in-body image stabilization versus in-lens stabilization?

Canon puts their image stabilization technology into individual lenses, while Minolta/Sony build it into the

camera bodies. From a user's perspective, is there any advantage to one over the other? (Besides the

obvious: if I have an IS lens, it will work on all my bodies; if I have an IS body, it will work with all my

lenses.)

I have a camera with AS in body and works great, I can use old lenses, at a cheaper price that an expensive IS lens.

 

It gives me 2.5f-stops extra but the new Sony alpha will give you 3,5 extra f-stops.

<<It gives me 2.5f-stops extra but the new Sony alpha will give you 3,5 extra f-stops.>>

 

It is important to differentiate what Sony /claims/ the Alpha will do and what it /actually/ will do. There have been no tests (that I've seen) that provide any real data on how many extra stops the Sony stabilization method will /actually/ give you.

If I presume that the 'in camera' stabilization is done from the image chip and software then it won't work with SLR optics unless, that is, there is a secondary chip doing the work. Even then what moves to correct the camera movement?

 

I suppose that gyros in the camera could move the main sensor to compensate, this would mean the chip would be on a floating gimbal or the sorts. Quite complex and very susceptable to damage. Particularly when cleaning.

Jeff >> Yes, it's the sensor that moves to compensate for movement.

 

It is possible to develop algorithums to compensate the movement with software but that requires lots of computational power and cannot be done effectively on the camera in real time.

<<Quite complex and very susceptable to damage. Particularly when cleaning.>>

 

Which may be one reason Sony provided in-camera dust removal by way of an anti-static layer and high-frequency vibrations.

I seem to remember that the AS in the minolta body corrects for one less degree of freedom than the IS in the lens. Is this true?

 

Also, even though canon IS lenses are expensive, they might still be less expensive than the non-IS minolta counterpart which means going with IS in the body might not save you any money at all.

One significant difference in terms of the user's experience is that, with in-lens stabilization,

you can see the stabilized image in the viewfinder! With in-body stabilization, the sensor is

stabilized (its position is shifted around based on gyros), so when the shutter opens, hand

shake is accounted for, but that doesn't help you frame or time your shot.

 

A side-effect of the in-body approach is that the same equipment that shifts the sensor can

(in many bodies) also be used shake dust off the surface of the sensor on power-up.

Somewhat off topic, but... I think it will be interesting to see how the image stabilization battle plays out over the next few years. Canon/Nikon want to protect the price premiums they can charge for IS lenses, while customers want to have IS working with all lenses. From an engineering standpoint, I suspect putting the IS in the body makes a lot more sense (less mass to move, and over a much smaller distance.)

 

This is a classic case of the established business model ("charge $400 extra PER LENS for IS") clashing with both the engineering factors and customer desires. My guess is that Canon and Nikon will both hold off on bringing IS into the DSLR camera bodies as long as possible, eventually giving in to demand and canabalizing their market for IS at the lens level. But what do I know?

In theory, you don't even need to turn off the camera. Just jump up and down with the lens cap off. It is natual move after chimping :-) You will soon see former Minolta macro shooter refine this to an art :-) :-)

I have little experience with build it into the camera bodies. But I had the opportunity to try a Minolta 7D for a couple of days. It worked flawless but for me it was annoying no to see the effect in the viewfinder but certainly I would adapt to this. As I shoot a lot with telephoto focal lengths in IS mode 2 (mostly the 4.0/500mm) with great success I was disappointed with the results of using the equivalent mode of the Minolta 7D and the 4.0/600mm.

I also found that particularly with these two lenses the IS worked more reliable with the Canon system. There was no such difference using zooms with shorter focal lengths (for example 28-135mm or 80-200mm). Maybe the IS in the lens is a advantage for telephoto lenses.

 

Regards Gerhard

Why cant they just use a bigger sensor to provide some sort of buffer (around a "cropped" frame) to accomodate for movement? So this will allow the camera to be used in either IS-on:"cropped frame" OR IS-off:"full frame mode". I do not think it requires that much more processing power! Maybe they can also use the LCD as the "viewfinder" in IS-on mode? BTW, I am new to SLR/digital technology so I am not sure if all this is possible or not :)

For those who may not have been following developments: the in-camera IS in the Minolta 7D (moving sensor) forms the basis for the in-camera IS in the new Sony, because Sony picked up Konica-Minolta's DSLR line. The name of the camera itself, the Alpha, actually comes from the Japanese name of the Maxxum line: Maxxum = Dynax = Alpha, but now it's all Alpha. The Alpha will use the old Maxxum mount.

