Does Image Stabilisation adversely effect image sharpness?

[Mads S]

Hi All,

I was reading about the technology of IS/OS or whatever you choose to call it, and it appears a little gyro moves a lens around in opposite sync with the tremors the objective is exposed to? If so, you'd imagine that pixel peepers could find slightly better image sharpness with non-shaken, non-IS'ed images compared to images where the IS was active? I have read that if you're doing Micro Adjust then you want to switch off the IS; I interpret that in the same direction, i.e. the IS affects the sharpness somewhat.

So my question is: If I have good support and fast shutter speeds, should I switch the IS off for the best results?

Thanks for chiming in,

[herbert]

I have read on Thom Hogan's site that the stabilisation mechanism operates at a frequency of 500 to 1000 hertz. So when shooting around 1/500 or 1/1000 of a second then the stabilisation element could move and adversely affect the image. When shooting fast shutter speeds you could benefit from turning it off.

Older stabilisation systems would still shake when on a stable tripod. So you were recommended to turn them off. Newer systems by Canon and Nikon are tripod sensing and turn themselves off anyway. The lens specifications should be able to tell you of this is the case.

Alex

[Colin Southern]

I think it's more "in theory" than in practice.

Why not do some tests and let us know what you discover?

Keeping in mind too, that the one over focal length rule for hand holding is only a rule of thumb for MINIMISING camera shake, not eliminating it -- so IS/VR may well still contribute something useful well above those speeds.

[Hero]

I have encountered a few odds with VR, but I'm not sure if the blame lies with VR or with me. For instance, I do a bit of dog-sports photography (coursing for sighthounds and dogfrisbee) and sometimes when I shoot for a certain point because I know the dog will pass that point, often the dog's coat looks a bit blurry when using VR. Whether this is because VR reacts to the appearance of the moving dog in the view or that it tries to arrest my unwitting tracking movement I don't know. But suffice to say that in those situations I now refrain from using VR.

[Glenn NK]

1) Most photographer's seem to have assumed that IS provides better sharpness - even to pixel peepers. That's why it's becoming so widely used. Even Sigma has OS lenses (and they are being sued by Nikon for patent infringement).

2) If IS/VR/OS is a problem when shooting at 1/500 or 1/1000 second, then it begs the question "why have or use IS"? At shutter speeds like that, IS may be redundant. The major benefit of IS is when the lighting conditions, ISO, and required f/stop conspire to put the shutter speed too slow. I've shot with my 24-105 at 1/15 second and obtained very sharp images.

3) Can IS/VR/OS "see" the moving dog, and react? I would think that only the focusing mechanism can "see" a moving object, but perhaps I'm not up to speed on this (pun intended).

I think Colin mentioned something about practice and theories. Good post Colin.

Glenn

[Colin Southern]

3) Can IS/VR/OS "see" the moving dog, and react? I would think that only the focusing mechanism can "see" a moving object, but perhaps I'm not up to speed on this (pun intended).

I'm reminded of an IS "gotcha" ... if folks are panning with an IS lens then it needs to be switched to mode II (Canon), or it interprets the horizontal movement as undesireable, and tries (in vain) to compensate for it -- with the resultant usually being a double image.

[William W]

Does Image Stabilisation adversely effect image sharpness?

Yes (in theory) it does.
But in practice it is very difficult to separate the adverse affect of Image Stabilization when IS is turned ON, because of how much IS favourably affects Image Sharpness by arresting camera movement.
So doing tests to isolate how “bad” the IS affect is, will prove very difficult, IMO.

Apropos “fast shutter speeds” . . . I suggest do some tests hand holding a 200mm lens and using 1/1000s and look at 100% 150% and 200%. – 1/1000s is not that fast for hand holding a telephoto lens, IF the file / negative is to be under close scrutiny.

Apropos “good support”. . . if that means “a very stable tripod” then most modern IS lenses (Canon) do not require the IS to be turned off as the lens’s IS system recognizes the stability: it’s worthy to note the still available new, EF 100~400F/4.5~5.6L IS USM, has an older IS system which does require it being turning OFF, when the lens is mounted on a tripod.

However – even IF you are using a tripod when using slower shutter speeds (consider 1/4s ~ 1/125s) you should consider using “Mirror Up” the help alleviate any sympathetic vibration along the lens which might be caused by the mirror’s action.

Another way to interpret the question (although I am sure it was not meant) is to compare two lenses one with and one without IS: the EF70 to 200F/2.8L USM and the EF70 to 200F/2.8L IS USM – the former is sharper than the latter.

