[Kenntak]

Demp10, I agree those are important points. I want to get into macro photography at some point, and being able to adjust the composition would certainly be useful.

[filoxophy]

Good points all around. FWIW, I didn't see any fanboyism in the initial post, just ignorance on the part of some regarding Pentax's IBIS, and the OP's desire to correct that.

Anyway, as long as these types of discussions are more "here's the facts about A and B--decide what's best for you" rather than "here's why you should choose B over A", I think they can be productive. So kudos to the OP.

As for me, give me IBIS or give me death!

[interested_observer]

Been following the thread since the initial post. Excellent discussion. I wanted to make a couple of additional points.

Regardless of which stabilization approach is best - is probably immaterial, other than for bragging rights. The both do the job, each approach has both advantages and disadvantages.

The point that I wanted to make is that in-body stabilization has been demonstrated to have a wider range of applications:

• O-GPS1 Star Tracking - a very novel and clever approach to star tracking, using the movable sensor. This was touched on previously, and has not been able to be duplicated by Canon and Nikon. Sony and Olympus also have the in camera in place to take advantage of the the approach.
• Digital Shifting - Pentax already (since the K7) has the ability to shift the sensor by hand both horiziontally and vertically - thereby simulating a shift lens (sorry no tilt).

• Sensor Shift In Action: Increasing Field of View - Pentax Tutorial Videos - PentaxForums.com

There are two other approaches that sensor placement has been demonstrated to provide additional functionality.

1. Increased Resolution - by moving the sensor a half pixel both horizontally and vertically, you can essentially triple the resolution of the sensor. The Hasselblad H4D-200MS has a 50MP sensor that is enhanced by this technique to 200MP through the use of combining multiple images offset by the 1/2 pixel.

   • Hasselblad's 200-megapixel camera: $45,000 | Deep Tech - CNET News

2. Converting a Bayer Filter pixel arrangement into a Foveon array - Again by capturing multiple images, with each capture the sensor being shifted by a pixel, you can have each color sensor (red, green and blue) stacked vertically - essentially like a Foveon sensor, thereby increasing the color depth.
3. Combing the previous two techniques together - by both increasing the color depth and increasing the resolution, the current 16MP sensor could easily be a 64MP sensor. Now it would be somewhat limited to landscapes, but it would be easy to add such a mode to the current camera bodies - all the technology is there.

(Credit: Hasselblad)


Sensor shifting does provide quite a bit of added flexibility to the camera body.

[carrrlangas]

Sensor shifting does provide quite a bit of added flexibility to the camera body.

In did. This is amazing. Now I am curious whether the SR mechanism (based on electromagnetic fields) has the precision to achieve the technices described.

[Class A]

I see many advantages for a sensor-based stabilisation scheme and personally prefer it.

One advantage of the lens-bases system, however, is that not only the viewfinder image is stabilised -- a questionable advantage, AFAIC, because you get less feedback about what you are doing wrong -- but also what the AF system sees. In particular in low light or with slow lenses, AF systems need higher integration times and any shake will not help to optimise the focus. There is at least one patent to move the AF optics along with the sensor in order to address this problem. So far, there is not implementation yet, though.

A disadvantage of the lens-based system is that it messes with the optical alignment of the lens. I've read claims that the bokeh of such lenses can suffer. I have no idea how the uniformity of sharpness or other properties (e.g., CA) are affected by strong shake compensation. The sensor-based system does not even have the potential for such issues.

[vanyagor]

Many good points, I will only comment on bad/weak ones.

In-lens stabilization may degrade ultimate image quality. I believe Nikon recommends turning it off to have the best quality.

This is true about any SR system and any manufacturer. Think of SR as the last reserve. Pentax K-5 manual also recommends to turn it off when it is not needed, for a reason.

In-lens stabilization is not necessarily perceptively better. The effectiveness of stabilization, whether it be in camera or in-lens, depends greatly on how the photographer holds and uses the camera.

This point is irrelevant. Of course everything depends on how you hold the camera. Use the tripod and the IS/VR/SR becomes irrelevant altogether.

One experienced photographer compared a nikkor AF-S 400mm f/2.8G ED VR to a Pentax FA*300mm f/2.8 ED [IF] and saw essentially no difference in stabilization. In fact, he gave the advantage to the Pentax system because it reduced the weight in the lens.

It does not make much sense to compare the weight/or the effect of stabilization of a 300mm and 400 mm lens

...It is also my understanding that stablization in the Pentax cameras is step-less, thereby allowing nearly infinite precision as is adjusts.

