Why we use in-LENS stabilization.

[Lightstream]

From Canon's latest white paper....

Some of Canon’s competitors have chosen to use in-body image stabilization. The technique involves moving the image sensor in a controlled fashion, based on signals from movement detecting sensors in the camera body. The obvious advantage of this system is that users have some sort of stabilization available with almost any lens they connect
to the body. Short focal length lenses require smaller sensor deflections; 24 or 28 mm lenses might need only 1 mm or so. Longer lenses necessitate much greater movement; 300 mm lenses would have to move the sensor about 5.5 mm (nearly 1/4”) to achieve the correction Canon gets with its IS system at the same focal length. This degree of
sensor movement is beyond the range of current technology. Short and “normal” focal length lenses need stabilization much less often than long lenses, so the lenses that need the most help get the least. Further, in cameras with smaller than full-frame, 35 mm film size sensors, equivalent focal lengths become longer, by a factor of 1.5 or 1.6, exacerbating the problem by making all lenses longer.

That explains a lot. Ever felt the IS gyros kick in on a bigger lens? They make more noise and they 'feel' a lot heavier than those on the shorter lenses. My 24-105 is so soft I can hardly even feel it, my 70-300 and 300 f/4L IS are a different story. (I do value the feedback though)


Interesting to note is that they won't rule it out.

At some point, in-body stabilization may improve to the point at which such technology
may be appropriate for certain segments of Canon’s DSLR range. It would be senseless
to rule out such a possibility. Even now, differences in unit cost are not enough to be
significant factors in such a decision. The bottom line is performance.

That does make sense, but it also says that the high end will remain optically stabilized. What would be neat is a lower-end body with integral IS and EF-compatible lens mount, that automatically turns off its body-based IS and gives preference to optical IS when a suitable lens is mounted. That way we get the best of all worlds.

[becknell]

Right, Canon chooses quality rather than convenience. That's just a little reason I like Canon.

[stupot]

what they didnt mention is that with all the jumping about of the sensor it may end up 'jumping' out of the image circle... then you're gonna get some nice vignetting...

[Coco-Puffs]

24 or 28 mm lenses might need only 1 mm or so. Longer lenses necessitate much greater movement; 300 mm lenses would have to move the sensor about 5.5 mm (nearly 1/4”) to achieve the correction Canon gets with its IS system at the same focal length. This degree of
sensor movement is beyond the range of current technology.

so its impossible but the competitors are doing it.

and its impossible only b/c they stick to the IS technology.
what if they use a new technology?

[Lani Kai]

so its impossible but the competitors are doing it.

The competitors aren't doing it. That's why with a telephoto lens the sensor-shift method is less effective. Since it's physically impossible to move the sensor 5.5mm, they have to make do with whatever they can manage--1 or 2 mm--meaning they can't fully correct for the blur when used in conjunction with a telephoto.

[dod]

That doesn't make sense to me. If you take an APS-c sensor what they're saying is that for a longer lens you need to shift the sensor about 25% of it's width to correct shake. If you look at the FOV of a 400mm focussed at infinity it's about 6 or 7 degrees. Shifting the sensor 5.5 mm wouldn't be correcting shake, it would be correcting someone swinging the lens about. (exagerated to make the point)

I must be missing something :(

[ron chappel]

The competitors aren't doing it. ... Since it's physically impossible to move the sensor 5.5mm, ...

It's nice to see canon talking about in body stabilization but it would be better if they had been more honest:evil:
The facts are that other manufactures ARE achieving great performance .The new Sony body claims 3.5 stops advantage -about the same claimed by canon's *best* IS versions.I'm not sure how far the sony moves it's sensor but the older minolta version moved up to 10mm.
Personally i'd like to see properly controlled tests rather than claims by the respective marketing departments.....


what they didnt mention is that with all the jumping about of the sensor it may end up 'jumping' out of the image circle... then you're gonna get some nice vignetting...

