Canon EOS IS Bodies

I was just perusing Bob Atkin's prediction list and it seems a big wish item is

an IS equipped body. What would happen if you mounted an IS lens to an IS body?

I don't think any systems offer both right now (IS bodies and IS lenses). I

guess you can always just shut it off but seems like there would be some sort of

interference here... as opposed to gaining 6 stops :)

 

Makes me think that Canon (so invested in the IS lens world) would have little

interest in making an IS SLR body. But what do I know.

 

Just food for thought...

 

 

aaron

Here's hoping Canon keeps the in-body IS for their consumer P&S cameras. It's my guess and speculative opinion that true IS is best when built into the lens. So, here's more hoping I.S. doesn't find its way into EOS bodies any time soon.

<<Here's hoping Canon keeps the in-body IS for their consumer P&S cameras.>>

 

It would be quite a change if Canon went to CCD-shift IS for their point and shoots as well.

The easy solution is just to turn off in-body IS when an IS lens is mounted.

 

Having owned a K100D with SR, in-body IS absolutely rocks for low-light work. Nothing like shooting with IS and a 50mm wide open in low light.

<p>Or the body could turn off the lens' IS. We already know this is possible, as some EOS bodies turn off IS when you use bulb or self-timer.</p>

The camera and lens are smart enough to figure it out. My guess is the best option would be for the lens IS to dominate, turning off the IS in the body if IS in the lens was detected.

 

If both operated it would be chaos!

 

I don't expect to see it, but I want it. There's a big difference.

 

I'm not sure what would make Canon do it. If they did do it I'd expect to first see it in an entry level DSLR like the Rebel, since the Sony and Pentax cameras have it. If Nikon took that route too, it might force Canon's hand.

 

I'm sure Canon would FAR rather sell IS lenses than make an IS body. Only economics could force them into doing it. If everybody else did, and they were losing sales, they just might. But not yet. There's certainly no technical reason why that can't do it, at least for APS-C sized sensors. For full frame it's a bit more difficult since there's a bigger mass to move and you'd have to find some extra room in the camera body to allow for movement, actuators etc.

Aaron, could you post a link to that list for those of us who are out of the loop?

 

Thanks

A full frame sensor might move outside the image circle as well. I think it will only work reliably with crop frame sensors.

 

I would like it too. It would be great to have IS with some small fast primes. But if Canon did this they would loose future IS lens sales for those cameras. It'd be like shooting yourself in the foot so you can keep up with the Jonses. I can't see it happening.

Ideally, I'd like both to be available. Canon is correct in saying that Lens-based IS is more effective, but that's not the same as saying sensor-based IS is ineffective. I'd certainly prefer (and be willing to pay for) IS on telephotos, but as a pretty serious low-light street shooter, in-body IS with short primes or zooms would be even more useful. A 24/1.4L and in-body IS on a 1.6x crop body would rock my world (A 20/1.4 would be even better).

 

And Canon actually has a big advantage with in-body IS as they have very few cropped-sensor dedicated lenses to worry about coverage with (And two of the 5 are already IS). In-body IS requires more coverage from the lens than lens IS.

I remember reading a year or so ago an interview with some Canon head honchos. They

specifically addressed CCD-shift stabilization by saying something like "At this time we

believe that lens based stabilization is better, but it is possible that we will use CCD shift in

the future." Of course, if you study or know the Japanese you know that beating around the

bush is normal. My guess is that if they see Canons being ditched for DSLRs like the

Sonyolta or the Pentax in droves they would market a CCD shift camera. I would not be

shocked if they had already constructed a prototype to experiment with the possibility.

http://www.bobatkins.com/photography/digital/canon_predictions_2007.html

Canon will do it. It is only a matter of time. They have said that while they believe lens stabilisation is better, they have also said that they can't provide IS for primes because there are not enough elements to play with.

 

The market will gravitate to IS with primes, and in body stabilisation offers this. It is most likely to feature at the entry level in drebels and 30D bodies.

 

Lens based IS in the future will probably be restricted to pro telephotos lenses, where it remains the optimal solution.

Not necessarily, if the Body IS (BIS) know about the Lens IS (LIS) and compensate for the rest of the shake (ROTS) !

ILA (I love Abbreviation)

"Not necessarily, if the Body IS (BIS) know about the Lens IS (LIS) and compensate for the rest of the shake (ROTS) ! ILA (I love Abbreviation)"

 

 

What do you mean "compensate for the rest of the shake". the amount of hand shake is the same whether the IS is on or off, therefore you cant say "for the rest of the shake" cause there was no hand shake being reduced when using IS in the first place.

 

IS only bends the light so that when light hits the sensor, the image appears unshaken, but surely , your hand still shakes , IS on or off.

