IS lens on an IS body?

[vic6string]

I just had a thought while replying to a post on the new Sigma 18-200 OS lens: What do you think would happen if you used this lens on an image stabilized body like the Sony Alpha or Olympus e510? The lens would send a much more stabilized picture to the sensor that would then stabilize it some more? Would the two systems work together to make it that much better, would they get in each others way, or would it do nothing at all? I have mentioned in a few posts that Canon and Nikon are going to have to put a cheap IS kit lens in their kits, and drop IS prices in general due to marketing concerns with all the Image stabilized cameras coming out now, but if an IS lens could work together with an IS body Canon and Nikon might just go that route. Even if they don't work in unison, but they don't get in each others' way, this could be a viable strategy in the marketing war against all these new cameras. It would be a huge boon if Canon were to figure out a way where in-camera IS could work in unison with only THEIR EFand EFS IS lenses. What do you all think?

[bieber]

In-camera IS is a gimmick, and not one Canon and Nikon are going to fall for. The Pentaxes and the Sonys need stupid little tricks like that to try and convince people their bodies are worth buying; Canon and Nikon could care less, they just have each other to compete with.

[vic6string]

Gimmicks sell newbies! And as far as gimmicks go, this one is useful (as good as the Canon and Nikon answer?: no, but a stop or two of IS is indeed useful) and it is a proven seller as Sony and many others have been using it to sell digicams and camcorders for years. In the eyes of someone who has never even considered a DSLR, seeing the "supersteadyshot" emblem is something he can recognize and appreciate. When you tell him the Sony has it, but the Nikon and Canons charge you a ton of extra money on every lens to get it, what do you think his choice is going to be? Like it or not, that is the consumer most coveted right now in the DSLR market, and Nikon and Canon will both have to come up with an answer to this soon if they plan to keep their huge market share. That said, I didn't mean for this to be an in-body vs in-lens IS thread, I really am curious about my original question.

[ed rader]

Gimmicks sell newbies! And as far as gimmicks go, this one is useful (as good as the Canon and Nikon answer?: no, but a stop or two of IS is indeed useful) and it is a proven seller as Sony and many others have been using it to sell digicams and camcorders for years. In the eyes of someone who has never even considered a DSLR, seeing the "supersteadyshot" emblem is something he can recognize and appreciate. When you tell him the Sony has it, but the Nikon and Canons charge you a ton of extra money on every lens to get it, what do you think his choice is going to be? Like it or not, that is the consumer most coveted right now in the DSLR market, and Nikon and Canon will both have to come up with an answer to this soon if they plan to keep their huge market share. That said, I didn't mean for this to be an in-body vs in-lens IS thread, I really am curious about my original question.

then why not ask your question on the proper forum. the canon DSLRs have no in-body IS .. as you already know.

ed rader

[vic6string]

then why not ask your question on the proper forum. the canon DSLRs have no in-body IS .. as you already know.

ed rader

Man Ed, this sounds like a Nikonian answer. I ask in this forum because 1) I am thinking about the possibility of Canon throwing some basic form of IS in body (more for marketing reasons than anything else) and 2) i find this forum to have some of the best minds out there when it comes to this stuff, and if anybody can give an intelligent, thought provoking answer to this question, chances are they will be a forum member here. I love how defensive some people get when you even think about mentioning in-camera IS. It would be incredibly interesting to see what some of these people would say if Canon does decide to go this route. Would it all of a sudden make in-camera IS cool?

[Jon]

If Canon were to offer in-body IS, they'd need to disable it when an IS lens was mounted. The IS mechanism's totally self-contained within the lens - it doesn't rely on anything but power from, nor does it send anything to, the body. So starting the body trying to smooth out shake while the lens is trying to smooth out shake will have them working at cross-purposes. The big problem with in-body IS is that it loses effectiveness with longer focal length lenses, where you need it most.

[aoleg]

They already tested the two anti-shake systems on Olympus. Here's the actual test: http://www.cameralabs.com/reviews/OlympusE510/Olympus_E510_with_Leica_14-50mm.shtml

Their conclusion, although preliminary, was that the in-body anti-shake was more effective than the one in the lens. Interestingly enough, if both systems are enabled, the camera uses the in-lens one during the focusing, and then disables it and uses the in-body one for the shot.

[asylumxl]

anyone who thinks in-body IS is a gimmick is short of a few braincells. Ofcourse it works, and it's effectiveness differs from make to make. Canon don't make inbuilt IS as it would mean they wouldn't be able to sell their IS lenses to people with more money than sense.