In technology or even natural sciences there is no such thing as infinite precision. As far as the mentioned step-lessness I guess it refers to the floating sensor design unlike in other sensor based SR systems. Note: it does not compare to in-lens systems, only to other sensor-shift techs. So I suggest to get rid of this part altogether or it needs additional explanation.

[Lowell Goudge]

I think that there are 2 aspects to IS that need to be conspired, first, and it was mentioned before is that IBIS has the ability to correct for rotational displacement, like what you do when you press the shutter, ILIS can't deal with rotation, so the IBIS wins because rotation will always be there in hand held situations,

BUT. What is important to realize is that image stabilization and the stops claimed are relative.

Consider film, there was a golden rule of 1/FL as a minimum shutter speed. This was to give "acceptable sharpness" on an 8x10 inch print. Acceptable sharpness is judged by the same principles as DOF with a circle of confusion of 1/100 of an inch at the print.

In film, not every one could shoot at 1/FL. It was a guide, and each person with their technique, could either do better or worse. Remember that point, IS does not give 3 (or what ever the claim is) stops over 1/FL, it gives 3 stops over what you can do.

Now lets move to digital IS and what pentax bodies can do.

The following were shot with a K7D and a Pentax SMC 300/4 plus SMC-F 1.7x AF converter. I.e. a 510mm focal length, at 1/40 of a second

For the shor, i was standing unsupported, holding the lens, not leaning against anything to either steady myself or the lens


and 100% crop of the head


Now, simple math considering focal length and crop factor says the rule of thumb should be 1/750 of a second, and that 1/40th is more than 4 stops gain.

This does not mean that Pentax offers 4 stops, what it means is that I had a very steady stance good technique, and did not rush the shot.

It is and always will be about technique

[NotSteve]

Quite an interesting thread. SR was definitely a feature that brought me to Pentax when I decided to get a DSLR.

I just wanted to add a point about sensor-based reduction. I believe this was pioneered by Konica-Minolta, and then cross-licensed to others. When the Dimage A200 came out, and I was thinking about upgrading from the Canon powershot A50, I found this a very attractive feature. Anyhow, just wanted to give KM their proper due. They made some interesting innovations (folded optics for point and shoots) before they left the digital camera business.

Here's a quote from a DPR review of the Dynax 7D.

"The Maxxum 7D's 'unique selling point' is its Anti-Shake stabilization system, unique among digital SLR's. Minolta first introduced this feature with the DiMAGE A1, it is unique in its operation because instead of stabilizing a lens element (as in a traditional image stabilization system) the sensor is stabilized. Inside the 7D its six megapixel CCD is mounted on a movable platform controlled by two actuators (x and y axis)."

I also find it interesting how the Japanese camera companies are fairly willing to cross-license to one another, or at least so it seems, versus the Apple/Samsung patent wars. I'm sure there's a good reason for this.

[Docrwm]

Quite an interesting thread. SR was definitely a feature that brought me to Pentax when I decided to get a DSLR.

I just wanted to add a point about sensor-based reduction. I believe this was pioneered by Konica-Minolta, and then cross-licensed to others. When the Dimage A200 came out, and I was thinking about upgrading from the Canon powershot A50, I found this a very attractive feature. Anyhow, just wanted to give KM their proper due. They made some interesting innovations (folded optics for point and shoots) before they left the digital camera business.

Here's a quote from a DPR review of the Dynax 7D.

"The Maxxum 7D's 'unique selling point' is its Anti-Shake stabilization system, unique among digital SLR's. Minolta first introduced this feature with the DiMAGE A1, it is unique in its operation because instead of stabilizing a lens element (as in a traditional image stabilization system) the sensor is stabilized. Inside the 7D its six megapixel CCD is mounted on a movable platform controlled by two actuators (x and y axis)."

I also find it interesting how the Japanese camera companies are fairly willing to cross-license to one another, or at least so it seems, versus the Apple/Samsung patent wars. I'm sure there's a good reason for this.

There is, kigyo keiretsu.

[Fogel70]

I just wanted to add a point about sensor-based reduction. I believe this was pioneered by Konica-Minolta, and then cross-licensed to others.

I doubt that Pentax need any license from Minolta, as Pentax use a completely different technology for SR.

[NotSteve]

I doubt that Pentax need any license from Minolta, as Pentax use a completely different technology for SR.

Thanks for the correction Fogel70 re: the issue of cross-licensing. I did a bit more reading up and yes, Pentax SR is their own particular implementation, with various Pentax patents associated with it. Though, I believe my other points still hold, the sensor shift in general was a Konica-Minolta innovation, and that there is less patent trolling between Japanese companies.

