Image stabilization - why on lens?

[kram]

I confess to not knowing anything about how IS lenses manage to reduce the impact of shakes. It may be a factor of balancing the movement with internal counter-movements?

That aside, why is it that IS has to be in the lens? A logic that I would like comments on : In a digital camera, if the sensor is bigger than the size of the actual recorded image, any information on shakes will get recorded which can then be corrected.

[weemannie]

I've read about how IS works - and it went straight over my head!:D

Suffice to say that it does work, and is of considerable assistance when hand holding.
If you introduce camera shake into a shot, then there is nothing you can do to correct that in post processing.

Many people here love their IS lenses, but I find them a lot heavier than their non-IS counterparts, and anyway, for my style of shooting, I use a tripod 99% of the time,so IS is unnecessary. :)

It entirely depends on your style of shooting.

[PaulB]

IS in the lens can be fine-tuned to that particular lens design.
IS is not usually needed on wide-angle/normal focal lengths.
IS in-camera is not yet proven over a long time period, there may be problems with electrical connections from the sensor to the PCB for instance. The jury is still out on this implementation.
You can see when lens IS kicks-in, there is no such indication when/or even if in-camera IS is working.

[tim]

I believe Canon put IS in the lens because that way it's compatible with both film and digital cameras. You can't move film around like you can a digital sensor, or more likely some kind of filter over the sensor. With film dying out I wouldn't be suprised to find IS moved into the lens, that's a competitive advantage of some other manufacturers right now. I've read that the lens is the best place to do image stabilisation, but so long as in-camera works ok the cost benefit would probably override that.

AFAIK it works by having two internal gyroscopes moving a lens about to counteract movements of the lens. It can't cope with much movement, perhaps a few mm in my 70-200 IS lens, but I find it quite useful. Like people said, motion blur can't be corrected after the shot's taken. Once you want to take more than a few pictures you need to get them pretty much correct in-camera, otherwise the post processing time will be unfeasible.

[kenyc]

Minolta put it in the camera body in theirs so you could use any lens. Since it also has the effect of moving the CMOS sensor it could be used to keep it dust-free :)

KAC

[chris clements]

Useability with film bodies is important, although will wane.

There's a direct analogy with the birth of AF, when Canon went for a tailor-made, self-contained solution in each lens, whilst the others went for a clunky driveshaft from the body.
We all know which came out on top!

[tmurphy]

Even if IS was embedded in the body, surely you'd still need the gyro's in the lens, as most of the movement would be in the lens, and only a fraction of that movement would be apparent at the body, as a tiny movement at the body is greatly increased at the far end of a 400mm lens!

Couldn't live without my IS on my 100-400L, and it's the only thing making me have to save up a little extra for my 70-200L because I want the IS version!

[chris clements]

If you have to, sell a kidney or your wife to raise the money for the 70-200L IS.

[ron chappel]

Having IS in the body is a FAR superior solution than having it in the lens-in spite of the naysayers 'amateur engineering' opinions:)
Sadly it looks like market forces and dumb luck will keep us (canon users) from having in body IS for the forseeable future-canon don't see any threat from minolta's system as yet so will keep on making lenses with the individual (and expensive) IS built in.

If you only need one IS lens then canon's system works out cheaper.If you wan't to use several lenses with IS then minolta is the go

[bauerman]

Why would they see a threat from Minolta - in the SLR market they are really not even on the same playing field. And as far as FAR superior goes - I think that has yet to be proven. Minolta is a niche manufacturer of DSLR's at this point and IS in the camera is niche technology.

[slin100]

Having IS in the body is a FAR superior solution than having it in the lens-in spite of the naysayers 'amateur engineering' opinions:)

I'm curious. Why is in-body IS superior to in-lens IS? I read that in-lens was theoretically superior because it's easier to make corrections closer to the lens' principal node than at the sensor plane.

[Longwatcher]

okay time for my limit knowledge of optical theory here.

IS is better in the lens because you can correct the shake with the least amount of movement by putting the correcting optics near the center of the lens. If placed correctly very little is needed despite any changes in focal length.

The gyroscopes do not move the optics, they only detect the movement intesity and direction and then tell servos when/where to move the IS portion of the lens.

The two fold problem with putting the IS in the body is
1. Electronics; your sensor would have to physically move at high speed; thus the longevity of your camera will be shortened.
2. the longer the lens, the more the sensor would have to move to compensate, thus the IS will probably not work as good as the focal length of the lens gets longer.

Lastly I would rather have a lens lose IS on me then the camera, because I can change lenses (or maybe use it without IS) and keep shooting, if camera sensor IS goes out, odds are good, the camera can't be used. I don't know about your kit, but mine contains 1-2 cameras and 3-7 lenses depending on travel, needs, adn risk.