[JeffreyG]

My understanding is that in-body is fine for shorter lenses and becomes less effective as the f.l. increases. This is because the magnitude of the shake is magnified by the longer lens...to the point where the sensor cannot move far enough to correct.

I think a line of image stabilized cameras plus IS lenses would be great, but I'm not holding my breath.

It would be cool if normal to short telephoto primes (50mm, 85mm) could all rely on the in body IS. Then you would only want / need to pay for lens IS in the long telephoto lenses. These would presumably turn off the body IS when they were used. That would be a slick system.

Their conclusion, although preliminary, was that the in-body anti-shake was more effective than the one in the lens. Interestingly enough, if both systems are enabled, the camera uses the in-lens one during the focusing, and then disables it and uses the in-body one for the shot.

The tested lens was not a long telephoto. I'm most interested in IS technologies with the longer lenses (I'l take it on shorter, but it's not as critical). One other problem for in-body IS with long telephoto lenses is the viewfinder is not stabilized. I find that to be a big help at 300mm.

[SpaceGuy1]

In-camera IS is a gimmick, and not one Canon and Nikon are going to fall for. The Pentaxes and the Sonys need stupid little tricks like that to try and convince people their bodies are worth buying; Canon and Nikon could care less, they just have each other to compete with.

I'm guessing you have never tried a sensor based stabilizer before? It is not a gimmick, it works. Learn to educate yourself so you won't have to say something to stupid.

[vic6string]

I never even knew Leica made an image stabilized lens. Can't judge body based IS vs lens based IS by that article as they only took one picture each way. He might have just had a steadier hand for that one shot. Hell, I have some shots I have taken at 1/15 and 55 mm with the kit that are perfect, and some shots at 1/60 with motion blur. As for the idea of all normal to short lenses using in-body IS and going to in-lens for the telephoto: it seems like a decent idea until you think about the other reason in-lens IS is better, that being the stabilized view in the viewfinder. No question in-lens is better, especially for longer lenses, but in bady has its merits on the shorter side. I still wonder if there were some way you could design both to work together. If anyone could, it would be Canon as they make great sensor shaking IS in their digicams and they make the best in-lens solutions.

[pepsi]

Double IS?!?! Would that mean I could handhold down to 8 seconds? :-D

[pakomo]

Double IS?!?! Would that mean I could handhold down to 8 seconds? :-D

I'm not too sure about that, but remember; the IS doesn't stop the world, so you would only see static objects, while all things that moved a tiny bit would be blurred :)

[Miyagi-san]

IS is for pansies, in lens, body or otherwise (tripods and monopods included)....you too beanbag users! bahhhhh!!!!

[pepsi]

IS is for pansies, in lens, body or otherwise (tripods and monopods included)....you too beanbag users! bahhhhh!!!!

Hey! Leave the "bean baggers" out of this! Hahhahaa!!

[longisland.km]

Why are some people so offended by in-body IS? It seems every once in a while some new technology gets implemented into the camera and people are all up in arms about it...

Once upon a time light meters were added to the camera. I'm sure some people thought that was stupid. Then they added autofocus. Some other people thought that was stupid too. Then they went digital. More people thought that was stupid.

Now the battlefront is IS...in-body or in-lens. The in-lens camp says in-body sucks. Maybe in a relative sense that is true, but its still better than hand holding without it. The in-body camp says it works on every lens. That is definitely true. The nutty thing is that there is absolutely no technical reason why both can not co-exist. The limitation would purely be what Canon marketing would want to limit their customers to (and that would be a great disservice). Why do I think it is a disservice? Well, its obvious that Canon is not going to make every lens IS. From an end user perspective, there are definitely instances where one or two stop IS would be beneficial.

All the camera needs some logic like this:

if (userwants = ISon) then
bodyIS=true
endif
if ((lens = hasIS) and (lensIS = true)) then
bodyIS=false
endif

If the lens has built in IS, it will be used. If it doesn't, then if the user has enabled bodyIS then it will be used. Simple.

For the people who are arguing that they don't want to pay extra for in-body IS...well, if the camera has a sensor cleaning function, the in-body IS will probably only require a change in programming, like how ABS and traction control both use the same slip sensor in different ways.

[Jon]

anyone who thinks in-body IS is a gimmick is short of a few braincells. Ofcourse it works, and it's effectiveness differs from make to make. Canon don't make inbuilt IS as it would mean they wouldn't be able to sell their IS lenses to people with more money than sense.I suggest you tone down your rhetoric until you learn a bit more about lens shake.