WW

[jcuknz]

I wonder if the whole question is academic after one has post processed and included some form of sharpening?
Last time I used a tripod for a two second exposure the subject matter was moving* but I was suprised to get a sharp result in PP. *so slowly I didn't realise it was happening until after several takes, except it kept on moving out of frame and spoiling the composition Most annoying

[Mads S]

Yes (in theory) it does.
But in practice it is very difficult to separate the adverse affect of Image Stabilization when IS is turned ON, because of how much IS favourably affects Image Sharpness by arresting camera movement.
So doing tests to isolate how “bad” the IS affect is, will prove very difficult, IMO.

Apropos “fast shutter speeds” . . . I suggest do some tests hand holding a 200mm lens and using 1/1000s and look at 100% 150% and 200%. – 1/1000s is not that fast for hand holding a telephoto lens, IF the file / negative is to be under close scrutiny.

Apropos “good support”. . . if that means “a very stable tripod” then most modern IS lenses (Canon) do not require the IS to be turned off as the lens’s IS system recognizes the stability: it’s worthy to note the still available new, EF 100~400F/4.5~5.6L IS USM, has an older IS system which does require it being turning OFF, when the lens is mounted on a tripod.

However – even IF you are using a tripod when using slower shutter speeds (consider 1/4s ~ 1/125s) you should consider using “Mirror Up” the help alleviate any sympathetic vibration along the lens which might be caused by the mirror’s action.

Another way to interpret the question (although I am sure it was not meant) is to compare two lenses one with and one without IS: the EF70 to 200F/2.8L USM and the EF70 to 200F/2.8L IS USM – the former is sharper than the latter.

WW

I think this is turning into a useful discussion. Basically my question could be re-phrased as; might I get (even) tack-sharper results by consciously switchin the OS on my Sigma 150-500mm OFF when I'm on a tripod and have decent light - and I think the answer to that one by now is "clear", YES i might. And I shall.
I have had surprisingly good results with hand-held slow shutter speeds with the same lens back when I was using it on a 5D - but the 'upgrade' to 1.6 crop factor on the 7D has made the magnification so high (800mm equivalent) that even the rather impressive OS on the Sigma doesn't really manage. So I think I'll just stop pretending that it is hand-holdable, and switch off the OS.
When the bank allows I'll be getting a Canon prime, but until then I want to get the most out of the Sigma, and this has probably helped along :-)
Anyone curious could check out http://birds.syndergaard.dk for examples of what the Sigma can do - WARNING not National Geographic standard but I'm having a great time making them :-)

Thanks for participating,

[Dave Humphries]

Last time this topic came up, complete with the link to Thom Hogan's article suggesting problems between 1/500 and 1/1000s, I did try some tests myself.

I was using my Nikon D5000 with the Nikon 70-300mm VR lens. They weren't 'blind' tests, I knew when it was on (and when it was off), so I suppose you could say I might have (even subconsciously) influenced the results.

What did I find? - Very close scrutiny of about 30 normal, hand held shots (of a perched bird), with shutter speeds in the range of 1/500 to 1/1500s, I found the sharpness slightly better, on balance of number shots, in favour of VR on. There wasn't much difference though, but when you add to this the risk of forgetting to switch it back on if the light drops and your exposure ventures below 1/500s and the decision was made! I shoot with it on at all times - handheld.

If I use a tripod (for me, very rare);
indoors (i.e. no wind); I probably would turn it off - if nothing else it saves battery power (although that's never an issue on the D5000) - if I remember!
outdoors; if risk of wind movement, I might shoot both with and without VR if time and conditions permit (and if I remember!)

I also agree with the sentiment that unless you are cropping excessively, the downsize and sharpen process for web use will mask any of these small differences.


On the dog's coat issue mentioned above, the lens's VR/IS/VC isn't looking at picture content, it works by detecting movement of the lens itself, literally the "camera shake" and corrects on that alone - if there is an apparent link in the way the coat displays it must be 'coincidental' dependent upon the precise image content (fine detail) and the amount of movement.

[William W]

I think this is turning into a useful discussion. Basically my question could be re-phrased as; might I get (even) tack-sharper results by consciously switchin the OS on my Sigma 150-500mm OFF when I'm on a tripod and have decent light - and I think the answer to that one by now is "clear", YES i might. And I shall.

I have not used many Sigma Lenses.
However I do speak with a lot of Photographers who do use Sigma Lenses and it occurs to me that the Sigma Corporation is not overly forthcoming with detailed information regarding the uses and functionality of their OS System.

I concur with your assessment and the idea of switching off the OS when the lens is secure & stable on a tripod.

