In body IS

[robrone9]

Hey Everybody,

Just wondering what the general consensus was on the forums about in body IS of late... my dad was in town this week, and had me snap some shots with his P&S, and it was pretty handy in some instances. Which got me thinking about the odds of seeing a Canon DSLR with in body IS anytime in the near future - and maybe wishful thinking with the approach of a new 5D seemingly eminent.

I wonder how long it is until they succumb to the pressure and add sensor shift technology to a DSLR (the same as they had to with sensor cleaning tech - though the IS would hopefully work better).

I would pay out the nose for a 5D with in body IS -- and Canon will probably make sure of that if it ever becomes reality. IMO Canon knows it needs to stop dallying with features and start competing again... Which brings me back around to the question,

What do you think about in body IS, and would you be interested in such a DSLR if it could provide 2 stops equivalence, 3 stops, 4 stops? Or would you just leave such a feature in the "off" position and rather not pay for it?

 

[DIGIC]

... because that would hurt badly the IS lens market.

Anyhow, if Canon EVER do it, they will make sure it only gives you 1-2 stops extra with the inbody IS, BUT together with the IS on the lens, it may give you 3-4 stops.

Something like that... meaning they will try hard to ensure this will not hurt much their IS lens market, which is a high profit (just look at the price diff between the 70-200 f2.8 IS and non-IS... it is ASTONISHING!)

--
DiG!C
http://www.pbase.com/hugoneto
(PBase Supporter)
http://digitalphotography.blog.pt/
(Digital Photography Techniques Blog)

 

[PhillipA]

I remember reading somewhere that to adequately damp movement on a 1.6 crop camera with a 300mm lens the sensor would have to shift by up to 10mm (3/8"). Perhaps that's why the in-body IS cameras have an indicator to let you know if the maximum IS level has been reached. It is obviously more effective to move a lens element at or near the nodal point of the lens a small amount than to move a sensor at the "film plane" a large amout. IS is generally required more the longer the lens is, whereas in-body IS becomes less effective the longer the lens is.

 

[Nathan Yan]

I have a feeling this will runaway into a long thread, but since you asked for general consensus here is what the general consensus is, more or less:

There is a lot of noise over the effectiveness of either lens-shift or sensor-shift. Despite general forum sentiment that lens-shift is universally better, this has yet to be proven in consistent, standardized tests (primarily because they don't exist). However I'm personally of the opinion that as of today, lens-shift is still more effective.

Many people aren't interested in sensor-shift IS because they regard it as inferior. There is no way to convince them that it is a legitimate alternative or even supplement.

Sensor-shift won't hurt anybody, other than perhaps increased manufacturing cost. For that it should be welcomed, but there are some who will gripe even then that Canon should not waste their time developing it, or that they wouldn't want to buy a camera that had it.

You will likely see sensor-shift on a consumer-level body before the pro-level 1d bodies.

It isn't likely to happen soon, because Canon is too invested and makes too much on IS lenses. If either of the big two (Canon or Nikon) jumps the gun, however, you can expect the other to hop onboard fairly quickly.

 

[Dave Peters]

I dont think its likely.

Correct me if I'm wrong but I dont think anyone has managed to produce an in-body IS system on a full frame sensor. You need extra image coverage over the normal sensor size because the sensor moves outside the normal coverage area. On a crop body with a full frame lens, you have that extra coverage.

In body IS would be nice but only because canon have some areas not covered by IS lenses - like Macro. Also for 3rd party lenses like Sigma.

I would think Canon are more likely to add IS to the remaining lenses rather than help the competition by effectively adding IS to all lenses via the body.

--
Dave Peters

 

[Nathan Yan]

It is obviously more effective to move a lens element at or near the
nodal point of the lens a small amount than to move a sensor at the
"film plane" a large amout. IS is generally required more the longer
the lens is, whereas in-body IS becomes less effective the longer the
lens is.

This doesn't make sense. A lens element would have to move much more than a sensor (now, I'll allow that a lens element may have more room to shift). Think about a long lens mounted on a camera body, and imagine it pivoting about the camera body (rotational shake). What moves more - the sensor, or the front element of the lens?

 

[Nathan Yan]

I dont think its likely.

Correct me if I'm wrong but I dont think anyone has managed to
produce an in-body IS system on a full frame sensor. You need extra
image coverage over the normal sensor size because the sensor moves
outside the normal coverage area. On a crop body with a full frame
lens, you have that extra coverage.

This is because the only companies who have produced full-frame sensors are the only two companies who have established lens-based IS systems. If or when Sony releases their full-frame, I would fully expect it to have sensor-shift IS. Until then neither you or I can say it can or can't be done, although I'll claim that there's really no reason it can't be done.

A full-frame sensor is not going to be significantly heavier. It's definitely not as heavy as a lens element! There should be no issues moving it at the required speed. You didn't say this but others have, so I'm pre-empting this.

