Why No D-SLR In-Camera Stabilization

[Voff]

If anything, the lens
base system would be more responsive as it is moving less mass.

Are you sure there is less mass involved?

In
addition, the lens system can be tuned to the particular properties
(mass distribution, dimensions, etc.) of the lens. A camera based
stabilization system does not know, in general, the lens properties
except for perhaps the lenses made by the same manufacturer.

Lens data is transmitted to the camera and used for calculating compensation.

Then
there are the advantages of stabilizing the image in the viewfinder
and in the AF sensor system.

That is purely a matter of taste. Personally I don't like stabilized viewfinder; I would like to see how much I shake. This helps me keeping more still increasing my chanches to get the shot.
A stabilized viewfinder/ optical IS also use for more power.

Having said all this, camera based stabilization is a good thing as
it increases the performance envelope of cameras that have it over
not having it. Like a lot of things, in the stabilization business
you pay more dollars, you get somewhat more performance. I expect
that both Nikon and Canon will implement some sort of body based IS
in their entry level DSLRs when and if they feel the competitive
heat from other brands.

What I have seen from pictoral results is that in body IS works. Even with a 600mm lens with 1,4X converter gets 3-4 stops gain which is basically as good as it gets.

The price difference is not necessarily related to performance but due to the fact that optical IS is far more expensive to implement. A stabilized prime lens often end up with same number of glass elements as a zoom lens due to the need of movable and aberation correcting elements.

 

[Voff]

While it may very well be proveable that one is better than the
other.

I'm sure some are better than others. But this have probably nothing to do with optical vs. in body but how well the software can predict and compensate for movements. Software can be upgraded; at least for cameras.....

 

[TrapperJohn]

On the extreme telephoto end, C/N may be right in that in lens IS would be more effective.

However, no company in the technology business likes to admit that they've been eclipsed by a new technology. Product image in a very image conscious business demand that they at least make an effort to support their current platform. Otherwise, they begin to look 'behind the times'.

Also, that might tend to irritate the customers who have invested heavily in IS lenses.

The main advantage of in body IS is that it's cheaper. Might be a little less effective, but the cost difference is huge, when looking at three or four lenses. However, if the in body IS that Pentax, Sony/Minolta, and Olympus now feature on their dslr's continues to drive sales, I wouldn't be surprised to see in body IS appear on some low end C/N bodies.

 

[Joseph S Wisniewski]

Lens based stabilizers can put the accelerometers much farther from
the center of gravity of the camera, so they produce much stronger
signals.

Doesn't matter where you put an angular rate sensor on a rigid
body. You don't think they use linear accelerometers?

Yes, actually, I think they do.

In pairs. Take the difference and it's a lot better at high frequencies than a rotational accelerometer.

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[Henry Richardson]

A stabilized prime lens often end up with same number of glass
elements as a zoom lens due to the need of movable and aberation
correcting elements.

It seems that IS lenses are also larger. I am still somewhat perplexed at the huge size difference between the Canon 17-55mm f2.8 IS and Tamron 17-50mm f2.8.

--
Henry Richardson
http://www.bakubo.com

 

[Joseph S Wisniewski]

The real reason is that Nikon and Canon are the market, they don't
need gimmicks to try to break into the market.

..........because I wouldn't mind a stabilized Nikon body to
go with my 85mm f1.4 and 135mm f2.0. Let the body stabilizer
disengage peacefully when the superior stabilizer of the 70-200mm
f2.8 is available.

Take no offense, but I thought that was a little funny.

I do take offense. It was not "a little funny", it was very funny. You don't appreciate humor, even when it borders on art.

You
mention it as just a gimmick that Nikon or Canon doesn't need. But
then stated you would like it on a Nikon.

A gimmick doesn't have to be useless, it just has to have a hype value.

Apologies for this, I'd rather quote Websters than dictionary.com, but Websters weighs 10 pounds and I'd have to retype it...

http://dictionary.reference.com/browse/gimmick

Dictionary.com Unabridged (v 1.1) - Cite This Source

gim·mick ˈgɪmɪk Pronunciation Key - Show Spelled Pronunciation[gim-ik] Pronunciation Key - Show IPA Pronunciation
–noun

1. an ingenious or novel device, scheme, or stratagem, esp. one designed to attract attention or increase appeal.

2. a concealed, usually devious aspect or feature of something, as a plan or deal: An offer that good must have a gimmick in it somewhere.