:lol: :lol: :lol:
This won't happen.Because only reduced frame cameras have in-body stabilization ,and it's only used with full frame lenses there is basically no risk of 'running off the edge '

[ron chappel]

In a way it's good to know canon are offering the best performance option (if not by alot)
but they SHOULD offer at least one body with IS.
Let's face it-in spite of all the claims it's obvious that in-body IS can do (at least) a reasonable job with long lenses.Added to that is the fact that all of us *will* use IS even with shorter focal lengths.
I for one am getting very annoyed with canon for stalling and making excuses on this :evil:

[stupot]

:lol: :lol: :lol:
This won't happen.Because only reduced frame cameras have in-body stabilization ,and it's only used with full frame lenses there is basically no risk of 'running off the edge '

yeah that makes sense:) but if canon were to introduce it on their crop bodies then it would surely be a problem with ef-s lenses (and dc lenses from sigma? i dont know what all the symbols mean! but im sure they do some reduced circle lenses...).

[SkipD]

That doesn't make sense to me. If you take an APS-c sensor what they're saying is that for a longer lens you need to shift the sensor about 25% of it's width to correct shake. If you look at the FOV of a 400mm focussed at infinity it's about 6 or 7 degrees. Shifting the sensor 5.5 mm wouldn't be correcting shake, it would be correcting someone swinging the lens about. (exagerated to make the point)

I must be missing something :(What you are missing is that Canon designed their IS so that it will work on ALL of their SLR cameras - APS-C DSLR's, 35mm film cameras, "Full-Frame" DSLR's, etc.

[Banbert]

It's nice to see canon talking about in body stabilization but it would be better if they had been more honest:evil:
The facts are that other manufactures ARE achieving great performance .The new Sony body claims 3.5 stops advantage -about the same claimed by canon's *best* IS versions.I'm not sure how far the sony moves it's sensor but the older minolta version moved up to 10mm.
Personally i'd like to see properly controlled tests rather than claims by the respective marketing departments.....
'

But are Sony saying that they achieve that across their whole range of lens, as that would seem to be the key point, achieving 3.5 stops on 17-55mm sounds a lot easier than on a 400mm.

Not really read any of the sony literature so sorry if thats obvious from the literature they have put out.

[dod]

What you are missing is that Canon designed their IS so that it will work on ALL of their SLR cameras - APS-C DSLR's, 35mm film cameras, "Full-Frame" DSLR's, etc.
That's not what I'm getting at. Say at a shooting distance of 50m a 400mm has a FOV of about 5.2m. A shift of 5mm at the sensor would equate to a total swing at the subject of about 2 metres laterally as far as I can work out. Would that be right? Seems a bit much, might be my calcs though :confused:

[Lani Kai]

So by that logic, with an AOV of 4 degrees (just guessing here) a lateral shift of 2m in a FOV of 5.2m would be a lens movement of roughly 1.5 degrees, no?
I think I may be doing something wrong here.

[red hot sheep]

yeah that makes sense:) but if canon were to introduce it on their crop bodies then it would surely be a problem with ef-s lenses (and dc lenses from sigma? i dont know what all the symbols mean! but im sure they do some reduced circle lenses...).

That's what I don't understand. So Sony cannot make reduced circle lenses even though they have no full frame camera?

[jondru]

Could you post a link to that white paper? I poked around the Canon site without success. I wonder if they have other interesting white papers?

thanks
vv
J.O.

[Dan Roeder]

Is it possible to make an IS unit that is separate from the lens but goes between the lens and the body? That way, you could buy an IS unit and attach all your lenses to it. Seems like this would be a good idea.

[Lightstream]

Could you post a link to that white paper? I poked around the Canon site without success. I wonder if they have other interesting white papers?

thanks
vv
J.O.

Bottom of the page. The page itself is a summary that introduces the camera. Enjoy.... :)

http://www.robgalbraith.com/bins/multi_page.asp?cid=7-7897-8492

[lostdoggy]

Maybe its because in- body IS won't work on film and Canon produced IS lenses before the DSLR explosion!!!

