One area where Canon will fall behind...

Did you read the entire post you replied to? The author was specifically challenging Canon to find a way to "combine" the IS in lens to anti-shake on the sensor, so that there was an additive or multiplicative effect. Frankly, I do believe that this is technically impossible.

The author explicitly acknowledged the "simply turn off the anti-shake CCD sensor" option and called that a dissapointment. Frankly, I don't consider that a dissapointment, I agree with the premise that this would be a fine (grand?) compromise/solution...

Thanks, kfarouki, just about to point that out myself...

I think the biggest loser in that situation (turning off sensor shift with IS lens mounted) would be Canon, as everybody out there with the 28-135, 24-105, 17-85, or 17-55 as their "standard" lens would disregard the sensor-shift feature as a reason to upgrade to the new body, and conversely most consumers owning a sensor-shift body released without the "coordinated" IS feature would certainly have no interest in buying IS glass, which is a good chunk of Canon's bread & butter.

-Blake

lukewarm said:

Don_D said:

So what you are saying is that IS is "pretty much free" in camera,
but costs a several hundred dollars when IS is included in a lens.
Is the "in camera" IS a technology that is significantly less
expensive to manufacture than IS "in lens"?

Click to expand...

Image Stabilization was a new technology back in the '90s. It was
considered a luxury. And as such, Canon could charge exorbitant
amounts for this luxury. The manufacturing costs of I.S. today
doesn't reflect the obscene amounts Canon is charging.

Click to expand...

I have it on good authority that in fact Canon takes a loss on every IS lens they sell! They price them so aggressively to help the consumer, not - as many insinuate - to pad their own pockets by grossly overcharging. It has been pointed out on the boards here before, it makes much more sense to buy IS in each and every lens at a $500 premium when you consider that even with just two such lenses in your collection, it would mean a savings even before you reached your 11th new camera body (with the $100 AS price gouging!)

Canon is helping us consumers. Or so I have heard. ;-)

--
-CW

lukewarm said:

People can make excuses for $1700 lenses having superior Image
stabilization than the ones found in-camera. That may be true.
But keep in mind that technology progresses. Most people upgrade
their dSLR bodies every few years, and the in-camera I.S. will
likely improve with each iteration. Therefore, your entire lens
collection will benefit over the years. On the other hand, your
entire $5000 collection of "L" lenses will not benefit from newer
technology.

Click to expand...

Are you predicting that "in lens" IS technology will improve beyond 3-4 stops?

Maybe the "in camera" IS will improve until it is as good as the present "in lens" IS?

Don
http://www.pbase.com/dond

blakedotfr said:

The way I see Canon implementing this (especially now that they
have sensor-shift dust buzzer technology in the field) is that
their sensor-based IS will be just as effective as everyone else's,
but hopefully since they already have so much IS glass in the field
they'lll be able to work out a way that the sensor-shift and
lens-based systems can work together, so you get 2 stops of
stability with a non-IS prime, and up to 6 (!) stops with a lens
like the 70-200 f/4L IS, as the sensor and lens stabilization
systems communicate and coordinate with eachother. The engineering
required to do this is certainly non-trivial and I don't know if
such a feat is even possible, but it certainly would be yet another
proverbial feather in Canon's cap next to tilt-shift, DO, all the
nice IS glass, low-noise CMOS sensors, FF digital, etc.

Click to expand...

The engineering for combining the two feedback systems would not be trivial. I can imagine them going into oscillations as they get out of sync with one another.

To prevent this, perhaps one system could handle low frequency shake and one high frequency shake...but if this were done there would have to be a switch on both to restore them to "all frequencies" if they were to be operated one at a time.

Understand this is little more than speculation on my part, but I'm sure that Canon engineers are looking into these two systems to explore improvements.
--
Don
http://www.pbase.com/dond

IS in lenses wears out. And IS in lenses is much more expensive... And you need it every lens

sugar

Don_D said:

Jeff Varszegi said:

And since Nikon is their biggest competitor (maybe soon to be
rivalled or exceeded by Pentax and Sony), it would be a coup for
them to differentiate themselves by offering in-body IS early.

Click to expand...

I don't know about a coup....
One disadvantage of in body IS is that you have to pay for it again
and again as you move up to newer camera bodies.
Is that worse than having to pay for it only once for each lens?
In body IS advantages/disadvantages would seem to depend on what IS
or non-IS lenses you own or plan to buy.
Maybe the next XXD will have two models..an IS in body option for
one and not the other....or maybe that's what you meant.
With lots of options it could get confusing!

--
Don
http://www.pbase.com/dond

Click to expand...