<<Why cant they just use a bigger sensor to provide some sort of buffer (around a "cropped" frame) to accomodate for movement?>>

 

Because bigger sensors cost big money.

<<Why cant they just use a bigger sensor to provide some sort of buffer (around a "cropped" frame) to accomodate for movement?>>

 

how do you just "accomodate" for movement?

 

Yes the bigger sensor will caputre the original position and the shifted position but the image will be blurred. To "un-blurr" an image that was blurred due to motion requires huge processing power. It can only be done with iterative techniques and these techniques require the knowledge of the motion path. Not only that, the optimal gain in these iterations does not have a predetermined optimal value and the worst part, even with all this information, a non-divergent solution to the iteration is not guarnteed.

I think the answer has already been given, with the stabalisation on lens you see this the effect in the viewfinder, quite usefull. Of course this is not an issue for non-SLR cameras.

I can imagine how useful it is to see the stabilising effect in the viewfinder, particularly with long lenses, but are there any other advantages to having IS in the lens rather than in the body?

let's have both. why not?

With lens stabilization you can tune the whole system to optimize stability. You need a very different set of control parameters to optimize the stabilty of a 17mm lens vs. a 600mm lens. For example the image moves around a lot faster and with much greater amplitude at 600mm than at 17mm.

 

For general purpose work at under 100mm they probably work about the same, but I've seen reports that suggest that for long telephoto work - where you need a lot more movement to compensate for lens shake - the Canon system works better.

 

There's no reason (other than pride and patents) you can't have both.

"let's have both. why not?"

 

Yes, what would happen, if you had both in-camera and in-lens IS - would you get 6 stops of stabilization?

Unless the two stabilizers are coupled appropriately, you can't have both. The lens is built so as to keep the image stable on an assumed fixed-relative-to-the-camera sensor. If this assumption is violated and the sensor is also jiggling around, the effect is that you'll get essentially no stops of improvement.

Thanks for the many comments; this answers some questions. Still, it seems that the

biggest advantage of in-lens stabilization is that you can see it in the viewfinder. (As for

tweaking the technology for the particular lens, I'll put that in the "engineering viewpoint"

category -- it's the same reason that Canon decided it made more sense to equip each

individual lens with its own AF motor, rather than building this into the camera body.)

<P>

I did want to addresss one comment: <I>Canon/Nikon want to protect the price premiums

they can charge for IS lenses, while customers want to have IS working with all lenses.</i>

<p>

As an economist by training, I have to think that Canon *could* devise some scheme that

implements in-body IS, while *improving* their profit margins -- especially given how

they've segmented their market. A person who uses a 30D might buy a single IS lens at the

$300 (or so) price premium that it involves, but might be willing to pay $500 for a camera

that provides it with all lenses. A 5D user might buy two or three lenses, but the price of

the camera could be jacked by $1000 for a body that provides it with all. A DRebel user

who would never buy an IS lens might still pay a couple hundred bucks more for a camera

that offers the ability.

<P>

Obviously, this is a very simplistic example (and I know little about the specifics of the

camera industry). But the basic principle should be the same: a better product package can

command a higher price, which translate into higher profits for the company. If Canon

doesn't choose to go this route, I'd presume there's a reason that they don't feel it's a

superior package, whether it's technology or performance of in-body versus in-camera

stablization.

There's no practical way to move the film around so IS/VR has to be in the lens and Canon/Nikon still have a lot of film users. Personally I would like to see IS/VR as a selectable option in the body so that you can turn it on for non-IS/VR lenses. Does anybody honestly think that Canon/Nikon will add IS/VR to every prime lens? I'd love to have it on my 28/55/85 f1.4 lenses. Maybe you could shoot by candlelight with that kind of setup.

<I>I've seen reports that suggest that for long telephoto work - where you need a lot more

movement to compensate for lens shake - the Canon system works better.</i><P>

 

I've heard that also. Never seen anything like an objective test, however.

I agree with Bob, I think lens based IS is likely to work better across an entire system. Why?

Because it can be individually optimized for a particular lens. The problems of a

supertelephoto are bound to be very different--or more demanding--than a wide zoom.

However for a point 'n shoot body IS is the way to go as the lens is fixed.