[ron chappel]

That's what I don't understand. So Sony cannot make reduced circle lenses even though they have no full frame camera?

In general this could indeed be a problem with in body stabilization.The reduced format lenses introduced with the minolta 7D/5D bodies obviously had enough coverage built in.
In practice most wide angles would probably have enough coverage anyway (due to the small movements required) but reduced coverage tele lenses such as those made by sigma could well give problems.

[Incredirebelz]

That's not what I'm getting at. Say at a shooting distance of 50m a 400mm has a FOV of about 5.2m. A shift of 5mm at the sensor would equate to a total swing at the subject of about 2 metres laterally as far as I can work out. Would that be right? Seems a bit much, might be my calcs though :confused:

mmm ya, 2m shift in an FOV of 5m sounds a bit extreme... altho i would not know as i never tried peeping at a 400mm focal length. I just did a quick estimate with my 70-200. I notice my actual image shift at 200mm focal length (subject ~70meter away) is around 20-30cm at max. ( and I have had about 1 L of coffee at work since 7am this morning). Maybe when they say a 5mm shift at the sensor refers to a 4stop compensation at a focal length of say, very long?

I guess exaggeration in promotional materail is normal... They're prolly right in that the in lens stabilization is more effective at long focal length... but for shooters with good bracing experiences, I am guessing that the in sensor compensation might already be quite sufficient...

[Incredirebelz]

I would also think that it would thebest if Canon would have both in camera IS and lens IS... the user can use either at their discretion... but then again, if the differences in benefit between the two aint really that noticeable, it's gonna hurt the sales on the IS-lenses...

[foty89]

Maybe its because in- body IS won't work on film and Canon produced IS lenses before the DSLR explosion!!!

I don't know if that is it exactly, but the first Canon IS lenses did pretty much predate digital stuff. Also, if I have read it correctly, nowhere does it say they could not do it in camera. It says that they can do it better on a lens to lens basis, as each system is specfic to each lens, and therefore gives better performance than a one size fits all approach. That line of reasoning makes a lot of sense to me. And I REALLY do wonder how effective in camera systems are across the range of lenses. While it may be more expensive, I prefer the Canon approach. I'd rather have to pay more AND get more, than pay less AND get less.

[Anders Östberg]

Canon will have in-body stabilization eventually, it's what the (at least consumer/prosumer) market will want.

[foty89]

Canon will have in-body stabilization eventually, it's what the (at least consumer/prosumer) market will want.

I think you are right no there, after all, Canon just added the anti dust system to combat competitors. Hopefully, they will have it set up to switch off when a lens with built in IS is used.

[ron chappel]

Canon will have in-body stabilization eventually, it's what the (at least consumer/prosumer) market will want.

I really hope so.
I had hoped they might introduce it as early as right now with the new 400D. Unfortunately not:cry:

[DavidEB]

I don't understand this at all. why would the amount of image plane movement change as a function of lens focal length? To negate vibration, the image plane moves just as much as the vibration itself. If the vibration comes from your caffeine tremor, it'll be the same at 17mm as at 200. So it shouldn't change (much) as FL changes.

maybe you could argue that longer lenses weigh more and so you shake them more, but maybe you could argue that longer lenses being heavier dampen vibrations better than lighter lenses, so they shake less. but I think those effects are likely to be small.

[cc10d]

Longer lenses magnify more than short ones. For any given movement of the camera, the longer lens will shift the image seen by the sensor, we see this in the viewfinder also. When not using IS we up the shutter speed to expose a shorter time slice of this movement, thus "stopping camera movement". That larger excursion of the IS is not to hold the camera still, it is to compensate for the image movement seen by the sensor.

[jedwards]

David,
Remember the stabilization works closest to the sensor itself. Imagine how a change in axis of the camera of 2 degrees would affect a 17mm shot vs. a 400mm shot. at 400mm, the entire frame would look different, while at 17mm it may not be noticeable. an IS system would need to move much more to compensate for the same axis shift at 400mm.