The other is coverage area. True, you will likely end up with more vignetting when using IS. However, this is not sensor-shift specific. Crop lenses with IS, such as the Canon 17-55IS and Canon 17-85IS, have this same problem. Most of the time you're not going to notice it (your "blur" and thus the region that vignetting affects might be 100 pixels or less). At worst you have vignetting, which is not a big deal to correct.

In body IS would be nice but only because canon have some areas not
covered by IS lenses - like Macro. Also for 3rd party lenses like
Sigma.

This would be the main advantage of sensor-shift to me. Support for lenses that Canon/Nikon refuse to implement IS on. Another is cost.

I would think Canon are more likely to add IS to the remaining lenses
rather than help the competition by effectively adding IS to all
lenses via the body.

Good point, I forget to mention this. Then again you have to ask what Canon should be more worried about - more people buying alternative lenses from Sigma/Tamron, or more people buying other bodies and thus different systems entirely?

 

[Trevor Allen]

I think body IS will need to be added, to both Canon & Nikon, or they will loose new sales to the competition.

 

[Marcel Knaapen]

The reason IS is so successful on the long lenses is that the image in the viewfinder is stabilized. This gives much more advantage than just being able to shoot with slower shutter speeds.

Especially sport and wildlife shooters (shooting moving subjects) that use long lenses appreciate mainly this aspect of IS, they need short shutter speeds to freeze action anyway. So IS is not just about allowing longer shutter speeds, but for stabilizing the image in the viewfinder.

CANON also use IS in some of their binoculars, allowing handholding binoculars with a magnification of 18x whereas normally max 10x is considered hand- holdable.

Physically moving the sensor is very tricky. A very small movement in the direction towards/ or from the lens will result in focus problems. Allowing movement up down and left right means there has to be some room for the sensor to move. this conflicts with the requirement that you want the distance lens-sensor to be as stable as possible (hey that's what calibration is all about!) tolerances will need to be higher and focus will not be stable.

In consumer (video) camera's in camera anti shake is only done by cropping the image and dynamically change the crop area. I don't think it prevents motion blur on stills!

 

[MikeA]

The first reply you got noted that putting image stabilization into lenses provides "high profit" to Canon. I don't know how anyone can make such a claim credibly without knowing what the actual ("high" or otherwise) profits are. Making precision equipment such as a lens -- and then adding yet more precision equipment (the stabilizing system) -- must be costly. So is an IS lens truly a "higher profit" item? Ok. What are the numbers...

The conventional wisdom about this always seems to be: Canon wouldn't want to add "on-board" IS, thereby irritating its customers who bought IS-equipped lenses. I don't see why this has to be an issue at all. Why not the following approach: in a camera body with sensor-based stabilization, the on-board system would be automatically switched off when an IS-equipped lens is mounted and its own IS is used. Turn that off or mount a non-IS-equipped lens, and you'd have a choice to use the sensor-based IS. Stabilization in the lens provides the obvious advantage of allowing you to see the stabilizing effect in the viewfinder. Sensor-based stabilization could provide another advantage -- some stabilization when none is available in the lens. Or will manufacturers start putting IS into every lens they make -- wide-angle lenses included -- thereby increasing the costs of all lenses by quite a bit? (Would that be a smart business model?) I can't imagine resenting a manufacturer for adding a potentially very useful feature. I'd appreciate having the option for it when I'm using non-IS lenses. I don't understand the "zero-sum" point of view about this...

 

[NigelMoore]

... that the stabilised image is not projected to the viewfinder, so you have no idea of how the shot will look until it's taken. Lens IS allows to to compose the shot as it will turn out.

This may or may not be a problem.
--
second frog in

 

[MikeA]

The reason IS is so successful on the long lenses is that the image
in the viewfinder is stabilized. This gives much more advantage than
just being able to shoot with slower shutter speeds.

A long-time fine-arts photographer who once posted articles on Luminous-Landscape.com wrote an article about his (then-)new Minolta digital -- this was not long before Minolta disappointed him big-time by closing up shop. He had their SLR with on-board stabilization. Of course he couldn't see the effect in "real time" in the viewfinder. He had to gauge it by looking at the indicators (LEDs, perhaps) in the viewfinder that displayed the extent to which the stabilization was working. That disadvantage aside, he said the system worked well for him and he included an example -- a pretty sharp image, hand-held and taken in low light, that he said he would not have been able to get otherwise. The system worked .

Especially sport and wildlife shooters (shooting moving subjects)
that use long lenses appreciate mainly this aspect of IS

I'm sure they do. And does that being an advantage somehow weigh against having an additional implementation of the feature, for other situations when it would be useful?

Physically moving the sensor is very tricky. A very small movement in
the direction towards/ or from the lens will result in focus
problems.