Just my knowledge,

[ron chappel]

I'm curious. Why is in-body IS superior to in-lens IS? I read that in-lens was theoretically superior because it's easier to make corrections closer to the lens' principal node than at the sensor plane.


Ah,ok.I better explain my hasty unclear statement:)
I mean that in-body IS is better because it can be used with a whole bunch of different lenses -that's all.

As for which is better engineering wise,it doesn't matter much.Both systems have to move precision parts (optics groups or sensor) in an extremely acurate way-there isn't really much difference when you get down to it.
The fact that the sensor has wires running to it is not a serious problem.
As for the sensor being heavier than the lens element group-that just means the sensor based IS will have to use more battery power to do the same.

Overall the lens type IS has slight performance advantages that are easily equalled by the sensor type IS at the expense of power consumption and (maybe) extra cost .
Overall the reason i'm so keen on the in-body IS idea is because it can be used with some incredible lenses that would normally never be made in IS versions.

I haven't heard that about the lens IS being better because it can correct close to the lens' principal node,but it does sound interesting ....i'll have to think about that abit :)

[PhotosGuy]

I wonder when they'll come up with a software IS? Since the cams are mostly computers anyway, it doesn't seem that it would be that hard a "fix" to me. (But I've been known to be wrong before!) ;-)

[kawter2]

it is all abou the

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$$$$$$$$$$$$$$$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$$$$$$$$$$$$$$$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$$$$$$$$$$$$$$$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$$$$$$$$$$$$$$$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$$$$$$$$$$$$$$$


Why sell the IS one time with each camera when they can sell it over and over again with each lens!!

[slin100]

Ah,ok.I better explain my hasty unclear statement:)
I mean that in-body IS is better because it can be used with a whole bunch of different lenses -that's all.

Ok, I was aware of this. Thanks for the clarification. I don't think one can equate the performance of in-body and in-lens IS so easily, power consumption issues notwithstanding. I've seen accounts, but no definitive evidence, that in-lens IS can counteract more camera shake than in-body, something like 1 extra stop. Until someone does a rigorous test, the jury is still out on this one.

[slin100]

I wonder when they'll come up with a software IS? Since the cams are mostly computers anyway, it doesn't seem that it would be that hard a "fix" to me. (But I've been known to be wrong before!) ;-)
I don't know if they still do, but Sony used electronic IS in their camcorders. I have an old Hi-8mm camcorder that has this feature. Back then, Canon had optical IS on their camcorders and it was regarded to be superior, which is not that surprising. Imagine trying to correct a smear of pixels in post processing.

[kram]

I was actually referring to the solution proposed by photosguy - the IS can be software based. After all, with shake, the sensor is getting various images of very close similarity - differentiated only by the extent of camera shake.

That would seem to imply, the oftware can merge all the images and correct for the IS. And absolutely true, having the IS on the camera is a matter of conveninence. I would assume all serious photographers get to somewhere near 1:5 body vs lens. And 1 body having IS removes a lot of bulk from the lenses too.

[lostdoggy]

One thing that KM 7D can't do is Stabalize the view in real time. Their IS will only become active moments before the shutter opens. Where as the In lenses IS you would get real time IS. I don't like the idea the the sensor is shaken like a milkshake everytime I depress the shutter button. And as far pricing is concern the KM 7D is priced at $1600 and there is report that on some lenses are vignetting especially on wideangles cause by the IS.

[Longwatcher]

Digital cameras 101 - if I remember my technolocgy correctly
each pixel/detector only records one value. This value is based on the number of photons detected that reached that pixel/detector before the value is recorded and dumped (actually I seem to remember it is dumped as it being recorded), which is why you end up with blown highlights, because the number of photons exceeded the capacity of the pixel/detector to record beyond a certain number.

The problem with using electronic IS on this kind of sensor is that if the photon from point A ends up in detector B's bin because of shake, there is no easy way for software to know this if at all. I believe Sony was just recording multiple takes of the data and then compiling them, which allowed for some electronic IS adjustment to a limited extent. as it could measure that the bright points had moved over to a different pixel for one or two images and then moved back.

Try this experiment as an example:
Set up about 9 glasses in a 3x3 set, now take a tenth cup and randomly spray water into the cups. Stop when one or two fo them are full. Now measure the levels in each of the nine cups. If you overflow a cup that is the same as blown highlights. You don't know how much it overflowed by. This is how your sensor works. to simulate a bayer sensor add one drop green dye to every other glass and then alternate red and blue dyes amoung the remainder (actually they do it in rows of red/green and blue/green).
Now take the center glass and empty it into a larger container and add half the water in the non-diaganol neighboring glasses and 1/4 the water in the diagonally neighboring glasses (actual amounts will differ based on scientific analysis). Now what color is the water in the large glass (and actually you could do some math to get a more accurate color instead of a muddy brown) anyway you get the idea.

