Image Stabilization vs. Image Quality . . Your Thoughts?

[FlyingSquirrel]

I can speak only from my own experience, as I do not have extensive knowledge of the technology of lenses and IS (as many others here obviously do have). My opinion / findings are, if the equipment is good quality, and the photographer understands the equipment and is skilled in its use, the posit that a lens with IS will produce lower quality images than one without, is not necessarily accurate.

That said, I have a Canon 500mm f4 IS II lens, and Canon does not make the exact same lens without IS. So technically I cannot address the exact claim that is being put forth. But I can tell you that regardless of whether or not my IS is on, I get razor sharp images. I simply do not see any degradation of image quality due to the IS.

[dje]

It actually helps to be as thick as Planck.....

Dear oh me, that would have to be a contender for "worst pun of the year"

[xpatUSA]

It actually helps to be as thick as Planck.....

(groan) . . which I Constantly am . . .

[Manfred M]

I can speak only from my own experience, as I do not have extensive knowledge of the technology of lenses and IS (as many others here obviously do have). My opinion / findings are, if the equipment is good quality, and the photographer understands the equipment and is skilled in its use, the posit that a lens with IS will produce lower quality images than one without, is not necessarily accurate.

That said, I have a Canon 500mm f4 IS II lens, and Canon does not make the exact same lens without IS. So technically I cannot address the exact claim that is being put forth. But I can tell you that regardless of whether or not my IS is on, I get razor sharp images. I simply do not see any degradation of image quality due to the IS.

Matt - so far as I can tell, the story of IS / VR not working well for long lenses seems to have originated in the pro sports photography community. Not doing any sports photography myself, I can't comment as to if this is really an issue.

I suspect that some of the early lenses may have not been quite up to snuff (I have the original version Nikon 80-400mm lens, which is the first Nikkor that incorporated VR, and it has a few quirks and the VR is not nearly as good as on my newer lenses). I also suspect that they already had a non-VR workflow so they stuck with their old shooting methods using monopods and other ways of shooting that I don't use.

So is it a real issue or is it a combination of the early technology coupled with a resistance to change in one particular user community? I suspect it is, but can't confirm it.

[pnodrog]

(groan) . . which I Constantly am . . .

That is even more obscure than mine......

[FlyingSquirrel]

Matt - so far as I can tell, the story of IS / VR not working well for long lenses seems to have originated in the pro sports photography community. Not doing any sports photography myself, I can't comment as to if this is really an issue.

I suspect that some of the early lenses may have not been quite up to snuff (I have the original version Nikon 80-400mm lens, which is the first Nikkor that incorporated VR, and it has a few quirks and the VR is not nearly as good as on my newer lenses). I also suspect that they already had a non-VR workflow so they stuck with their old shooting methods using monopods and other ways of shooting that I don't use.

So is it a real issue or is it a combination of the early technology coupled with a resistance to change in one particular user community? I suspect it is, but can't confirm it.

Good points, Manfred. Old tech is certainly suspect. My 500mm is the newer version and is fantastic. I will say I shoot wildlife / birds, which is basically similar to shooting sports. One thing I do on my shots is I always try to have as fast of a shutter speed as is reasonable. I think the IS would be more likely to cause problems at slower shutter speeds. Just a guess. Also, the newer canon IS lenses such as mine are OK to use on tripod, with IS set to ON. The lens can sense when it's on a tripod and adjusts the IS mode accordingly. Certainly a great lens, and well worth the painfully high price.

[mrinmoyvk]

The phenomenon you are seeing is called "jitter" and is pretty typical in any control system that is constantly trying to make a correction and overshooting a bit, correcting for that, etc. If the system were perfectly damped, you'd likely not see this at all.

If you shooting on a tripod with IS / VR turned on, I would expect to see more of this as the lens tries to correct for shake that isn't there. I'd expect that this would be more prevalent on older lenses as IS / VR seems to have become more sophisticated with each new lens release. The earliest stabilized lenses might only gain you a stop or two, but the latest ones are claiming gains of four stops or more.

