Lens-based vs sensor-based stabilization

[nyer82]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

 

[brightcolours]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

"is that the case still" or "Is that not the case anymore"

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

 

[BarnET]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

Yup it is not better. That canon says that has anything to do with marketing and zero with the technology. That said there are benefits for both systems.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

The only brand that makes decent size sensors with this technology is olympus. And yes these are very effective and actualy more effective as most lens based systems. It also works with all lenses. an huge benefit. Be aware that there were some early problems with shuttershock.

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

Lens based systems also have 1 major benefit with Dslr's.

They stabilize the viewfinder. So it becomes much easier to compose with longer lenses. Then i noticed something with my tamron 17-50mm on the IBIS Pentax. Sometimes i got massive vignetting in 1 corner.

The system shifts the sensor but it does not shift the image circle. So when it moved to it's far side it actually moved to the edge of the image circle of the lens. causing this massive vignetting in the top right corner. This will not happen with an lens based system or when using fullframe glass on an Apsc sensor shift. Since then the image circle would be large enough to allow the movement.

About technical difficulty with wide aperture lenses. These lenses have larger optics which are heavier. Making them move at these frequencies is an challenge an smaller lighter group is much easier to move arround. So yes it does mean higher cost becouse of the technological difficulty.

 

[007peter]

I came from Sony Minolta to Canon. IBIS works, but it doesn't produce as great of a result as In Lens Stabilization. 2 problems to keep in mind

1. "Floating" Moveable Sensor is a BAD IDEA. IBIS works by moving sensor about. This create more problem than its worth. If your battery runs lows, the magnetic charge isn't holding the sensor as securely as it was fully charged.

2. Moving Sensor constantly for motion compensation is a big battery drain. Sensor is heavier, required more torque, and more magnetic power which drains battery quicker. Where as Canon In-Lens-Stabilization is always off in a stand-by mode, and only engage itself when you 1/2 press the shutter.

3. In Body Stabilization is prone to Over-Compensation or Under-Compensation depends on lens you mount on. Heavier lens required more compensation, lighter lens required less compensation. But when you have IBIS, sensor is limited in it range of motion compensation. Depends on the camera calibration, this could result in over-compensation and produced Stabilization Induced Blurring.

If you search M43 forum and E-PM2, E-PM1, you'll noticed many m43 owners advice turning off IBIS at 1/25s, 1/100s, or 1/125s. There are certain shutte-speed where m43 owners find to IBIS will actually induced blur.

After owning Sony and play test Olympus E-PM2, I find that In Lens Stabilization prefer by Panasonic, Canon, and now Sony FE/NEX mount to be the prefer method.

 

[nyer82]

Supposedly the EPM2 doesn't have the latest 5-axis stabilization, only 3-axis... I guess 5 would be the best of the technology at the moment.

 

[nyer82]

Would that be the absolute best current implementation of sensor-based stabilization?

If so, let's compare *that* with what we get from Canon, that's sort of what I'm asking about in this thread.

 

[brightcolours]

I came from Sony Minolta to Canon. IBIS works, but it doesn't produce as great of a result as In Lens Stabilization. 2 problems to keep in mind

1. "Floating" Moveable Sensor is a BAD IDEA. IBIS works by moving sensor about. This create more problem than its worth. If your battery runs lows, the magnetic charge isn't holding the sensor as securely as it was fully charged.

2. Moving Sensor constantly for motion compensation is a big battery drain. Sensor is heavier, required more torque, and more magnetic power which drains battery quicker. Where as Canon In-Lens-Stabilization is always off in a stand-by mode, and only engage itself when you 1/2 press the shutter.

3. In Body Stabilization is prone to Over-Compensation or Under-Compensation depends on lens you mount on. Heavier lens required more compensation, lighter lens required less compensation. But when you have IBIS, sensor is limited in it range of motion compensation. Depends on the camera calibration, this could result in over-compensation and produced Stabilization Induced Blurring.

4. In lens stabilization can be tailored to the lens specifically. If a lens for instance has severe focal length changes over its focus range, with IBIS the camera will need to know the exact focal length (but the exact focal length is not communicated). This makes IBIS less precise. For instance with superzoom lenses that shrink from 200mm at infinity to 130mm at MFD.

If you search M43 forum and E-PM2, E-PM1, you'll noticed many m43 owners advice turning off IBIS at 1/25s, 1/100s, or 1/125s. There are certain shutte-speed where m43 owners find to IBIS will actually induced blur.

Also on the more expensive M1/5/10 models shutter slap seems to induce sensor movement.

After owning Sony and play test Olympus E-PM2, I find that In Lens Stabilization prefer by Panasonic, Canon, and now Sony FE/NEX mount to be the prefer method.

And Nikon.

 

[nyer82]

4. In lens stabilization can be tailored to the lens specifically. If a lens for instance has severe focal length changes over its focus range, with IBIS the camera will need to know the exact focal length (but the exact focal length is not communicated). This makes IBIS less precise. For instance with superzoom lenses that shrink from 200mm at infinity to 130mm at MFD.