I think canon takes the correct approach with this, the IS system can be tuned to the lens optics.

[ron chappel]

I don't understand this at all. why would the amount of image plane movement change as a function of lens focal length? ....

Look through your camera viewfinder with a wide angle lens attached. Swing from side to side
Now do the same with a tele lens attached.Quite a difference huh!:)
So that's why they have to move the sensor more with longer lenses -simply because the image moves more

[gef]

I'd like to see IS on body and in lens. For the lenses without IS, the body can provide the IS. Then on the wider lense like my 17-70, the body can provide the IS. I would still consider IS on the telephoto lens so long as it was better then what the body could provide. I figure Canon will provide IS on the body someday, and if that happens I may have to upgrade my body. The dust removal is a nice feature I'd like to have, but how effective is it at removing dust?

[dandan]

The dust removal is a nice feature I'd like to have, but how affective is it at removing dust?

according to luminous-landscape its pretty durned effective

[Anders Östberg]

according to luminous-landscape its pretty durned effective
The camera is brand new, I can't see how LL can make a judgement already.

[Jon]

That's not what I'm getting at. Say at a shooting distance of 50m a 400mm has a FOV of about 5.2m. A shift of 5mm at the sensor would equate to a total swing at the subject of about 2 metres laterally as far as I can work out. Would that be right? Seems a bit much, might be my calcs though :confused:Using actual numbers, a 400 mm has a (diagonal) FoV of 57 m at 1000 m on an APS-C body. AoV is 3.2 deg. Assuming Canon's stated 5.5 mm shift applies to this lens (not to, say, the 600), that would correspond to the lens slipping 0.8 deg. off target, or around 14 m. Bringing that down to your 50 m. example, that would amount to about 0.7 m "subject shift". That's less than 1 deg. wobble for a pretty big, pretty heavy lens.

[Jon]

I don't understand this at all. why would the amount of image plane movement change as a function of lens focal length? To negate vibration, the image plane moves just as much as the vibration itself. If the vibration comes from your caffeine tremor, it'll be the same at 17mm as at 200. So it shouldn't change (much) as FL changes.

maybe you could argue that longer lenses weigh more and so you shake them more, but maybe you could argue that longer lenses being heavier dampen vibrations better than lighter lenses, so they shake less. but I think those effects are likely to be small.For an example of why the long lens will need more compensation, try holding an OED close to you, and then at arms' length. Which way does it shake more? Or try holding a short ruler steadily in front of you, then a meter stick. That little bit of motion close in is amplified by the length. It's the old physics calculation of a lever and a fulcrum. The greater the difference in distances betweeen the effort and the fulcrum and the fulcrum and the weight the less effort (and greater displacement) you need to lift the weight.

[Lightstream]

The camera is brand new, I can't see how LL can make a judgement already.

A couple of months back, in the middle of the night, some Canon guy packs an innoculous brown box off to LL together with embargo and NDA documents.. :mrgreen:

They probably got the prerelease units, DPReview usually does as well but they didn't this time and they are pretty ticked about it. So we know Canon does send some out to the Chosen Few.

[Jeff1]

Why can’t Canon put an Image Stabilizer in a DSLR like Pentax K100D or a P&S like S2IS? Instead of paying hundreds more to have it in your lens? Do they have a good reason or are they only concerned with the bottom line. I am considering getting the K100D over the 400D for this reason, not to mention my budget. Any thoughts on this subject??

[Balliolman]

Well, the competition are putting the squeeze on Canon so give it time, Jeff, give it time ...

[hemuni]

Here's another vote for body IS - maybe the next rebel?

[HughScot]

The new Sony DSLR body has IS but I read on one of these forums that it did not work as well as IS in the lens. I have no idea what the technical reason may be so if someone else does please enlighten us.

