« Last Day, Last Chance! Our Book Sale Is Ending Today | Main | Telephoto Reach and OIS vs. IBIS »

Monday, 10 July 2017

Comments

I hadn't realised IS only came into being (for camera lenses) as late as 1995 - I had thought it was there from the beginning of the EOS system. But I'm probably thinking about USM which was there (theoretically) from the beginning.

Being a Canon shooter I'm firmly on the in-lens side of the fence. One argument that I have always found persuasive is this. In-lens systems allow the solution to be tailored to the requirements of that particular lens. The stabilization requirements for e.g. a 35mm lens and a 300mm lens will probably be different because the amount of image movement is likely to be different. Therefore the solution, which should be tailored to the requirement, might also be different. (The same is true of AF systems, and diaphragm movements, of course.) That can only be done with an in-lens solution. Surely an IBIS system is a one size fits all solution, which can't be ideal for most lenses?

I like the idea of hybrid solutions, however.

[The first USM lens was the EF 35–135mm ƒ/4–5.6 USM lens of 1990. --Mike]

Hmmm. That "long lens" thing is really counterintuitive. I would think the corrections needed would be _greater_ with a long lens because every photographer twitch is magnified 8x in a 300mm lens, or something like that. Maybe if you are moving only the iris . . . ??? Would love to understand the mechanics of all that better if there are any engineers in the TOP-verse.

[It's because of the location of the correction. A small group of elements in the middle of the lens is all that has to be moved. IANAE. --Mike]

"Olympus claims that IS technology is limited by the movement of the earth:

Is this Olympus way of trying to prevent customer whining if they don't get sharp photos during an earthquake?

“Olympus claims that IS technology is limited by the movement of the earth.” The other day I tried it with the 300mm, IBIS + OIS.
It's true! As soon as I touched the release button the EVF image was completely frozen and I could feel the tectonic plate vibrate under my feet.

The multi-seconds hand-held exposures with Olympus M1 mark ii are real -- I've done it, and you can read about it by others in various places. But it requires combining IBIS and OIS in one special (12-100 mm) lens. Incidentally, one thing that OIS cannot do is correct for rotation about the axis of the lens. That's why IBIS is only "5-axis". The sixth axis would be the AF by moving the image plane that you describe. But it is limited in its effect to short focal length lenses. Even two generations ago, I found you could take pretty nice macro photos handheld with objects a few inches from the lens with the Olympus E-PL5 and their inexpensive 60 mm macro lens.

We (I) tend to think of Canon as a camera company that makes lenses. Maybe they're a lens company that makes cameras.

At list prices, the dollar value of my lenses is much greater than that of my single camera body.

This makes me miss my clunky Canon 28-135 IS lens. It had a reassuring "settling" while it stabilized the image. I could see it through the viewfinder and it made me confident.

I think the EOS M5 already has in body stabilization that works with lens stabilization
https://www.usa.canon.com/internet/portal/us/home/products/details/cameras/eos-m-series-digital-cameras/eos-m5-ef-m-18-150mm-is-stm

Tom Burke's comment is one I've read before from others, but equally valid (from the other side) is that IBIS offers IS with all lenses.

I've shot Sony A mount, digicams that use IS, Nikon F mount and Sony E mount. The IS on my Nikkor 70-200/2.8 is excellent. But the IS that I get with a 30/1.4 or 28/2 on the A6500 beats the VR that I don't get with the Nikkor 35/1.8 (or the 85/1.8 for that matter). Tamron makes a couple VR primes for Nikon and Canon, but otherwise, lens-based IS is only good if it's in the lens.

Overall, I guess my preference would be to have both and make them work together, like Panasonic lenses do. Failing that, I have a mild preference for sensor-based IS simply because it is available for all lenses. But it's not a showstopper - I did switch from Sony to Nikon some years back knowing I'd be giving up IS with the 35 & 85. To me, IS is like sensor quality ... at this point, it's good enough and ubiquitous enough that I can take it for granted and make my buying decisions based on other factors.

So, for S. Wolters, photography makes the earth move. Interesting. There's probably a name for that.