I thought exact focal length was communicated (at least for m4/3 lenses). Wouldn't it show up in the Exif, or is that not exact enough? How exact does it even need to be?

Current Canon marketing.

 

[brightcolours]

4. In lens stabilization can be tailored to the lens specifically. If a lens for instance has severe focal length changes over its focus range, with IBIS the camera will need to know the exact focal length (but the exact focal length is not communicated). This makes IBIS less precise. For instance with superzoom lenses that shrink from 200mm at infinity to 130mm at MFD.

I thought exact focal length was communicated (at least for m4/3 lenses). Wouldn't it show up in the Exif, or is that not exact enough? How exact does it even need to be?

Exif will say "200mm" when in fact the real focal length can be 135mm.

Current Canon marketing.

 

[nyer82]

Really? I've never had the EXIF be that much off when I've been zooming.

I actually have Magic Lantern so I can see the zoom markings and what the lens is reporting as focal length. I'm using the 55-250mm... maybe you have a better lens. Anyway, if I set the lens at 135mm, it'll generally report 135mm, but although I can zoom in for a tiny bit and have it remain 135mm until it'll change to 146mm.

If I zoom out in super teeny increments, I manage to stay at 135mm until I get to 131mm, so there's a bit of "play" where the exif is reporting 135mm.. I guess there's some rounding occuring there. My guess is that anywhere from 133mm-140mm that lens is going to report 135mm. (I used the harmonic means to get the range there).

So under your theory, the OIS would stabilize that focal range somewhat differently, and the sensor-based stabilization might be stabilizing using 135mm despite the actual focal length varying somewhat in that range there. I'm not sure how much of a difference it'd make though.

 

[007peter]

Current Canon marketing.

Canon marketing is tell you the TRUTH. Whether you want to believe it that is up to you.

I would strongly encourage you to pickup and try a Canon 70-200 F4 or F/2.8 IS, compare that against the smaller lighter 55-250mm IS

The 1st time I try my Canon 70-200 F4 IS I was shocked by the power of its IS. The (IS) induced KICK is so strong and the Gyro spin so fast, that I can get away with nearly 4~5 stop IS with great result. I also had a 55-250mm IS for traveling portability, but its (IS) is muted and less powerful, since a plastic body lens is so light that it simply doesn't need a set of powerful gyros to deliver the similar result.

The beauty of (IS) in lens is that you can TAILOR the power-level to each lens exactly, perfectly match. But you can't do that with In-body-Stabilization. There is only so much compensation moving a sensor can compensate. Judging by the numerous reported problem in M43 forum about certain shutter-speed to avoids.

Try switing a Tennis Racket vs a badminton racquet, try holding them steady for 1 minutes. You're realized you can't possibly hold the 2 different rackets using the same wrist power. Heavier tennis racket required more power and strong compensation. That is a law of physics.

IBIS is fun and cheap, but keep in mind, you ran into over/under compensation problem soon or later. That is why many M43 shooter advice people to TURN IF OFF when you want absolutely the sharpest photos without blur.

Another problem with IBIS is that moving sensor can contribute to Jello-Warpping effect in Video.

Panasonic, Canon, and Sony are the 3 experts in Professional Video circuits (check NAB shows). It's not coincident that all 3 uses In-Lens-Stabilization for their Video. Moving sensor in video is just a bad idea.

 

[Abu Mahendra]

Really? I've never had the EXIF be that much off when I've been zooming.

I actually have Magic Lantern so I can see the zoom markings and what the lens is reporting as focal length. I'm using the 55-250mm... maybe you have a better lens. Anyway, if I set the lens at 135mm, it'll generally report 135mm, but although I can zoom in for a tiny bit and have it remain 135mm until it'll change to 146mm.

If I zoom out in super teeny increments, I manage to stay at 135mm until I get to 131mm, so there's a bit of "play" where the exif is reporting 135mm.. I guess there's some rounding occuring there. My guess is that anywhere from 133mm-140mm that lens is going to report 135mm. (I used the harmonic means to get the range there).

So under your theory, the OIS would stabilize that focal range somewhat differently, and the sensor-based stabilization might be stabilizing using 135mm despite the actual focal length varying somewhat in that range there. I'm not sure how much of a difference it'd make though.

Google "Focus breathing". What gets reported in EXIF is the nominal focal length value which, depending on focusing distance, may differ from the real focal length.

 

[blue hour]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

The best solution would be to offer both approaches within one system.

Indeed Panasonic does.

The GX 7 comes with a sensor based in body stabilization

and most panasonic zooms offer Mega OIS lens based stabilization.

While using a prime without OIS, you simply turn on the in body stabilizer.

Besides:

Every lens ois is going to fail after a certain (admittedly long) period.

Bodies get replaced quite often, lenses can last for decades.

Roger Cicala at lens rentals once wrote that those tiny, high frequency movement parts won't last forever.

 

[brightcolours]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

The best solution would be to offer both approaches within one system.

Indeed Panasonic does.

The GX 7 comes with a sensor based in body stabilization

and most panasonic zooms offer Mega OIS lens based stabilization.

While using a prime without OIS, you simply turn on the in body stabilizer.