--
-----------------------------------------------------------------
I can crop at the long end myself if I want to

some humble pictures : http://www.flickr.com/photos/67259727@N00/

blakedotfr said:

The way I see Canon implementing this (especially now that they
have sensor-shift dust buzzer technology in the field) is that
their sensor-based IS will be just as effective as everyone else's,
but hopefully since they already have so much IS glass in the field
they'lll be able to work out a way that the sensor-shift and
lens-based systems can work together, so you get 2 stops of
stability with a non-IS prime, and up to 6 (!) stops with a lens
like the 70-200 f/4L IS, as the sensor and lens stabilization
systems communicate and coordinate with eachother.

Click to expand...

We shouldn't forget that IS systems just take out the shake....once it is taken out within a certain frequency range and amplitude, there are no more improvements possible.

I don't know if improvements can be made to extend the frequency and amplitude range over which the "in lens" IS works or whether the present systems, for all intents and purposes completely cover the needed corrections.

Don
http://www.pbase.com/dond

my 2cents

I think that Canon will offer IS in the consumer bodies in the future depending upon what Nikon does and if sony takes a bigger share of the market. That or have a usable 12000 ISO.

With the only other significant manufacturer of IS lens being Nikon. I think the route that canon will go to is have IS available in consumer bodies but it will disable the IS in the Lens. With the pro bodies have the IS available in the Body combined with the IS in the Lens.

i.e.

500D 2 stop IS with a 28-135 IS = 2 stops of IS on body or 1.5 Stops in the lens. So one would obviously choose the 2 Stops of IS in the body.

1ds Mk IV 2stop IS with same 28-135 lens would be able to yeald 3.5 stops.

I would also guess that there is going to have to be a limit on effectiveness of IS whether in body or not.

1ds MkIV 2Stops IS with 70-200 4 Stops = 5 stops not 6.

Just a guess.
--
Jay
Equipment on Profile

lukewarm said:

One area where Canon will fall behind Sony & Pentax is the
convenience of I.S. with prime lenses.
I can only imagine how nice it must be to use the reasonably priced
50mm, 85mm, 135mm, and 200mm prime lenses with anti-shake.

Click to expand...

Why?

If your subjects are moving, IS will do nothing to stop subject motion. Lens speed is far more important than IS because IS can do nothing to increase shutter speed. If the available light is so weak as to make hand held shooting impossible because of slow shutter speeds, a tripod will always be the better solution.

Most importantly, a sensor that is mounted in a movable socket will mean less long-term reliability. If in-body IS fails due to wear on its moving parts and sensor misalignment, it will fail with all lenses, and the camera is unusable.

Finally, there is no doubt that in-lens IS gives much better possibilities of tailoring the effect to each individual lens, and since the stabilization works directly on the optical system you can see its effect in the viewfinder.

Per Inge Oestmoen, Norway

I have no way to prove it, but I believe that the sensor shake IS and optical IS would coexist without any special programming or switches. And the effect would be additive when necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens. Camera AS detects movement of the projected image. So in cases where the lens completely stills the image, the camera AS will have nothing to do. If the shaking is not completely stopped by the lens, the camera AS will adjust for the remaining shake.

BTW, I don't think that IS costs $500 per lens. My 28-135IS only cost me $399. The 17-85IS was $600. And Canon's 500mm F4LIS sells for $5,500, compared to Nikon's 500mm F4 which is not a VR lens, yet it sells for $7,000. Doesn't look like Canon is gouging to me.

Big plus for in-camera antishake: It would provide stabilization for my Sigma 50-500mm zoom and my Canon 100mm F2.8 macro and my Sigma 70-300mm APO Macro Super and my Canon 50mm F1.8 and the Sigma 18-200mm that I would like to have (but only with IS) and ...well you get the idea. Some lenses will never be available in IS versions, especially 3rd party lenses, and in camera antishake would be so welcome.
--
Bill
20D circa 2004, Rebel XT circa 2005, Nikon F circa 1967

I bet Canon will have inbody IS in a year or two.

I would be willing to upgrade to get it.

--
jerryk.smugmug.com

wburychka said:

BTW, I don't think that IS costs $500 per lens. My 28-135IS only
cost me $399. The 17-85IS was $600. And Canon's 500mm F4LIS sells
for $5,500, compared to Nikon's 500mm F4 which is not a VR lens,
yet it sells for $7,000.

Click to expand...

To get a perfect apples to apples comparison compare the prices of the Canon 70-200mm f2.8L to the IS version and the Canon 70-200mm f4L to the IS version.