There's a reason the 1/f.l. rule applies; the longer the lens, the more linear deflection of the imaging rays from a small angular motion. So at longer focal lengths, the sensor will have to deflect correspondingly further to contain the shake because it's at the end of the path. Lens-based IS puts the moving elements at a nodal point of the lens so it's relatively independent of the focal length. It also allows the manufacturer to design the IS specifically for the characteristics of the lens.

The British magazine Photography, testing the Sony Alpha, found that the IS degraded as the focal length increased so that the 2+ stops improvement the camera was able to provide with a "normal" lens had decreased to a little under 1 stop at 300 mm. They published their test photos too; the decrease in effectiveness was consistent with increasing focal length. Olympus may be able to manage better with theirs, but they're still facing the same basic problem that a 0.5 deg. axial deflection on a 30 mm lens is barely noticeable in the 40 deg angle of view of an APS-C camera. That same 0.5 deg. deflection is much more pronounced on a 4.3 deg. AoV 300 mm lens on the same body.

[AngryCorgi]

In-camera IS is a gimmick, and not one Canon and Nikon are going to fall for. The Pentaxes and the Sonys need stupid little tricks like that to try and convince people their bodies are worth buying; Canon and Nikon could care less, they just have each other to compete with.

Ummm...

http://www.nikonusa.com/images/products/25552_180.jpg
http://www.nikonusa.com/images/products/25555_180.jpg
http://www.nikonusa.com/images/products/25561_180.jpg

...these must not be Nikon cameras then. :rolleyes:

Not saying sensor-shift is the best answer by any means, but Nikon thinks its good enough to sell on *some* of its consumer digicams. It's probably a super-huge pain to make the more complex lens-shift systems work for smaller cameras, thus bringing price higher. It's notable that Canon has not skimped even on the affordable digicams, and only offers optical shifting systems. Nikon is starting to move to 100% optic shifting on VR digicams too, but these remain as a testament that they fell for the "gimmick" as you call it.

[AngryCorgi]

I suggest you tone down your rhetoric until you learn a bit more about lens shake.

There's a reason the 1/f.l. rule applies; the longer the lens, the more linear deflection of the imaging rays from a small angular motion. So at longer focal lengths, the sensor will have to deflect correspondingly further to contain the shake because it's at the end of the path. Lens-based IS puts the moving elements at a nodal point of the lens so it's relatively independent of the focal length. It also allows the manufacturer to design the IS specifically for the characteristics of the lens.

The British magazine Photography, testing the Sony Alpha, found that the IS degraded as the focal length increased so that the 2+ stops improvement the camera was able to provide with a "normal" lens had decreased to a little under 1 stop at 300 mm. They published their test photos too; the decrease in effectiveness was consistent with increasing focal length. Olympus may be able to manage better with theirs, but they're still facing the same basic problem that a 0.5 deg. axial deflection on a 30 mm lens is barely noticeable in the 40 deg angle of view of an APS-C camera. That same 0.5 deg. deflection is much more pronounced on a 4.3 deg. AoV 300 mm lens on the same body.

Maybe it's just me, but the IS on my 70-200 f4L IS is not nearly as effective with a 1.4x and a 2x teleconverter. Still halts slight tremors, but nothing like it does at 70mm. I am sure optical stabilization is more efficient all-around (having played with some lens shift cams), but I am sure it's limitations are to the lens it is built into, allowing additional glass between it and the sensor (TCs) to degrade the performance.

[Jon]

Maybe it's just me, but the IS on my 70-200 f4L IS is not nearly as effective with a 1.4x and a 2x teleconverter. Still halts slight tremors, but nothing like it does at 70mm. I am sure optical stabilization is more efficient all-around (having played with some lens shift cams), but I am sure it's limitations are to the lens it is built into, allowing additional glass between it and the sensor (TCs) to degrade the performance.And if you read my description of how sensor-based IS works and what happened in tests on the Sony, you'll see that the sensor-based IS would not be able to do any better, even ssuming they eventully get it working as well. 0.5 deg. of angular displacement at 200 mm on about 6.4 deg. is 1/7 the sensor width. Put a 2x TC so you've got an AoV of 3.2 deg. and you need to offset an APS-C sensor about 1/3 its width. And if you're using an EF-S lens, you've moved outside the image circle, since that much motion would require an image circle that's 41 mm across. So kiss your UWA lenses good bye and learn about the cos^4 law that has some 5D owners moaning. Just because it sounds like the obvious solution doesn't mean that it will stand up to examination.