Besides:

Every lens ois is going to fail after a certain (admittedly long) period.

Bodies get replaced quite often, lenses can last for decades.

Replacing your Olympus 4/3rds body with a new one is already quite problematic.

And lets see how long Sony supports A-mount?

Roger Cicala at lens rentals once wrote that those tiny, high frequency movement parts won't last forever.

 

[blue hour]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

The best solution would be to offer both approaches within one system.

Indeed Panasonic does.

The GX 7 comes with a sensor based in body stabilization

and most panasonic zooms offer Mega OIS lens based stabilization.

While using a prime without OIS, you simply turn on the in body stabilizer.

Besides:

Every lens ois is going to fail after a certain (admittedly long) period.

Bodies get replaced quite often, lenses can last for decades.

Replacing your Olympus 4/3rds body with a new one is already quite problematic.

In fact I'd appreciate if Canon would implement an in body stabilization in addition to lens based IS.

Panasonic was just an example, a working proof of concept.

And lets see how long Sony supports A-mount?

As long as sales are sufficient, that's what most companies ought to do.

They could introduce in body stabilization in their E mount cameras.

Roger Cicala at lens rentals once wrote that those tiny, high frequency movement parts won't last forever.

 

[brightcolours]

Canon's probably never going to implement the sensor kind. Anyyyyway, I remember not too long ago there was somewhat of a consensus that lens-based stabilization was definitely a bit better. Canon has *always* said that, which makes sense because they never offered the other option. I wonder now that the technology has grown up somewhat, if that's as true as it once was.

The new sensor-based tech seems to be stabilized among more axes (the best version is stabilized at 5-axes) so I wonder how good it ranks up with current lens-based technology. In the beginning, I remember reading that it was a negligible difference at shorter focal lengths, but at longer ones the lens-based was definitely superior. I think that was when the sensor-based tech wasn't as mature though. Is that the case anymore?

---

[Practically, the the downside with lens-based systems is that you don't get stabilization with every lens, and you sure-as-hell don't get it with vintage ones. There's also supposed to be some kinda difficulty (either expense or practical-wise) with adding stabilization to wide-aperture lenses. Notice the smaller max apertures on Canon's stabilized updates to their primes.]

The best solution would be to offer both approaches within one system.

Indeed Panasonic does.

The GX 7 comes with a sensor based in body stabilization

and most panasonic zooms offer Mega OIS lens based stabilization.

While using a prime without OIS, you simply turn on the in body stabilizer.

Besides:

Every lens ois is going to fail after a certain (admittedly long) period.

Bodies get replaced quite often, lenses can last for decades.

Replacing your Olympus 4/3rds body with a new one is already quite problematic.

In fact I'd appreciate if Canon would implement an in body stabilization in addition to lens based IS.

Panasonic was just an example, a working proof of concept.

And lets see how long Sony supports A-mount?

As long as sales are sufficient, that's what most companies ought to do.

They could introduce in body stabilization in their E mount cameras.

You think they have just forgotten to do so?

Roger Cicala at lens rentals once wrote that those tiny, high frequency movement parts won't last forever.

 

[blue hour]

And lets see how long Sony supports A-mount?

As long as sales are sufficient, that's what most companies ought to do.

They could introduce in body stabilization in their E mount cameras.

You think they have just forgotten to do so?

The Sony Alpha A77 Mark II was announced May 1, 2014.

A new feature allows a stabilized EVF view as well.

I assume we will see a A99 Mark II during Photokina.

I guess Sony repositions A-mount towards an advanced photographer audience.

I wouldn't like to get entangeld in any brand discussion.

My bottom line simply is: Both lens and in body stabilization is possible within one system.

I think every manufacturer could offer a hybrid solution.

I like choice.

 

[nyer82]

Yeah hybrid is definitely best. With a lens-based system attached, I guess it would simply shut off the sensor-stabilization. With an unstabilized lens, it would use the sensor system. Best of all worlds, that one.

Although surely Canon and Nikon must have realized this would cannibalize sales of many of their optical stabilized lenses... perhaps not the very long focal length ones, but I feel like sensor-based stabilization is advanced enough for normal to wide lenses (although still unsure). It would also boost the value of old used, unstabilized lenses even more and they don't get a cut of the used sales.

 

[nyer82]

Good answer. It's not that I don't want to believe it, it's that I'm forcing you guys to explain it by acting contrary.

New question. What about normal to wide-angle stabilization? Would that be more of a toss-up between the two systems?

 

[cberry]

Honestly, I think Canon is right. It's not just about focal length but also camera/lens weight and distribution, oscillation patterns and so on. It's an interesting field of mathematics that requires the camera and lens combination to be evaluated together, as a system. You'd have more firmware upgrades required for every lens if it was the other way round.

The advantages of separate IS systems is that new versions of lenses can bring about improvements in the technology too. Having both sensor and Lens IS is a waste when the latest tech in lenses gives you so much more stability.

I think your answer is that Canon can charge double for the 70-200mm IS II version of the f/4 and people still think that's a bargain compared with the Non IS version.

cb