[catsith]

i had a play with the alpha for the same reason, and i have to say, it is a lovely camera. I too can't afford the prices canon want for their IS lenses, so i will probably end up buying the alpha. While it may not work as well as the dedicated lenses, at least it is better than no IS to me.

[Nick_C]

That's really what im waiting for before I upgrade again, at the moment my 350D is perfect for any situation.

What would force me to upgrade is a 22Mp body with IS, I might have to wait a while for that though ;-) it will happen though, im sure of it, probably sooner than later too at the current rate of technology.

Nick :-)

[Nick_C]

Canon should put IS into every body so any lens benefits from a little IS assistance, but continue with IS lenses for those demanding the best! at least that way we get a choice.

Nick :-)

[hemuni]

Canon should put IS into every body so any lens benefits from a little IS assistance, but continue with IS lenses for those demanding the best! at least that way we get a choice.

Nick :-)
Exactly! - people seem to think that one thing rules out the other. Just have the camera turn of body IS whenever a non compatible lens is attached.

Its NOT either or!

You could even have a crop option for large lenses, FF and EF-s compatibility.

Body IS might or might not work on long lenses, so what? it will still help the shorter ones.

Another thing people seems to forget is that lens IS ads an extra lens element and this has been reported to degrade IQ (compared to their non IS siblings) - specifically in 300/4.0 IS and 70-200/2.8 IS.

Canon should spend their time on making the best of this feature, instead of trying to explain why they are falling behind.

I love the idea of having the IS in a separate lens unit, like an extender (posted earlier in this thread by Dan Roder). Again it might not be as good as lens IS, but it will be a h*ll of a lot better than none.

Common canon - lets see those stabilized body's!

[BryanP]

Canon should put IS into every body so any lens benefits from a little IS assistance, but continue with IS lenses for those demanding the best! at least that way we get a choice.

Nick :-)

Having IS in body sounds like a great thing, but I would never rush it into a body if it can't even perform effectively.

I'm not saying other manufacturers' cameras are not effective in using in-body IS though.

It's good to have the option, but I don't think anyone would disagree that IS is MOST useful for long focal lengths. It's not too much of an issue shooting at short focal lengths when it comes to motion blur as it is holding lets say a 300mm lens. I find it pretty pointless having an IS feature and not even being able to use it in long focal lengths when it matters the most.

[hemuni]

...I find it pretty pointless having an IS feature and not even being able to use it in long focal lengths when it matters the most.
There's a difference between not being able to use it and the effect being weaker on a long lens, maybe it only gains you one stop of light, but that's still one stop.

According to dpreview, the alpha gains about 2 stops in a non scientific test. I don't know wich lenses where used, but that sounds pretty respectable to me. Especially considering this is a new system.

I think the point here is not which system is the best, but that adding body IS would mean instant IS for the entire backcatalog of EF lenses.

In the end this discussion is trivial. I'm sure there where quite a few opponents to the original lens based IS back in the days - not to mention AF (who will ever need that!?).

Camera producers will produce what sells and if Sony gets a hit with the alpha, Canon will eventually have to answer.

[BryanP]

There's a difference between not being able to use it and the effect being weaker on a long lens, maybe it only gains you one stop of light, but that's still one stop.

According to dpreview, the alpha gains about 2 stops in a non scientific test. I don't know wich lenses where used, but that sounds pretty respectable to me. Especially considering this is a new system.

I think the point here is not which system is the best, but that adding body IS would mean instant IS for the entire backcatalog of EF lenses.

In the end this discussion is trivial. I'm sure there where quite a few opponents to the original lens based IS back in the days - not to mention AF (who will ever need that!?).

Camera producers will produce what sells and if Sony gets a hit with the alpha, Canon will eventually have to answer.

It definitely is trivial. I guess if we're talking about magnitudes like that (of IS effectiveness), then it is fine. I was assuming a much less effective stabilization.

[Jon]

Well, the August Photography Monthly had an Alpha review that showed what looked like barely 1 stop, or less, with a moderately long lens. Most people could get that by just improving their technique.