If camera is on tripod with IS/VR set to ON, why would lens try to correct the vibration which is not there? I mean floating element would move to correct the deflection of camera but as camera is not deflecting, why would it move? Or is it like, it is always moving regardless of camera deflection?

Question somes after reading many posts saying turn off the IS/VR if using tripod...

[Donald]

If camera is on tripod with IS/VR set to ON, why would lens try to correct the vibration which is not there? I mean floating element would move to correct the deflection of camera but as camera is not deflecting, why would it move? Or is it like, it is always moving regardless of camera deflection?

Question somes after reading many posts saying turn off the IS/VR if using tripod...

It depends on the lens. You need to know what the manufacturer recommends. With some lenses, it's okay to leave IS/VR on, on others it is not.

[Inkanyezi]

Yes I would certainly not go as far as ignorance is bliss. However often I see long discussions on the finer points of technology when in practice they have no observable influence on the result except in extreme circumstances.

A hazard of the job. A number of the members (I have to include myself here) have technical backgrounds and our interests in some of the minutia as to how things are put together or work will influence the equipment we buy. As an example, I stayed away from both Sony and Olympus cameras because of how they implemented image stabilization.

An interesting statement, and I cannot but agree on the point that how things work is interesting, and we may have different takes on the importance of one or another technical feature, or a method, about how it might influence our achievements. In this case, it is about image-making, and how we might assume that a particular technique might enhance or ruin what we try to achieve. However, as this mostly is discussed as moot points, very little real world experience is included in our decisions about which system we might prefer. The exception of course is that in sports photography, real world experience mostly excludes the use of the stabilising feature, for a very simple technical reason.

I'll delve a bit into this technical reason here.
To understand the problem of stabilisation in sports photography, the simplest metaphor that most will understand is how we use the computer mouse, as in a few respects it shares the control algorithms of image stabilisation, whether within the lens or moving the sensor.
Stabilisation always will have limits, it does not work on an infinite space angle. When intentionally moving the camera, this limit will be reached over and over, and each time you hit that wall, where stabilisation will no longer be possible in the chosen direction, the system must find a new "fix" direction. It does this by swiftly moving the relevant element to the other side, in order to start over. This is similar to when we lift our mouse to get a new point to start, when we are about to slide it past the edge of the mouse mat, or whatever impedes further movement in that direction.

So the sports photographer, with stabilisation "on", will many times have the lens causing maximal blur, moving at a higher pace than the actual camera movement, only in order to find a new direction in which to stabilise. This will happen regardless of where in the system the stabilising component is situated, lens or sensor, same problem.

Now over to the decision of choosing one method of stabilisation over the other. I wonder why? I have seen so many discussions about it before, and I agree, that the stroke of the sensor will have to be very short, which might be too little for a very long focal length, compared to the possibility of optically stabilising a very narrow angle of view. However this caveat is valid only for very long focal length. On the other hand, optical stabilisation always will misalign lens components, while sensor stabilisation does not, which is an argument against optical stabilisation in favour of sensor based solutions. They also have the advantage of not having too many loose things rattling about in the lens, where the moving elements are rather heavy in comparison.

Those are of course technical caveats that may or may not be properly resolved.

There is however nothing that excludes having the best of two worlds. Cameras with in-body stabilisation can work with optical or sensor stabilisation at the choice of the operator, while a camera with no sensor stabilisation will only be stabilised with lenses that have optical stabilisation. Considering the fact that some lenses are impossible to stabilise optically, in-body stabilisation will have a great importance for very fast lenses of symmetrical build. There is no way to stabilise Gauss type lenses, which is a usual design for very fast lenses. Even if they are made in non-Gauss fashion, their components will invariably be heavy, which technically is a disadvantage, while in-body stabilisation will cope with them in exactly the same way as any other lens.