3. a hidden mechanical device by which a magician works a trick or a gambler controls a game of chance.
4. Electronics Informal. a capacitor formed by intertwining two insulated wires.
–verb (used with object)

5. to equip or embellish with unnecessary features, esp. in order to increase salability, acceptance, etc. (often fol. by up): to gimmick up a sports car with chrome and racing stripes.
–verb (used without object)
6. to resort to gimmickry, esp. habitually.

By the first definition, sensor cleaning systems are gimmicks, they're ingenious and they're designed to attract attention.

Nikon and Canon tend to focus their attention gathering ingenuity on competing with each other, while Pentax, Oly, Fuji, etc. try to carve a niche for themselves.

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[DavidMaven]

... the makers of in-camera stab. systems are , combined, outselling both Nikon and Canon... NOT!

--
[email protected]

 

[Joseph S Wisniewski]

Actually, sensor stabilization is more effective - google it and
you will find there is indeed article on this, with respect to
current technology.

I find articles that say both ways. But none that come anywhere close to how I tested it.

In summary and if you study the physics of it, displacement
increases as you go further away from the source of vibration or
disturbance (hand) and physics also tells you that the response lag
will also increase.

Oh, indeed I have studied the physics of it. If you do the same, you'll realize that a lens based system changes the angle of light, and therefore very small displacements have great effect.

A lens based system aims the gun to hit the target, a sensor based system moves the target around to try and catch the bullets.

What is being said is, in the past, not many electromechanical
systems are able to stabilize a photoreceptor media (in this case
film + film holder) fast enough due to its weight, etc. without the
compromises of size and power consumption and accuracy, not too
mention the speeds of the response required.

Yup.

It's best to tacke this further away (i.e. at the lens) where there
is indeed a response lag and a tilting floating lens mechanism
which can simply resolve this.

A response lag between the lens and the image?

With the introduction of light fixed photo sensors (CCD/ CMOS),
tiny piezo electro motors works efficiently to curb (counter
effect) vibrations very close to the source. It's less expensive
too.

That part is true, it is less expensive.

That said, it's all about business sense, as these two large camera
manufacturers still have a large pool of consumers out there with
SLRs (digital and film) without in-body stabilization.

I did point that out.

It's also
more difficult to change if you already have a whole range of SLRs
behind you without IS.

Not really. They don't have to "change", they just have to "add" to the existing systems. And since either Nikon or Canon, all by themselves, has a bigger SLR R&D budget than all the competitors put together, it really just is a question of what do they want to do.

And not to mention a factory line that makes
lens with IS and the right to the patents for a fee.

You're right, you shouldn't mention that.

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[Joseph S Wisniewski]

Over time, there has been many reports on this and most seem to
favor the lens based system although the differences are not large.

Again actually no.
What there HAS been is lots of posts where people say exactly what
you do, but how many proper actual comparison tests?

As far as I know, only one. Mine. I offer it for a quite reasonable fee to camera makers, large distributors, advertising agencies, and lawyers involved in advertising or product performance claims.

While it may very well be proveable that one is better than the
other. I believe it comes down to the individual (everyone is
different). I also believe (unsubstantiated opinion and all) that
Canon only keeps it in lens to maintain existing value of and for
their heavily spending customers and Nikon follows on this
issue....if Canon changed Nikon would also.

Can your theory explain why Sony charges $600 more for a 70-200mm f2.8 without stabilization than Nikon and Canon charge for the same lens with stabilization. And it's not optical quality, the Sony is not the leader of that pack.

I do not believe I could get any more stops with in lens than I can
get with in body but since I do not have any in lense IS it is
really not an issue.

In body at least appears to be value for money in any case.

I think lens systems are an overall better value, but body systems have the added ability to stabilize lenses that would be very difficult with lens based systems, like an 85mm f1.4 or a 105mm or 135mm f2.0.

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[Joseph S Wisniewski]

A stabilized prime lens often end up with same number of glass
elements as a zoom lens due to the need of movable and aberation
correcting elements.

It seems that IS lenses are also larger. I am still somewhat
perplexed at the huge size difference between the Canon 17-55mm
f2.8 IS and Tamron 17-50mm f2.8.

The Canon is weather sealed, and is internal focusing and zooming. The Tamron, while a fine lens, extends while zooming, is not sealed, and uses a lot lighter materials than the Canon.

The Tamron is a 17-50 while the Canon is 17-55mm. Now, 10% longer doesn't seem like much, until you consider that the weight of a lens is proportional to the third power of focal length, so you're dealing with 30% more weight just for that extra 5mm.

Consider this pair of sealed, pro quality 17-55mm f2.8 internal zoom lenses...