Now for IS. Have someone come up with the amount that should go in each glass without you seeing the amount that is supposed to go into the glass.
Have that person try to pour that amount into each appropriate glass, while a second person is shaking the glasses using a place mat they are on. now you try to figure out how much was supposed to go in each glass.

To simulate a mechanical IS on the sensor have someone stand on a three story building to pour the water while moving a foot to either side, try to make it go into the correct glass at ground level. Then try again with them only 1 floor up and they can only move 6 inches, relatively much easier yes?

Hopefully I did not mess up my analogy since I had a break in the middle, but also hopefully you get a feel for what is happening so you can understand why somethings are better then others when it comes to digital sensors and technology.

BTW: for pixel/detector size use bigger/small glasses. Way easier to catch photons (or water) in bigger detectors (glasses).

Just the science as best I can explain it and remember it, I am open to corrections.

[PaulB]

Canon have IS in their lenses because they needed it to work with film all those years back - no sensor to shake then!

[KennyG]

Can I add to this discussion? With IS in the lens you can see the effect of IS when using a DSLR. With it applied to the sensor, or in software you couldn't. Great for the P&S boys looking at images on an EVF however, but we don't have that type of design, thank god.

From my discussions with Canon engineering a few years back, IS in the lens allowed them to optmise IS for each lens. I would not like IS in the camera body for a number of reasons, the one size fits all being the main one. Plus, I like the idea I can move my lenses onto any EOS body and retain IS.

[tim]

Why sell the IS one time with each camera when they can sell it over and over again with each lens!!

If they don't someone else will. How many new SLR purchasers would start with nikon setups if they had IS in the camera instead of in the lens?

[KennyG]

As 90% of people buying DSLR's are relative beginners, then probably anyone who would believe the hype. Funny how even Nikon use stabilized lenses. Doesn't that tell you something when the two top DSLR brands take the same approach.

[lostdoggy]

The original AF actuator was body mounted and later Canon introduce Lens mounted AF. The reason for this is because there were limitation in the size of lens that can actually be mounted on the body and torque needed to focus the lens. The limitation exist in in camera IS. There will definitely be vignetting in wide angle lenses and vignetting beyond the 3 stop. It at present will not work on tripods and panning.

[Viking Joe]

As 90% of people buying DSLR's are relative beginners, then probably anyone who would believe the hype. Funny how even Nikon use stabilized lenses. Doesn't that tell you something when the two top DSLR brands take the same approach.

I didn't know that Nikon had stabilized lenses. I thought Canon had an exclusive patent on them, and that's why Tamron, Sigma, and other third parties didn't make IS lenses for the Canons. I guess I was wrong.

[tim]

I didn't know that Nikon had stabilized lenses. I thought Canon had an exclusive patent on them, and that's why Tamron, Sigma, and other third parties didn't make IS lenses for the Canons. I guess I was wrong.

http://www.camerahobby.com/Review-70-200mmVR.htm

[lostdoggy]

I didn't know that Nikon had stabilized lenses. I thought Canon had an exclusive patent on them, and that's why Tamron, Sigma, and other third parties didn't make IS lenses for the Canons. I guess I was wrong.

Sigma does have a IS lens for Noink. 80-400 EX OS, OS for Optical Stablizer.
http://www.sigma-photo.com/lenses/lenses_all_details.asp?id=3272&navigator=3

[Phil V]

I didn't know that Nikon had stabilized lenses. I thought Canon had an exclusive patent on them, and that's why Tamron, Sigma, and other third parties didn't make IS lenses for the Canons. I guess I was wrong.

Nikon have several VR (vibration reduction) lenses.

Sigma don't only produce IS lenses for Sigma and Nikon, you can have EOS mount too.

[ron chappel]

There is speculation that nikon had the first patent on image stabilization but canon somehow got around it.
Nikon were certainly the first to bring out a stabilized product- a top of the range point and shoot film camera with VR built in.That was about 1994 i think.

Whatever the case was,canon have really taken the lead since.
Now minolta have the chance to make in-body stabilization the thing to have.

[kram]

Just read this news article about Konica Minolta's foray into low end DSLRs - and they feature an anti-shake mechanism.

http://today.reuters.co.uk/News/NewsArticle.aspx?type=technologyNews&storyID=2005-07-15T122857Z_01_N15104929_RTRIDST_0_TECH-TECH-JAPAN-KONICAMINOLTA-DC.XML

link for those interested in the entire article.

Maybe one day, it will be a common feature across many DSLRs.