Docwrm, thanks for your point about the keiretsu. Something else to read more about.

[clockwork247]

I am involved in a photography author’s Facebook group. One issue that has come up is image stabilization. I first had to correct the members’ impressions that only Olympus and maybe Sony have image stabilization built into the camera body—Pentax of course has it also. After that, one of the photography experts stated that image stabilization built into the lens is much better, and that Canon and Nikon made the right choice in having it done that way. I commented that the image stabilization works great in the Pentax camera bodies. It is always there--I don’t have to make sure the lens has it. He then posted a long commentary on that matter.

I am a newbie, and I would like to respond simply for the purpose of ensuring that the facts are fairly presented with regard to Pentax. I am not trying to win any debates, just trying to correct any possible misconceptions. The poster may be perfectly right for all I know.

Anyway, the poster stated that basic physics makes it difficult, if not impossible, for stabilization to be more effective at the sensor. The movements of the sensor have to be much larger, while the movements of the lens mechanisms are measured in tiny fractions of a millimeter, and for sensor-based stabilization to be effective it has to be moved up to millimeters at a time. An example is a laser pointer and paper, where it is much easier to keep the laser pointer still than it is to try to move the target around to keep the pointer centered.

Image stabilization in the body also suffers from the inability to know exactly how the lens is zoomed and focused, which means that its correction is inherently somewhat inexact. Every lens would have to be accurately profiled and that data would have to be either stored in the camera or lens for truly accurate correction.

The advantage that sensor-based stabilization has is that if you have multiple lenses it is cheaper, as you only have one stabilization mechanism to deal with. You buy the mechanism once, and it works with all of your lenses.

The reason that Canon and Nikon have elected to do correction in the lenses is that it is easier to do and can be more responsive. Current IS/VR technologies are giving about four stops of improvement, which is far more than what sensor-based correction can do, especially on long lenses.

It is not that sensor-based stabilization is ineffective, but it is inherently much more difficult to pull off well, and it has a lot of limitations, especially as you start using telephoto lenses. It was invented for cost savings, not because it is a better way of doing it.

OS is better, I have tested this with my own lens with canon and my sony IBIS, the canon OS does a better job.

it also shows you through your VF, the IBIS can't do that.

it's better in most ways, except 2

1) it needs an extra glass element if i remember correctly, and some people says sigma OS lens isn't as sharp as the non OS version (same lens but different mount, not different lens).
2) it adds costs to the manufacture, and I can assure you those costs will some how find it ways back into the users wallet.

[Fogel70]

OS is better, I have tested this with my own lens with canon and my sony IBIS, the canon OS does a better job.

it also shows you through your VF, the IBIS can't do that.

it's better in most ways, except 2

1) it needs an extra glass element if i remember correctly, and some people says sigma OS lens isn't as sharp as the non OS version (same lens but different mount, not different lens).
2) it adds costs to the manufacture, and I can assure you those costs will some how find it ways back into the ,users wallet.

Usually as optical stabilization is added, it more than one extra lens element added. Stabilized lenses often have 2+ more lens elements of the non stabilized version of the lens. But this will vary depending on lens design.

There are a some other disadvantages with optical stabilization.
- It adds size and weight. A optical stabilized version of a consumer lens if often 20-35% heavier, and the size might add up as much too.
- It's not possible to implement on all types of lenses (fast normal or wide angle prime lenses).
- More moving parts in the lens add to the risk of the lens needing service. (a lens might have a life of 25+ years, but a DSLR probably get replaced within 5-7 years).
- Older lenses might have less effective stabilization (or no stabilization at all), so you might not get as much value out of used lenses.

[stevebrot]

I have had some pretty good luck with SR on my K10D. As Lowell noted above, it is helpful to have good technique to start with, but I have shot as slow as 1/8s with my Jupiter-9 85mm with very acceptable results hand-held and regularly shoot at 1/15s with shorter focal lengths. Where I have found that it really shines is when shooting with moderately long glass in good light. Under those conditions, the results are quite comparable to tripod-mounted shots. I would post a few examples, but Flickr is down for maintenance.


Steve

[carrrlangas]

OS is better, I have tested this with my own lens with canon and my sony IBIS, the canon OS does a better job.

it also shows you through your VF, the IBIS can't do that.

it's better in most ways, except 2

1) it needs an extra glass element if i remember correctly, and some people says sigma OS lens isn't as sharp as the non OS version (same lens but different mount, not different lens).
2) it adds costs to the manufacture, and I can assure you those costs will some how find it ways back into the users wallet.

Did you even read the whole thread?