a tripod is likely to be at a much higher frequency than that caused by hand motion (quoted by Canon as being 1 to 5 cycles per second) and might be higher than the IS can cope with.

For my practical purposes, the solution is simply to turn off the IS when using a tripod. My answer to Roy is that the reason I have had IS turned on was simply laziness and neglect. If I was more serious about astro-photography, I would buy a more modern and more appropriate lens.

[loosecanon]

By coincidence I was looking last night at some images taken a couple of weeks ago with a Canon 70-300L lens which were softer than expected when I used a tripod, but crisp when I had hand held the lens. Thinking why, I put down the softness to failure to turn off the IS.

[xpatUSA]

Thanks for all the replies. Sorry for not responding earlier but I was busy yesterday.

Dan's reference to Canon's article was interesting but to me still does not explain the engineering aspects of the IS system. I suppose it is quite complicated and without a full study of the system it is not possible to understand. One interesting comment that the article makes is that the IS system itself generates some motion.

Only when you turn the camera on and the IS system "erects", Tony, irrespective of what the article says.

Also, i would guess that any vibration on a tripod is likely to be at a much higher frequency than that caused by hand motion (quoted by Canon as being 1 to 5 cycles per second) and might be higher than the IS can cope with.

Could be on the right track there. Just as we can assess a camera sensor by viewing it's spatial response to a sharp edge, we can also assess a position control system by viewing it's time response to a pulse input (a knock) or to a step input (sudden change of position). These events being measured in milliseconds rather that seconds. With such a event, the IS system will certainly do "something" but it's anybody's guess what that something is, depending on whose stuff we are talking about.

Generally, the preferred response to a step input is a quick rise with very little overshoot, say 2-5%. But some systems might go more than that and might even oscillate back and forth for a short while (known as "ringing").

Tried to keep this simple but not too dumbed-down. FYI, I usually turn Sigma's OS off when on a tripod.

[William W]

. . . I was caught out the other day when we were out photographing the stars. Many, but not all, of my pictures showed what looked like camera movement. I hadn't turned off the IS and i am prepared to accept now that this the cause of the problem . . . the Canon 24-105mm, not the latest version. I was using 20 second exposures . . .

Noted the Exposure Time: It is likely (actually quite probable) that the “Feedback Loop”, for want of a better phrase, was set up by the Mirror Slap, even if you did use ‘Mirror Up Technique’. Alternatively your hand movement when releasing the Shutter.

I am pretty sure that, in at least some of the newer versions of IS, the mirror slap can set up enough "feedback loop" movement for Camera Shake Blur to be noticeable because it continues during long exposures – at least I have experienced exactly this phenomenon with IS lenses that have been released within the last four years - even when using a remote shutter release.


BTW the same consequence of the Mirror Slap can apply with NON IS Lenses, especially telephoto lenses, for shutter speeds from about 1/15th second to 2 seconds, the mirror slap can be enough to vibrate the lens and for those range of exposures times, the vibration can be caught as camera shake blur.

So irrespective of FOCAL LENGTH and SHUTTER SPEED, provided the Tripod is stable and on solid ground, it is best the TRIPOD technique to:

1. > turn OFF IS
2. > use Mirror Up Technique
3. > use a remote release

***

Additional to Dan's reference to the CPN document, there's a reasonable description of Image Stabilization in "EF Lens Work III - The Eye of EOS". I'm renovating my Study and my Library is in storage at the moment, so can't give you page references, but I have seen some extracts of this book in pdf form, on the www.

WW

[William W]

. . . If I was more serious about astro-photography, I would buy a more modern and more appropriate lens. . . .

In the case that you get more serious, (or someone else reading is serious) you don’t need “more modern lens” only “more appropriate lens”.

Assuming you have a 135 Format (aka “full frame”) Canon DSLR, then the (original) EF 24 F/1.4L is a cracker lens for Astrophotography.

WW

[apersson850]

I've seen behaviour in the IS systems that indicate that they tend to drift, when there's no "normal" vibration input (hand shake) to them. Thus they tend to slide around the frame, when not getting any shake they can recognize to combat.