[Poco]

Come on people, think outside the box. What about an IS Extender? A 1x extender (or higher magnification) with IS built in could allow you to add IS to any lens. I'll bet that with no magnification it could be done with limited lens elements and image degradation. Heck, if Canon built it into the 2x tele-extender it could help compensate for the lost aperture.

I wonder if anyone has a patent on that?

[CyberDyneSystems]

It's really about Legacy.

A Quick Lesson On Image Stabilization History

1. Canon invented IS for affordable FILM SLR systems.

2. It was installed on the lenses before Digital really existed. It was film only. At that time, obviously, the lens was the only option. Canon had a fleet of IS lenses before it ever really got into the DSLR market.

3. Minolta's solution is an innovation for sure.. but it was only possible/pracitcal for digital. making Minolta about 8 years behind Canon regarding adoption of IS.

4. With the exception of EFS, Canon's entire lens line is legacy holdovers from 35mm.. just like the IS. IS will not be reomved from the lens for Canon untill Canon (if Canon ) decides to migrate it's DSLR system away from the Legacy EF mount lenses. (when pigs fly http://members.bellatlantic.net/~vze2hdzu/pics/cow.gif or about 10 years from now.. which ever comes first)

[PhotosGuy]

From my discussions with Canon engineering a few years back, IS in the lens allowed them to optmise IS for each lens. I would not like IS in the camera body for a number of reasons, the one size fits all being the main one. Plus, I like the idea I can move my lenses onto any EOS body and retain IS. Another reason would be, if the IS goes out on one lens, you can 'restore' it by putting on another. Redundency rules when you're shooting for your supper! ;-)

[Croasdail]

As 90% of people buying DSLR's are relative beginners, then probably anyone who would believe the hype. Funny how even Nikon use stabilized lenses. Doesn't that tell you something when the two top DSLR brands take the same approach.

It tells me one thing.. that both companies have a huge install base that they need to protect. If we learn anything from history is that true inovation comes from the smaller companies because they have to do radical things to be relivant. The good news is that good technology eventually makes it into the bigger companies. Never say never - if the argument and the economies of scale prove out the technology - anything can happen - like a Mac running and Intel processor. I am glad there are companies like Sigma and Minolta kicking the big guys in the shins - it will keep them from becoming complacent.

And lastly, don't assume anything is absolute in technology - almost every rule has been broken and will continue to be. That is why I don't mind spending on lenses right now because there is a whole lot more change coming on the camera body side - the games have just begun.

[Cadwell]

I didn't know that Nikon had stabilized lenses. I thought Canon had an exclusive patent on them, and that's why Tamron, Sigma, and other third parties didn't make IS lenses for the Canons. I guess I was wrong.

Sigma does make an image stabilised lens for the Canon mount. The Sigma APO 80-400mm OS. The third party manufacturers are a little behind the game in terms of developing image stabilised lenses though.

From a personal standpoint I am not a big fan of IS. It's not particularly useful for what I do but having had a Canon IS L series lens fail on me due to faults in the IS system, I am glad I was in a position to swap it out for a non-image stabilsed Sigma EX lens and keep shooting... I'd rather not have IS in the body.

[ScottE]

With in-lens image stabilization a deflector lens near the nodal point is pivoted vertically or horizontally to counteract lens movement. The amount of deflection is very small because it is near the nodal point.

With in-camera image stabilization the image sensor must be moved vertically and horizontally parallel to the image plane. Mechanically this has several problems such as the mass of the image sensor, the greater distance that has to be moved and the more complex design of a sliding vs a pivoting mechanism. Congradulations to Konica Minolta for being able to design and produce this mechanism. Only time will tell how well it performs and how durable it is.

Someone still has to design and produce an electronic image stabilization mechanism for an SLR. This would use a sensor larger than the image size and each of exposure would consist of a series of tiny time bits, say 1/10,000 of a second. Each of these bits would be aligned and cropped electronically to produce the final image. The computer processing required for a large image is a major problem as is the reliability of the algorithm to detect camera movement and separate it from either subject motion or camera panning. If these problems can ever be resolved, this system should be better and more durable than either mechanical stabilizing method.

[kram]

Absolutely with you Scott. In fact, given that we already have cropped sensors, the sensor being bigger than the image recorded is not going to be an issue. The crop %age may go up marginally. Esp. when the crop %age is likely to come down with advancement in sensor technology, it could be partly used up for IS.

The key is memory of each sensor to record more than 1 image - it may mean changes to what we consider exposure today - so that the processor can put together an image correcting for movement...

Of course, I dont expect to see it from Canon anytime in the next 5 yrs, unless a rival DSLR does start capturing marketshare due to IS.

[SolPics]

I'm not sure exactly how it works but WOW! It makes the difference between sharp and blur at 200MM.