Mike,
I have a strong feeling that OIS sometimes mal functions and causes blurred images. In those situations I use my non stabilized 50mm lens (mostly in low light situations). I think it has given me better images. This may be just a subjective sensation but I believe it is there for sure. It may be because I tend to support the camera on some stable objects in such situations. It is said that the image stabilization should be switched off when using a tripod. I too am for hand held use of camera and hate to switch between stabilization and non stabilization.

PS I've also read that canon has both IBIS and OIS in video mode, but defaults to OIS in Stills mode, so I'm not sure which is accurate

Just for kicks, I did a quick search of the U.S. Patent Office with the terms "camera image stabilization" and "optical image stabilization". (Its really quite easy at the Patent Office website,) Since most foreign companies also file patents in the U.S., it should give an indication of who is innovating in a field. Pairing the above terms with the company names, Canon had 25, Nikon had 6, and Fuji, Pentax, Panasonic, Samsung, Konica Minolta, and others all had one or more potentially relevant patents filed. Since a single device or development can result in several patents, the number is only an approximation of the level of activity, but it seems that Canon is the most active in this area.

This is at best a semi-sequitur, but your essay reminded me that, as a teenager, I was reliably able to shoot my father’s prewar Zeiss Contax II with its collapsible 50mm f/2 Sonnar lens at 1/10 second, and often could get away with 1/5 second.

I don't know whether that was attributable to my steady hands (they were a lot steadier at 17 than they are at 70) or the design of the Contax’s focal plane shutter which, unlike those of the Leicas of that era, moved vertically rather than horizontally across the film. It certainly had nothing to do with the mass of the camera, which makes even my Fuji X-T2 feel like a brick.

Alas, the Contax was lost during one of my parents’ long-ago moves, but I do have a “Kiev” carbon copy — http://www.ChrisKern.Net/miscl/kievRangefinder.jpg — which continued to be manufactured for many years after the invading Red Army seized the Contax manufacturing equipment in Dresden and Jena, and relocated it to Ukraine. My father, apparently missing his old camera, picked it up during a business trip to Eastern Europe while he was serving as a foreign service officer during the 1970s. I've never tried to run any film through it, but everything seems to be in working order so maybe a modest test is in order.

Nikon's argument for VR being in-lens rather than in-camera:

http://www.nikonusa.com/Images/Micro-Sites/VR/in_lens_vr/

(I'm thinking an EVF is the key to IBIS working effectively. An OVF would give the blurred image Nikon describes)

Yes Patrick, unfortunately a certain A. Einstein seems to have the patents.

Canon's optical image stabilization system has been among their crown jewels for many years, and it keeps getting better. This week, for example, I'm working with Canon's EF 200-400 F4L IS USM which has an integrated and optically matched 1.4 teleconverter. This sucker is a 9 lb honker, Canon's most optically-perfect zoom but certainly the heaviest. Not something designed for handheld shooting. But I have taken several usable handheld images with it using that optical image stabilizing system. The OIS in this lens, and a few other more recent models, features three modes, set on the barrel, where you tell the lens what to expect.

And speaking of binos, I second William's applause for Canon's image stabilizers in big binos! I have a pair similar to his, an 18x50, and they can produce a very smooth handheld view that would be othewise impossible with such a narrow field of view.

It's my understanding that Canon basically OWNS optical image stabilization and licenses the technology to many other optical companies.

The dual IS in my Panasonic GX80 works superbly. Better than IBIS alone in my Olympus cameras, but not quite as spectacular as dual IS with an Olympus camera and (suitably equipped) lens combo. I fear your recent negative experience with an eBay GX8 might have put you off Panasonic's excellent technology: unlike even recent Olympus cameras, the GX80 has a wonderfully smooth and quiet shutter too with no evidence of the shutter shock prevalent in earlier models.

"One argument that I have always found persuasive is this. In-lens systems allow the solution to be tailored to the requirements of that particular lens. The stabilization requirements for e.g. a 35mm lens and a 300mm lens will probably be different because the amount of image movement is likely to be different. Therefore the solution, which should be tailored to the requirement, might also be different."

Right; lens focal length is needed.

Wrong; the AF lenses and camera bodies communicate, so the IBIS knows the FL.

Also: The IBIS systems I know of, Oly, Panny and Sony, all allow the FL of adapted manual focus lenses to be entered.