Nikon 754g 112mm long, 86mm diameter, no stabilizer
Canon 645g 110mm long, 84mm diameter, stabilized

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[Joseph S Wisniewski]

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

 

[Andrew dB]

It's best to tacke this further away (i.e. at the lens) where there
is indeed a response lag and a tilting floating lens mechanism
which can simply resolve this.

A response lag between the lens and the image?

It might only be a few hundred picoseconds but a lot can happen in that time (on the subatomic scale). ;-)

 

[Henry Richardson]

The Canon is weather sealed, and is internal focusing and zooming.
The Tamron, while a fine lens, extends while zooming, is not
sealed, and uses a lot lighter materials than the Canon.

I am talking about the Canon 17-55mm f2.8 IS lens. I'm not sure which Canon lens you are talking about. The 17-55mm extends when you zoom just like the Tamron and the Tamron has internal focusing, I believe. You can even see photos of it here:

http://www.photozone.de/8Reviews/lenses/canon_1755_28/index.htm

The Canon is also not weather sealed and, in fact, there are lots of complaints of dust being sucked into it.

The Tamron is a 17-50 while the Canon is 17-55mm. Now, 10% longer
doesn't seem like much, until you consider that the weight of a
lens is proportional to the third power of focal length, so you're
dealing with 30% more weight just for that extra 5mm.

Yes, the Canon is marked at 17-55mm vs. the Tamron at 17-50mm. The Canon is widely reported to be not as wide at 17mm as the Tamron though. The Canon is probably more like 18mm.

Consider this pair of sealed, pro quality 17-55mm f2.8 internal
zoom lenses...

Nikon 754g 112mm long, 86mm diameter, no stabilizer
Canon 645g 110mm long, 84mm diameter, stabilized

The Canon is not sealed, but for the price one would expect it to be a sealed L lens.

Of course, with the USM in the Canon I would prefer it, but I was just commenting on the big difference in size of these two otherwise comparable lenses.

--
Henry Richardson
http://www.bakubo.com

 

[Mike Fried]

1. Provide solid scientifically derived proof that lens based IS is
better than in-camera IS.

It is better than in-camera IS because you can see the effect of IS in the viewfinder when you're framing. That's not scientific evidence that the results you get are better, but it is a feature of in-lens IS that differentiates its value from in-body IS. This might not be all that critical at 35mm to 135mm equivalent FoV, but on a 600mm lens on a 1.6x body (960mm FoV equivalent), the camera shake seriously impacts framing, and IS makes composing images significantly better.

Also, no implementation of IS in body currently works on "full-frame" sensor sizes (which is to say that no manufacturer today makes such a system not that it isn't possible). Larger sensors have additional benefits including the ability to have narrower DoF and more surface area per pixel (at same resolution). The larger surface area leads to more sensitivity, which somewhat reduces the need for IS.

IS in a lens can be transferred from camera to camera. IS in a camera works with all lenses. Most of Canon's and Nikon's newest lenses have IS systems built-in. I was using the 600mm f/4L IS two weekends ago, and I'm quite glad that the IS unit was built into that lens, just like I'm quite glad that the AF unit was built into the lens. The IS worked very, very well and it was designed for the focal length actually used by the lens.

2. DSLR users tend to keep lenses and change bodies rather than the
other way around. So even if Canikon upgraded the in-lens IS, the
user probably wouldn't.

Really? Have you done or are you referring to a study that actually suggests this, or is this your "gut feeling"? What about SLR users and DSLR users? IS has been around longer than DSLRs, and it works on SLRs. Before you ask for "scientific proof", maybe you should back up your claims.

It's really hard to scientifically say if one technology in general (with several vendors implementing variations) is better than another technology (with several other vendors implementing variations). Phil and co. review these cameras and estimate how many acceptably sharp shots they get in their reviews, but it's an imperfect review. The most you can get out of their reviews is a rough feeling for how well the technologies implemented in the specific reviewed products work. To what extent the technology works for any given user depends on their technique/how well they hold a camera.

My keeper rate is much higher for my IS lenses, particularly in low light at telephoto focal lengths. For wider angles (15-30mm x1.6), softness from camera shake at slower speeds just isn't an issue I have a problem with.

-Mike
http://demosaic.blogspot.com

 

[Henry Richardson]

2. DSLR users tend to keep lenses and change bodies rather than the
other way around. So even if Canikon upgraded the in-lens IS, the
user probably wouldn't.

Really? Have you done or are you referring to a study that actually
suggests this, or is this your "gut feeling"?

I'm wondering how many people that bought a Canon 28-135mm IS ten or more years ago have upgraded to the several iterations of newer versions of the Canon 28-135mm IS with later generations of IS? Probably none since Canon has not updated it. In fact, how many Canon IS lenses have ever had their IS upgraded to later technology? The only lens that I know of is the 75-300mm IS finally got replaced with the 70-300mm IS recently. It took, what, about 15 years?