HOWEVER if I were doing the testing I would still use ‘Mirror Up’ and I would also be conservative about measuring the stability of the tripod – what ‘looks stable’ is sometimes not so stable: a centre hung counter weight and sand bags on the feet are very useful.

BUT – it is very unlikely that you’ll notice any difference in a 10 x 8 print anyway . . .

ALL that stated – I am all for getting the best product possible out of the Camera and applaud your efforts for pursuing same.

Good luck with it – and I would like to read your report.

WW

PS –Unsolicited comment:
Your ‘early morning shots’ utilize the light most effectively; compared with many other shots in your portfolio.

The Bullfinch is a particularly nice set of four images.

Thanks for sharing– I don’t shoot Birds – I have no patience for the task: but I enjoyed your work.

[rpcrowe]

I recently shot (and posted) a series of images of a polo match. I shot these at high shutter speeds (because I wanted to shoot at, or close to, maximum aperture to allow selective focus and minimize the background clutter) and I shot in IS Mode-II on a Canon 300mm f/4L IS lens. The shots were dead-on sharp and the IS seemed not to have any adverse impact on the imagery.

I suspect that in real-life shooting, a lens with a Canon IS system of the 300mm f/4L or later variety will not cause image problems. This is especially true since the 300mm f/4L IS and later image stabilization systems incorporate the Mode-II which allows panning. BTW: the 300mm f/4L IS lens incorporates an earlier version (I think it is version two) of IS. The IS systems in my 70-200mm f/4L IS and 17-55mm f/2.8 IS lenses are even more efficient than the IS in my 300mm f/4L IS lens!

[Mads S]

PS –Unsolicited comment:
Your ‘early morning shots’ utilize the light most effectively; compared with many other shots in your portfolio.

The Bullfinch is a particularly nice set of four images.

Thanks for sharing– I don’t shoot Birds – I have no patience for the task: but I enjoyed your work.

Thanks for the comments William - I'm funny in that I can't really be bothered to photograph anything BUT birds, another example of how different we all are :-)

As a birder, a good bird shot to me is a nice, dynamic image of something that you don't see every day - so although the Bullfinch is a pretty bird, it isn't my most valued set - I tend to like the waders and the owls from Brazil a lot :-D

Thanks again, and I'll post back when I have tried shooting without OS for a bit.

Mads S

[blakemcguire]

I have also heard an interesting point from an interview with a sports photographer. He said he never uses VR, and literally has the VR switch glued/taped in the off position. Reason being, he is using long lenses to shoot tight, and the VR activating makes the image "jump," which can cause him to lose the split second he needs to make that cover shot. At the same time, he's shooting with one of the newest 1d models (I'm a nikon user and therefore don't remember which revision they're on) so he's always up in the 2000/s and up anyway for capturing action.

Not quite so relevant to sharpness, but I think it's a useful point.

[dan marchant]

I have read that if you're doing Micro Adjust then you want to switch off the IS; I interpret that in the same direction, i.e. the IS affects the sharpness somewhat.

So my question is: If I have good support and fast shutter speeds, should I switch the IS off for the best results?

Micro Adjust should be done with the camera on a tripod. IS/VR should be switched off when using a tripod because it is specifically designed to counteract camera shake when the camera is hand held (unless you have a bad tripod and it is very windy). So yes, IS can have a negative impact when you use it while the camera is on a tripod. But when hand held it will almost always improve things (except in certain extreme conditions such as the sports photographer mentioned above).

I have encountered a few odds with VR, but I'm not sure if the blame lies with VR or with me. For instance, I do a bit of dog-sports photography (coursing for sighthounds and dogfrisbee) and sometimes when I shoot for a certain point because I know the dog will pass that point, often the dog's coat looks a bit blurry when using VR. Whether this is because VR reacts to the appearance of the moving dog in the view or that it tries to arrest my unwitting tracking movement I don't know. But suffice to say that in those situations I now refrain from using VR.

VR/IS is camera stabilisation, not subject stabilisation. It counteracts the physical movement of the camera due to shaking hands. It has no way of knowing if part of the image is moving - that isn't what it is for.

If your moving subject isn't sharp it is either because the shutter speed isn't fast enough to freeze the action or possibly because the AF is struggling to lock onto the target and hasn't actually achieved a lock. Alternatively it could be that it is locked but is fractionally front/back focusing.

[Colin Southern]

IS/VR should be switched off when using a tripod because it is specifically designed to counteract camera shake when the camera is hand held (unless you have a bad tripod and it is very windy). So yes, IS can have a negative impact when you use it while the camera is on a tripod.

In Canon land, the IS units from generation II onwards have tripod detection, which put the unit into a standby mode when on a tripod - so no "need" to turn them off on these models.