There is also an advantage with in-body stabilisation, which is impossible to attain by optical stabilisation; at least without adding further linearly working optical components. The lens, being essentially rotationally symmetric, cannot stabilise for roll movement, but the sensor can. Most newbie camera shake includes roll movement, caused by tapping (not pressing) the shutter release that is placed at one side, inducing roll movement.

And again - the designer may choose the ability to use both systems simultaneously. When a camera has both optical and sensor based stabilisation, the system may choose the best suited method for each type of camera shake. With a very long lens, there might be an advantage to have stabilisation in the lens, but as it cannot control roll, roll stabilisation may be applied at the sensor simultaneously with pitch and yaw stabilisation in the lens. It is probably not done, but the possibility is there. Why exclude one system? That's what makes me curious about the arguments for choosing one system over the other. Why exclusion?

[Dave Humphries]

There are many technically valid points there Urban, some I'd already been aware of, others not. I like your mouse and mat metaphor.

With regard to panning, do you know if any lens or body systems can detect the motion of pan and instead of doing the 'lift (mouse) and rapid reposition', could 'zero out' the (relatively) steady state velocity of the pan so the IS/VR just deals with the irregularities? Maybe it's not as practical an idea as I think.

I have experienced problems with the edge of the mouse mat when following birds of prey in flight

A few years back, I did some comparative tests with and without IS and at various shutter speeds, averaged over 10 shots; from memory, I was slightly better off with VR on. However, that may not be the case for other subjects, for reasons you describe.

[Inkanyezi]

I don't know of any system that would assess panning automagically in order to stop stabilisation for yaw. I certainly know that my Olympus OM-D does not. However, as I use many old manual lenses and also am curious about how much shake the system will accept before the image is blurred, I have tried the limits for stabilisation and got up to four seconds for free hand shots. They will not be perfectly sharp, but sufficiently, if I hold the camera reasonably in the same direction. It is however difficult as the viewfinder is blacked out, so I have tried it with an external sight device mounted on the camera, and it is amazing how well it works. As long as the fixed point is kept in the same direction, the image is stabilised.

When panning slowly, the stabilisation system first responds by keeping the first direction a while, then slowly following. Panning a bit faster makes it unstable.

I guess it is one of the trickier problems to solve, assessing a panning movement and stopping correction in that axis. Most systems have manual override for panning.

The only µ4/3 lenses I have are both Panasonic lenses with optical stabilisation "MEGA OIS" but I set them to off position and use the in-body stabilisation, which seems to work better. With other lenses, I set the focal length manually and get stabilisation also with the old manual lenses or when shooting macro with extensions. When using extension it is added to the focal length setting, so at 1:1 with a 50 mm lens, I set it to 100. With a 10 mm extension, I set it to 60.

[Manfred M]

Urban - I believe I now understand why you have written what you have in this latest as well as you previous posting one this subject.

First things first. I would suggest you go back and check your camera's specs. When I was out looking for a mFT camera around 15 months ago the only mFT camera that was available at the time that offered both in-lens and in-body stabilization was the Panasonic GX7. I had two Panasonic lenses that I use with my Panasonic video camera (14-140mm and 100-300mm) and was looking for a small still camera I could use those on.

None of the Olympus models I looked at utilized the in lens stabilization and used the in-body stabilization only. The GX7 uses the in-camera stabilization only with lenses that do not have the Mega OIS functionality. According to my manual it is not possible to use both in-lens and in-camera stabilization systems at the same time. Thinking about it for a few seconds; trying to design a system that simultaneously applies both in-lens and in-body systems is bound to not work as the two systems would essentially be fighting each others corrections and giving you a worse result that having no stabilization at all.

I suspect the reason that you find that you got better results with the in-lens stabilization turned off is that your camera was not able to use it and you were most likely shooting with no stabilization at all. I'd suggest that you might want to look at your camera manual to see what it says.