--
Henry Richardson
http://www.hrich.com
http://www.printroom.com/pro/intrepid

wburychka said:

I have no way to prove it, but I believe that the sensor shake IS
and optical IS would coexist without any special programming or
switches. And the effect would be additive when
necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

Click to expand...

I believe that existing in-body IS stabilizes using a motion sensor as well, not by detecting motion within the image. It is after all trying to counteract motion of the camera, which may be hard to deduce (for instance) from a moving image of a flight of geese against the blue sky.

If both systems work by motion detection, they could still work together, but it would get more complicated by far and wouldn't work well without communication between lens and camera, which the current protocol surely wasn't designed for.

wburychka said:

BTW, I don't think that IS costs $500 per lens. My 28-135IS only
cost me $399. The 17-85IS was $600. And Canon's 500mm F4LIS sells
for $5,500, compared to Nikon's 500mm F4 which is not a VR lens,
yet it sells for $7,000. Doesn't look like Canon is gouging to me.

Click to expand...

Au contraire. Check the prices of the 70-200 f/4L with and without IS, or the prices of the f/2.8 versions. I agree with you that it probably doesn't COST nearly that much, but they charge what the market will bear. In addition the implementation cost for in-lens IS is surely higher than in-body, due to the requirement for extra high-grade optics, etc.

wburychka said:

Big plus for in-camera antishake: It would provide stabilization
for my Sigma 50-500mm zoom and my Canon 100mm F2.8 macro and my
Sigma 70-300mm APO Macro Super and my Canon 50mm F1.8 and the Sigma
18-200mm that I would like to have (but only with IS) and ...well
you get the idea. Some lenses will never be available in IS
versions, especially 3rd party lenses, and in camera antishake
would be so welcome.
--
Bill
20D circa 2004, Rebel XT circa 2005, Nikon F circa 1967

Click to expand...

wburychka said:

I have no way to prove it, but I believe that the sensor shake IS
and optical IS would coexist without any special programming or
switches. And the effect would be additive when
necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

Click to expand...

Hi Bill,

You may be correct, and I have no way to know either. As you point out, the two feedback systems are independant, one working on the lens movement, the other on the projected image.

However, in your example above, you neatly separated the systems in time...you have the lens system working first, then the camera IS adjusting for the "remaining" shake.

In reality, the two feedback systems are operating in simultaneously in real time both trying to do the same job. This has potential for oscillations.
Consider this possibility:

Time o: The lens servo detects a physical lens movement (that has resulted in the image shifting up +1) and starts to correct -1. At the same time the camera sensor sees the image move up +1 and moves the sensor to correct -1.

Time 1: The image sensor sees the image corrected -1-1=-2, when only -1 was called for....the combined two systems have overcorrected. Can this overcorrection grow into an oscillation if the next motion and set of corrections produce an overshoot in the same direction?

Don
http://www.pbase.com/dond

Don_D said:

wburychka said:

I have no way to prove it, but I believe that the sensor shake IS
and optical IS would coexist without any special programming or
switches. And the effect would be additive when
necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

Click to expand...

Hi Bill,
You may be correct, and I have no way to know either. As you point
out, the two feedback systems are independant, one working on the
lens movement, the other on the projected image.
However, in your example above, you neatly separated the systems in
time...you have the lens system working first, then the camera IS
adjusting for the "remaining" shake.
In reality, the two feedback systems are operating in
simultaneously in real time both trying to do the same job.

Click to expand...

I don't think so. If the possibility of an image-based stabilization system is real, it could of course only deduce shake from the already-stabilized image coming from the lens. This of course has implications for lag. In any event I'm sure that in-body systems currently employ motion sensors.

Jeff Varszegi said:

Don_D said:

wburychka said:

I have no way to prove it, but I believe that the sensor shake IS
and optical IS would coexist without any special programming or
switches. And the effect would be additive when
necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

Click to expand...

Hi Bill,
You may be correct, and I have no way to know either. As you point
out, the two feedback systems are independant, one working on the
lens movement, the other on the projected image.
However, in your example above, you neatly separated the systems in
time...you have the lens system working first, then the camera IS
adjusting for the "remaining" shake.
In reality, the two feedback systems are operating in
simultaneously in real time both trying to do the same job.

Click to expand...

I don't think so. If the possibility of an image-based
stabilization system is real, it could of course only deduce shake
from the already-stabilized image coming from the lens.

Click to expand...

I don't follow....why wouldn't the in camera IS sense the shake (image movement) immediatedly while the lens system was trying to correct (there must be some lens/image movement in order to correct)....these corrections don't happen in zero time....and both systems are active.

Jeff Varszegi said:

This of course has implications for lag.

Click to expand...

Are you saying that the two system operate on different sets of non-overlapping frequencies?