[AB26]

I cannot imagine that lens manufacturers will add stabilisation to a lens if it is going to adversely effect image quality. Have not read anything in any review yet, that says anything to this affect. This is quite interesting and worth the effort to investigate.
I do however believe if the lens is used according to the manufacturers instructions such problems would be eliminated.

Nikon calls it Vibration Reduction. Vibration suggests small movements to be countered by the lens. Panning is no small movement and when panning VR is probably eliminated. If the VR system can sence stability on a tripod I would think it would also sence when the movement is too much to handle and it would then disengage.
VR does however give you an additional 3-4 stops.

Thanks people you have given me another subject to investigate, so let me go and try and find out form the pro's.

Regards,
Andre

[AB26]

Hi again,

Go to http://www.bythom.com/nikon-vr.htm for an explination on the subject.

Cheers,
Andre

[Mads S]

Hi again,

Go to http://www.bythom.com/nikon-vr.htm for an explination on the subject.

Cheers,
Andre

Awesome stuff Andre, thanks! I'm pleased to have arrived at the same conclusion "intuitively", by simply thinking of the mechanics of the system and wondering.

The OS of my cheap Sigma lens is amazingly good when on a full-frame house and hand-holding "impossible" shots. But on the crop factor house (7D), where it becomes an 800mm, it just can't keep up.

Have a look at the image, in case the resolution here does it some justice. It was shot handheld at difficult angles from the car, 1/100s, and in my view shows just how good the OS is. For some reason you must click the image 'down' to get the best resolution. Having said that I will still be switching it off for most work from now on...



It'll be OFF from now on,

Mads S

[Manfred M]

The correct answer to this question is of course “it depends”. There are so many variables in play here that it is impossible to provide a definitive answer, but at a high level, from best to worst, all things being equal:

1. Non stabilized lens on a tripod;
2. Stabilized lens with stabilization turned off on a tripod; and
3. Hand-held stabilized lens.

It is quite unfortunate that Thom Hogan wrote and posted that article, because he does have much of it wrong. For whatever reason, the camera and lens manufacturers have not been particularly open about how image stabilization (IS) works; letting people speculate as to how these systems actually work. Some are more qualified to speculate than others.

There are two issues at play here, the basic physics of both motion (kinematics) and optics and then the practical engineering solutions as to how the physics are implemented through the various processors, physical components and controls. I’m going to stick with the high level physics as much as possible.

The only way to counteract camera motion is to move either the sensor (as Sony and Olympus do) or deflect a lens element (as per the method Canon and Nikon use). The ideal is to track exactly at the same speed as the camera is moving, both up and down and side to side, but in the opposite direction of the way the camera is moving. This means that the subject matter appears to be stationary to the sensor, in spite of the fact that the camera is actually moving at a velocity that varies during the exposure. There is a bit of a problem here though, because we don’t actually want to track at the current speed, but rather at the speed that the camera will be moving during exposure.

The IS lenses are equipped with accelerometers, and not surprisingly, these measure acceleration (and deceleration), i.e. the rate at which the velocity is changing. This is a very important bit of information, as acceleration and deceleration tends to be fairly consistent, even if the speed that the camera is moving is not. Engineers look at acceleration and deceleration as the same property and when describing acceleration, will do so with positive numbers, and deceleration with negative numbers. Something that has zero acceleration is moving at constant velocity.

I saw someone refer to these devices as “electronic gyroscopes” in an earlier posting. This is actually a bit backwards; 3-axis accelerometers can be used to build electronic gyroscopes, but not the other way around.

The next piece of information that is important is that the IS system does is sample or measure the acceleration of the camera at a 1 kHz frequency, or 1000 time per second. This allows us to monitor how consistent the acceleration is, and to correct for its impact. The technical name for changes in the rate of acceleration is “jerk”.

The camera designer, using the on-board processor can now design and program the IS mechanism to track the camera motion and correct for it. This is, of course, when theory and the real world collide and the engineering solutions start trading off various “real world” issues in pursuit of “real world” solutions. Real world components have don’t react instantaneously, may not position totally accurately, be affected by hysteresis and ambient temperature, etc. all mean that the results are going to be better than a hand-held shot, but probably not as good as with a shot from a tripod. I suspect there may be optical issues as well, but would leave that answer to someone who has more knowledge regarding lens design than I do.

The reason I rank a fixed element lens more highly than a IS lens that is locked is that the floating element can never be positioned as accurately as a fixed element, and that will lead to some (minor) level of distortion. I suspect this might be the reason that Nikon did not design their pro f/2.8 24-70mm lens with VR.