--
Henry Richardson
http://www.bakubo.com

 

[becksi]

I believe that in-camera stabilization is NOT as effective as lens
stabilization. It's as simple as that.

I think that is a highly relative statement and entirely depends on how you define effictiveness or what is more of value to you. If you look at the individual lens you may be right about IS beinf more powerful in-lens compared to in-camera. However, in camera has the major advantage that all of your lenses benefit from IS.

And as stabilization technology improves, you can always upgrade to
it via the lens. If your body had stabilization version 1.0, you're
stuck with it as long as you have your SLR body, even if version
2.0 is on the market

I want to disagree. The body of DSLRs are rather the disposable part than the lenses. Lenses retain their value much better and are usually the parts that you keep, vs. the body becomes obsolete more quickly and is replaced more frequently than the a lens (in general). So you can take advantage of improvements of in-body IS more quickly and widely than in-lens.

From a cost stand point... at the price difference of some IS lenses vs. their non-IS counterparts and body upgrade is almost as easily justied.

 

[dkloi]

Doesn't matter where you put an angular rate sensor on a rigid
body. You don't think they use linear accelerometers?

Yes, actually, I think they do.

In pairs. Take the difference and it's a lot better at high
frequencies than a rotational accelerometer.

In other words, you want to extract an angular acceleration signal from two differenced displaced linear accelerometers. Either way it's the same thing, the angular signal will be the same wherever you are on the rigid body. Hence it still doesn't matter whether you put the sensors in the lens or the body.

MEMS angular rate sensors have excellent performance, typical performance of 300 degrees/s, 0.1 degrees/s/root Hz, 0.1% non-linearity, kHz bandwith. Most importantly, this performance is cheap.

Taking the difference of two linear accelerometers which are displaced by millimetres will not give you a very clean signal. Taking the difference of two similar quantities usually doesn't give you a very good signal to noise ratio, especially if the individual signals are subject to drift. And the frequencies of physiological tremor aren't all that high, maximum of tens of hertz, typically in the teens.

I don't really think that coupled linear accelerometers would be used in this application. They'd measure angular acceleration, not the ideal signal since the correction signal (displacement of decentering group or image sensor) is the angular displacement which is the acceleration signal twice integrated, rather than just a single integral of angular rate leading to even greater low frequency noise sensitivity. Since the frequencies of interest are quite low, drift of each of the paired linear accelerometers would be disasterous.

You'll have to go back to the drawing board with this idea.

Cheers,
Daniel.

 

[whvick]

"I expect that both Nikon and Canon will implement some sort of body based IS in their entry level DSLRs when and if they feel the competitive heat from other brands."

Which may be soon as I am seeing more and more POTDs won by Pentax K10Ds and in the past it was mostly Nikons and Canons.
No prejudice here ... I have a canon D60.
whvick
--

 

[Jim Salvas]

Let's say that image stabilization does not exist yet. Imagine a photographic world without it.

Now, imagine that a manufacturer comes to you for your opinion and says they plan to implement an IS system. Further, they say they can do it equally well either way: in-lens or in-body. They want your opinion on which method they should choose.

Which would you tell them you wanted?
--
Jim

 

[neil holmes]

Can your theory explain why Sony charges $600 more for a 70-200mm
f2.8 without stabilization than Nikon and Canon charge for the same
lens with stabilization. And it's not optical quality, the Sony is
not the leader of that pack.

Clearly the cost is not ONLY a IS/non IS issue and other factors come into it.

In body at least appears to be value for money in any case.

I think lens systems are an overall better value, but body systems
have the added ability to stabilize lenses that would be very
difficult with lens based systems, like an 85mm f1.4 or a 105mm or
135mm f2.0.

--
Normally, a signature this small can't open its own jumpgate.

Ciao! Joe

http://www.swissarmyfork.com

I do not understand how an in lens could be regarded as better value.

For the price of IS in one lens alone you can buy a dslr (AND lenses) that will stabilise ALL your lenses. I am very happy with my (Canon speak) 50 1.2 IS, 135 1.8 IS, Tamron 17-35 2.8-4 IS and now Tamron 300 2.8 IS They and all my other lenses are stabilised and It cost me nothing extra for that and I would not be able to hand hold them at any slower speeds than I would regularly use if they where in lense stabilised. My shorter lenses are hand holdable doen to 1/2 a second and tele to around 1/10, both slower than I would use in practice.

If its better value for you to pay each time you buy a lens you are welcome to it.

neil