Jeff Varszegi said:

In any event I'm sure that
in-body systems currently employ motion sensors.

Click to expand...

Don't you think such a system could also oscillate?

--
Don
http://www.pbase.com/dond

Don_D said:

Jeff Varszegi said:

In any event I'm sure that
in-body systems currently employ motion sensors.

Click to expand...

Click to expand...

The Pentax system does employ motion sensors and then moves the sensor to compensate.

I am still concerned that such a system, when combined with an "in lens" IS system, could oscillate.

I also don't see how that an in camera IS system used in conjunction with a lens with IS would produce any further image stabilization....it's not cumulative, once the shake is compensated you're done....I don't see any 6x stop improvement for a combined system as some have suggested.
Don
http://www.pbase.com/dond

Lens and in camera IS measure and correct the camera shake not image shake. They will do that independently and will not know what the other is doing unless there is a communication between the two. If let them both do it the net effect will to overcorrection.

Technology for sensor to detect the image shake will be too difficult now. You'll need a very fast sample rate and high S/N plus great processing power.

wburychka said:

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

wburychka said:

--

Click to expand...

Bill
20D circa 2004, Rebel XT circa 2005, Nikon F circa 1967

Click to expand...

Don_D said:

Don_D said:

Jeff Varszegi said:

In any event I'm sure that
in-body systems currently employ motion sensors.

Click to expand...

Click to expand...

The Pentax system does employ motion sensors and then moves the
sensor to compensate.
I am still concerned that such a system, when combined with an "in
lens" IS system, could oscillate.
I also don't see how that an in camera IS system used in
conjunction with a lens with IS would produce any further image
stabilization....it's not cumulative, once the shake is compensated
you're done....I don't see any 6x stop improvement for a combined
system as some have suggested.
Don
http://www.pbase.com/dond

Click to expand...

Someone else supposed that the in-body system would be based on sensing of the image, where it would be cumulative with in-lens IS. As you now realize, this isn't so. I never said that the effects of in-body motion-sensing IS would be cumulative with in-lens motion-sensing IS. You now understand what I was getting at.

About oscillations: there wouldn't necessarily be feedback, unless the in-body IS introduced vibrations that in-lens IS attempted to correct; however, since both systems would shift in response to each motion, each compensation would be overcompensation.

Don_D said:

Jeff Varszegi said:

Don_D said:

wburychka said:

I have no way to prove it, but I believe that the sensor shake IS
and optical IS would coexist without any special programming or
switches. And the effect would be additive when
necessary--automatically.

Consider IS lenses. The IS compensates for shaking of the lens.
Camera AS detects movement of the projected image. So in cases
where the lens completely stills the image, the camera AS will have
nothing to do. If the shaking is not completely stopped by the
lens, the camera AS will adjust for the remaining shake.

Click to expand...

Hi Bill,
You may be correct, and I have no way to know either. As you point
out, the two feedback systems are independant, one working on the
lens movement, the other on the projected image.
However, in your example above, you neatly separated the systems in
time...you have the lens system working first, then the camera IS
adjusting for the "remaining" shake.
In reality, the two feedback systems are operating in
simultaneously in real time both trying to do the same job.

Click to expand...

I don't think so. If the possibility of an image-based
stabilization system is real, it could of course only deduce shake
from the already-stabilized image coming from the lens.

Click to expand...

I don't follow....why wouldn't the in camera IS sense the shake
(image movement) immediatedly while the lens system was trying to
correct (there must be some lens/image movement in order to
correct)....

Click to expand...

Imagine that you look through an IS lens, and are capable of wiggling the camera in response to the shifting of the image. Would you begin compensating based on the shifting of the image BEFORE the lens IS had time to take action? Of course not, because you are always observing the result of the lens IS.

In the (wrong) scenario of image-sensing body IS, the body IS is in the same boat. Put simply, Its inputs are the outputs of the lens IS system. It can ONLY take effect after in-lens IS.

Don_D said:

....these corrections don't happen in zero time....and both
systems are active.

Click to expand...

Right, hence my statement about lag.

Don_D said:

Don_D said:

This of course has implications for lag.

Click to expand...

Are you saying that the two system operate on different sets of
non-overlapping frequencies?

Click to expand...

No, just that the body IS wouldn't be able to start its work on a particular vibration until after the in-lens IS had taken its best shot.

Don_D said:

Don_D said:

In any event I'm sure that
in-body systems currently employ motion sensors.

Click to expand...

Don't you think such a system could also oscillate?

Click to expand...

Well, it would have to oscillate by design in some way, in its function of dampening oscillations. There wouldn't necessarily be any feedback.