This can be very exciting, and terrifying! I popped the 8 mm Oly Lenscap Fisheye on my E-M5, put it up to my eye, and half pressed the shutter release. Bang, Crash! Camera tried to jump out of my hand - Turn IT OFF!

The last non-AF lens I had used on that body was 600 mm. No damage, and I learned my lesson - for now.

Less clear is how important focal distance may be. The geometry says it should matter. Oly bodies put focal distance in EXIF* Panny and Sony don't, which doesn't mean they don't know it, at least approximately.

And yet, the Oly IBIS I've used with MF lenses has worked quite well.

* Not terribly accurate, esp. at closest and longest.

On the subject of patents: When the Oly IBIS was introduced, I recall that the talk among the 4/3 discussion forums was that, since Oly already had a patent for shaking the sensor (for cleaning) then that allowed them to get around the prior patents that Minolta would have held for their IBIS. This is all off the top of my head and gleaned from aforementioned net-forums so I guess it falls under the category of "interesting if true".

Great story and finally an explanation for the differing lens- versus body-based stabilization approaches favored by different brands. On my Pentax K3-II, I use Carl Zeiss or Takumar lenses from the 1960s with an adapter, and as has been mentioned with respect to the K-1, get 4 or 5 stops better results than the old "1/focal length" limit for handheld shots. 135mm lens, 1/10 or 1/8 of a second exposure time, works most of the time in low light. Funny that Canon would let itself be compromised in how it offers shake reduction to its customers, just due to its historical connection with binoculars, rather than acknowledge the better usefulness of the in-body stabilization used by its rivals and somehow acquire it for Canon's own models. At least that's my feeling from using my Pentax. Funny that both Canon and Nikon, usually so superior about their technology, are the odd men out in the implementation of this useful feature. Yet another reason to keep one's mind open about cameras from different brands.

Mike,

I too have a 7D languishing in a drawer having developed the black frame problem. Also I had to tape the battery drawer closed. Otherwise I agree, it was a great camera.

As I remember, Bo's correct that IBIS was a Minolta development that Sony inherited when they gained Minolta and their engineers. As I recall comparative tests at around that time, The IBIS was was equivalent up until close to 200mm and after that there was a small advantage to OIS. With developments since then this may or may not still be true.
Further, it was generally felt that OIS was better for video because the mecahanism was further from the recording. I recall speculation that this was why in the earlier days of micro ft, the Panasonic versions (that did better with video) used OIS while the more still photography orientated Oly used IBIS.

I've found my Olympus IBIS to be quite good at longer focal lengths, and the EM1 mark ll I rented was pretty much miraculous with telephoto lengths. My K1 is not as good, but it is reliably decent, and I can count on 1/13 of a second usually being sharp at 50mm with my not-so-steady hands (just slow enough to show a little water blur if you want it).

OIS requires an extra lens element, which means the lens design has to be compromised to do so. Also, OSS seems problematical in wide aperture lenses, because few such lenses have OIS.

Thus IBIS just makes a lot more sense. And IBIS works a lot better with wider angles, allowing you to get really long exposures with such lenses.

And OIS can still be used even though the body has IBIS. So IBIS is a no-brainer, especially considering how affordable it is. Cameras like the EM-10 MK II show that IBIS can be implemented very affordably.

I know that the moment I buy another camera Sony will announce pixel shifting like Hasselblad and Pentax have.

With all the computational power and image recognition in cameras today, some camera company should get object tracking to shift the sensor to lock onto an object when you are panning. It would be really good for racecar photography for instance.

My problem with long exposures has always been the subject moving , not so much the camera. People, and trees in the wind are the worst. Its almost like being backfocused. The background is sharp but the person you are photographing is blurred. That's why I always shot with a leaf shutter and a hint of fill flash way back in the dark ages. It would always sharpen things up.

Oh, one more thing. Luis Alvarez at Bell & Howell patented the in lens optical stabilization we know now using what is essentially a reverse telescope pivoting in the lens in 1963. Kodak patented a system using a bunch of moving prisms in 1945.

Guided missiles were the big application of most of the image stabilization systems in the 60s and they used moving sensors in the form of video tube.