[AngryCorgi]

And if you read my description of how sensor-based IS works and what happened in tests on the Sony, you'll see that the sensor-based IS would not be able to do any better, even ssuming they eventully get it working as well. 0.5 deg. of angular displacement at 200 mm on about 6.4 deg. is 1/7 the sensor width. Put a 2x TC so you've got an AoV of 3.2 deg. and you need to offset an APS-C sensor about 1/3 its width. And if you're using an EF-S lens, you've moved outside the image circle, since that much motion would require an image circle that's 41 mm across. So kiss your UWA lenses good bye and learn about the cos^4 law that has some 5D owners moaning. Just because it sounds like the obvious solution doesn't mean that it will stand up to examination.

Did you think I was arguing against anything you stated? Just putting in the (obvious) viewpoint that optical systems have limits as well. Anywho....

[tdaugharty]

IS is for pansies, in lens, body or otherwise (tripods and monopods included)....you too beanbag users! bahhhhh!!!!

Handheld 2.8 IS shot @ 1/15 - Tack sharp .. Got one? Ha! I'll answer for ya. No!

[Miyagi-san]

Handheld 2.8 IS shot @ 1/15 - Tack sharp .. Got one? Ha! I'll answer for ya. No!

Actually yes. (although I insist it must be viewed from a 30 foot distance minimum!! :D )

[longisland.km]

Jon,

I get your technical argument that at longer focal lengths, the in-body IS faces limitations due to the amount of deflection that would be needed in the sensor. Its like you're telling people that in-lens IS is kind of like having a car, while in-body IS is kind of like having a bike. So, if you're travelling, a car is better than a bike. I would agree with that. But, having no IS is kind of like walking. While having a bike is not as good as a car, it sure beats walking!

In any case, where IS is really needed, Canon has in-lens IS available as an option. As in your example, the 70-200 has IS (well, if you paid for the IS version). How about the 85 f/1.8, or the 24-70 f/2.8L? Wouldn't it be nice to have the **option** of having one or two stop IS on that? I've heard all the arguments about IS not being needed on the shorter lenses. I call BS on that. I have tried taking some shots at a friend's wedding at ISO1600 wide open on my Tamron 17-50 f/2.8 (yep, I lust for the 17-55 f/2.8 IS). Those crazy Manhattan churches surrounded by tall buildings let almost no light in. One or two stop IS would have helped the keeper rate in that situation when I was getting alot of 1/15 shutter speeds.

-

And if you read my description of how sensor-based IS works and what happened in tests on the Sony, you'll see that the sensor-based IS would not be able to do any better, even ssuming they eventully get it working as well. 0.5 deg. of angular displacement at 200 mm on about 6.4 deg. is 1/7 the sensor width. Put a 2x TC so you've got an AoV of 3.2 deg. and you need to offset an APS-C sensor about 1/3 its width. And if you're using an EF-S lens, you've moved outside the image circle, since that much motion would require an image circle that's 41 mm across. So kiss your UWA lenses good bye and learn about the cos^4 law that has some 5D owners moaning. Just because it sounds like the obvious solution doesn't mean that it will stand up to examination.

[vic6string]

Longisland is dead on. Yes, we all know (at least those of us who have taken the time to read about it or use it) that in-lens is better, and that becomes more appearant the longer you go, but not everyone needs to go long, or wants to pay the premium. Now, if you are buying a 5D or a 1 series, you probably have the money and the experience to desire in-lens IS, but I would guess that a very high number of XT and XTi users will never buy a lens after their first (the kit or maybe one of the popular kit replacements like the Tamron or Sigma 17-50s) To those people (who are a quickly growing segment of the DSLR market)in-body IS would be much better than no IS, and at that focal range, in-body IS is quite effective. As someone mentioned earlier, Canon already has a shaking sensor in place, so it is not like they have to completely retool the line, just add some lines of code.

[Jon]

Longisland is dead on. Yes, we all know (at least those of us who have taken the time to read about it or use it) that in-lens is better, and that becomes more appearant the longer you go, but not everyone needs to go long, or wants to pay the premium. Now, if you are buying a 5D or a 1 series, you probably have the money and the experience to desire in-lens IS, but I would guess that a very high number of XT and XTi users will never buy a lens after their first (the kit or maybe one of the popular kit replacements like the Tamron or Sigma 17-50s) To those people (who are a quickly growing segment of the DSLR market)in-body IS would be much better than no IS, and at that focal range, in-body IS is quite effective. As someone mentioned earlier, Canon already has a shaking sensor in place, so it is not like they have to completely retool the line, just add some lines of code.Actully, it is like they'll have to retool. The IS sensor needs to be precisely-controlled in specific directions, and a gyro-sensing mechanism has to be incorporated; the dust-clearing action just has to shake (randomly,as it were; precise control not needed) the low-pass filter (not the sensor).