Is this so difficult that no other manufacturers have done it? Obviously they have. Does it work? Apparently it does. It certainly worked for the photographer who wrote the article on Luminous-landscape. Whether or not the on-board system is the ideal, best-of-all-possible-worlds solution doesn't weigh against its potential value.

In consumer (video) camera's in camera anti shake is only done by
cropping the image and dynamically change the crop area. I don't
think it prevents motion blur on stills!

If it doesn't work, then why have manufacturers spent a no-doubt-considerable amount of money putting it into their d-SLRs? Perhaps people having real-world experience with sensor-based image stabilization can comment (real-world experience being a whole lot more useful than speculation).

 

[Dave Peters]

Adding IS to the body may lose customers to Sigma, Tamron etc because their lenses are then stabilised. Currently if you want IS you have to buy Canon - with a couple of exceptions.

Canon now have a lower cost IS and have launched cheaper lenses with IS so I dont think it is adding much to the cost now. Similar arguments were used by Minolta about adding motors to all lenses when they had one in the body with a mechanical drive. Now almost everyone agrees the motor in the lens is better and there are cheap lenses with them.

--
Dave Peters

 

[Dave Peters]

If it doesn't work, then why have manufacturers spent a
no-doubt-considerable amount of money putting it into their d-SLRs?
Perhaps people having real-world experience with sensor-based image
stabilization can comment (real-world experience being a whole lot
more useful than speculation).

I have both the Minolta 7D (sensor AS) and Canon (lens IS). Both systems work well and I would be happy using either from the point of view of stabilisation. I use the Minolta for butterflies hand held as I can't buy an IS Canon macro lens. I bought the Canon for long lenses as the Minolta system was hacked down by Sony and Sigma 300-800 was not available for Minolta/Sony.

On the cameralabs site there is a nice video showing an Olympus camera with live view. They switch between lens IS and body IS to show they both work equally well and then demonstrate that when used together its a disaster.

--
Dave Peters

 

[Trevor Allen]

Yes, but I seen threads on here where, the Bigma (50 to 500mm) has been used with good effect on an IS body, making a cheaper option, & have been tempted myself to try this.

 

[Antony Ellguy]

If we are talking full frame then you need a sensor larger than 36mm x 24mm if you want to move it to stabilise an image. The other manufacturers can do it because they use smaller sensors.

 

[deeyay]

I have had in body IS in my Minolta 7D that I sold a while ago. It was great in the sense that it stabilized all of my lenses. I find it as good for wide angle lenses as for telezooms. Many argue that it's wortless for wide angles (probably because Canon don't have stabilized WA lenses) but I do not think so, I could hand held 11mm shots for 1/2sec which is very usefull for me. I do miss that on my Canon 5D, that my wide angles ain't stabilized. It helped so much for dim light shots with wide angle.

In body IS is bad for some reasons too I think. If it breaks down, it cripples the whole camera leaving it impossible to use. If it's in the lens and breaks down you an still use the camera at least (not sure about the lens?). Also with lens IS you see how the IS works in real-time, this makes it easier to see the actual result and time the shot well. With the in body IS it was more of a hit and miss because I could not see the 5 step "shake meter" on the side of the viewfinder very well at the same time I'm supposed to frame the shot.

Both have some positives and negavtives, it depends on what you as a photographer prefers and what you need for your shooting.
--
Picture agency

• http://danieljansson.se

Web portfolio

• http://danieljansson.com

 

[John Motts]

It is obviously more effective to move a lens element at or near the
nodal point of the lens a small amount than to move a sensor at the
"film plane" a large amout. IS is generally required more the longer
the lens is, whereas in-body IS becomes less effective the longer the
lens is.

This doesn't make sense. A lens element would have to move much more
than a sensor (now, I'll allow that a lens element may have more room
to shift). Think about a long lens mounted on a camera body, and
imagine it pivoting about the camera body (rotational shake). What
moves more - the sensor, or the front element of the lens?

Apologies if I haven't understood here, but surely if the greatest movement is at the end of the lens then surely that is where the IS should take place?

 

[Dave Peters]

Yes, but I seen threads on here where, the Bigma (50 to 500mm) has
been used with good effect on an IS body, making a cheaper option, &
have been tempted myself to try this.

It depends what you want to do with the results and what percentage of shots you want to work. The IQ of this lens is not that great - but web images will look fine as will small/medium prints. The focus works by reporting a false aperture to the camera which has to have an effect on accuracy - hence reduced keeper rate.

I would rather have the 100-400 IS and enlarge a bit more to get the 500 reach.

--
Dave Peters

 

[Dave Peters]

Apologies if I haven't understood here, but surely if the greatest
movement is at the end of the lens then surely that is where the IS
should take place?

Exactly the opposite. You can correct the image with a smaller movement of the elements if you insert them in the lens where the image size is smaller. You may need higher precision but you an use smaller corrective elements hence less movement distance and less mass to move. The lens tube is rigid so you can calculate the correction required at any point along its length.

--
Dave Peters