I remember that the way to hotrod Sony cameras for the half inch reel to reel video portapacks was to get a government surplus Newvicon tube that was reputedly out of a vietnam era missile guidance system. Super sensitive all the way into the infra red, and good recovery from camera flashes and I suppose bomb explosions at night.

IS correcting for the earth's rotation speed would be interesting. We're moving about (depending on the distance from the equator)1600 - 1700 km/hr. So, accurate correction would involve camera onboard GPS being used to calculate your location speed. The closer you get to a pole, the slower you travel.
I can't wait, my old Gitzo weighs 8 lbs.
https://image.gsfc.nasa.gov/poetry/ask/a10840.html

to Hugh, who wrote, "Guided missiles were the big application of most of the image stabilization systems in the 60s and they used moving sensors in the form of video tube.

I remember that the way to hotrod Sony cameras for the half inch reel to reel video portapacks was to get a government surplus Newvicon tube that was reputedly out of a vietnam era missile guidance system. Super sensitive all the way into the infra red, and good recovery from camera flashes and I suppose bomb explosions at night. "

Nowadays it's subject tracking, facial recognition, and ever-more detailed and enlargable image sensors. The primary applications of these minor miracles is obviously military. When we focus our cameras, we're using tech created to aim weapons of war. Does anyone think this was all done for artistic purposes?

Working in low light as I do a lot, I also have the problem that subject motion is often the problem -- musicians for sure, even people just talking. (And roller derby of course!)

I did some vaguely careful testing of Nikon VR (and it's not the newer VRII) some years back and found it fairly disappointing, but actually I'm generally disappointed by optical VR, and very frequently get surprisingly useful results from IBIS. My test reports are here. Also the test target I used is available there as a TIFF if you want to do your own testing.

@mouse: I didn't know that IBIS systems could behave differently depending on the FL of the attached lens. Thank you.

As popular still cameras are starting to develop impressive video capabiities, we will begin to see different philosophies of image stabilization required to stabilize around the kind of intentional motions that video or fast sports action introduce. For example, Leica's OIS 90-280 will recognize a pan movement, and ignore sideways motion while stabilizing vertical movement. Fuji's and Olympus' OIS long lenses tend to freeze all motion, so a movement tends to be sluggish at first, followed by a slight overshoot, which can be distracting for video.

opps, forgot to link to that 1963 patent https://www.google.com/patents/US3378326

In-body stabilization does off a few other things: a degree of astro-tracing, sensor tilt-shift and high-resolution images. Various Pentax cameras offer some or all of them while Olympus offer high-resolution images. For that reason I think IBIS is well worth having even if one leaves it switched off a lot of the time, though the best system for longer focal lengths is as you say one which combines stabilization in both body and lens.

It may be worth keeping an eye on the high-resolution capability of IBIS. At present it is limited to tripod-only use because even very small movements mess up the alignment of the final image which is made up of several images shifted half or a single pixel apart. However, if enough in-camera processing becomes available, then hand-held high-resolution shots may become possible with any "shake" eradicated by software processing. That could be a very powerful feature, I think. As it is, my Olympus M43 camera delivers 64 mpx high-res files with better colour and acuity than the regular images, though still tripod-only. That's quite something from a "little" camera.

Minolta supposedly had a plan to make a stabilized camera that moves the film plane somehow but they found it too difficult to make it work in practice. So they were ready when digital came along.
Their A2 was also the first digital camera with a high resolution EVF. Years before others.

"I have a strong feeling that OIS sometimes mal functions and causes blurred images."

That could be shutter shock (or mirror slap if you're using a DSLR) impacting the moving lens element which is usually more sensitive to shutter shock than anything else.

I've never been a binocular person (well, I do have two eyes last time I looked) and now I know that OIS is available. But can you get AF as well? Even if it was just a "focus now" button.

I was working as a industrial cinematographer in the 70s. In one of the helicopter shoots we used what I think was called a Dynalens -- a cantankerous rig with two parallel glass plates stuck in front of the lens. If I remember right, there as a viscous fluid between the glass plates and the glass plates moved to correct for vibrations. It worked. Kinda. Sometimes. I remember a complicated helicopter shoot where it died in the middle of the run. Finished the run handheld with a wide-angle lens.
Modern stabilization is so much simpler, lighter and more reliable for the user.

The comments to this entry are closed.