It's completely safe to assume that if Canon ever did introduce in-body IS, there would be a protocol to avoid any conflict with lens IS. Most likely this would involve disabling one or the other, but in any event I think it very unlikely that there would be any benefit beyond the better of the two.

 

The downside of in-camera IS is that it only affects what goes onto the sensor, not (a) what you see in the finder nor (b) what the AF system sees. I certainly like seeing a steady image in the finder, and if I do need to fine-tune the focusing I find that makes it easier to do so. I strongly suspect (does anyone actually know?) that the AF system works better with a stabilised image falling on it, and if that is correct then it is a significant consideration.

AFAIK, IS lenses can't compensate for rotation. So when the body detects an IS lens, it would make sense to let the body compensate for rotation, but not for vertical and horisontal movement. I.e. the center of the sensor should not move relative to the camera, but the corners may move to keep the sensor's orientation stable relative to gravity.

IS in body sucks - that's why canon makes IS lenses.

Michael: Canon introduced Lens IS back when Body IS wasn't possible.

 

Body IS works, and works well(Having owned a K100D, it's very much not hype). Lens IS works even better, but is impractical on small lenses, especially primes shorter than 100mm. The combination of both would be best.

I object that lens IS is not practical for wide lenses. Both the EF-S17-55 and the EF-S 17-85 have IS. I own the EF-S 17-85 and IS works wonders in low light.

 

However, IF Canon can master it, combining body IS AND lens IS for a total of 4 or 5 (?) f-stops gain would be a major breakthrough. The new EF 70-200/4IS lens gives 4 stops (Canon claims) with Lens IS only. Wildlife photographers with EF 400, 500 and 600mm lenses would jump on this opportunity. Can you imagine shooting with EF600+2xTC at 1/30?

The disadvantage of lens IS is that it need 2 seconds to stabilize the image properly, as said in the manual. In an in body IS, you dont have to wait 2 seconds, you can shoot instantly , cause the in body IS starts moving just before the shutter is open, and possible save power.

 

It would be for the best , if you can override the selection between the 2 types of IS. If ever Canon makes in body IS.

>> If both operated it would be chaos!

 

To date I have not head a single technological explanation which would state why it is physically impossible to have both activated at the same time and adding LIS to BIS.

 

Of course, the simplest solution would be that BIS will shut itself down if LIS is activated.

 

 

Happy shooting,

Yakim.

"Yakim Peled Photo.net Patron Prolific Poster, feb 15, 2007; 07:11 a.m.

>> If both operated it would be chaos!

 

To date I have not head a single technological explanation which would state why it is physically impossible to have both activated at the same time and adding LIS to BIS."

 

Yakim,

 

If a non-technological explanation works would you consider it :-)

 

If the two systems are not coordinated, then it's obvious chaos would reign: While LIS is correcting a particular movement, BIS is correcting the same movement -- the result is overcorrection.

 

If the two systems are coordinated, then they need a protocol decide who is going to be the Boss. Then the Boss needs to know everything about his own capabilities and everything about Employee's capabilities so that when a movement needs to be corrected, s/he can decide whether one or the other of them is going to do the correction, or whether they're going to cooperate. Having made that decision, the Boss and then has to communicate it to the Employee.

 

Oops! Oh well, maybe they'll have time to correct the next movement :-)

Sorry. I fail to understand why that procedure prevents the coordination of both systems.

 

And besides, what is overcorrection? If I shoot at 1/8000 and IS is on, where is the problem?

 

Happy shooting,

Yakim.

Sigh... Sorry, guess I'm not as good at metaphoric explanation as I thought.

C

o-ordination isn't magic that just happens. You have two CPUs each with a program, and a physical system that the program controls. The programs are written to control their systems, not to co-ordinate with another program/system.

 

Co-ordinating two separate systems is itself a separate co-ordination problem which neither program is programmed to do.

 

In this case, one system, LIS, generally performs better than the other, i.e., over a wide range of inputs it makes better outputs. So the tricky part is adding a third program which decides which system is better suited to handle a given input and to communicate that decision to the two systems so that, for example, both of them don't try to correct the same input.

 

But all this co-ordination takes time. The upshot is that response is slower and more complicated than what you get from just letting either one of the existing programs/systems handle the job. The co-ordination problem is not intractable, but generally its more complicated than either of the basic control programs -- controlling controllers is harder than controlling a system.

 

Try another analogy: A car has two equal but separate sets of controls -- steering wheel, brake pedal, accelerator. One set is in the front next to the dividing lane, and the other set is in the rear next to the curb. These controls are the physical system.

 

Both you and your partner operate the car simultaneously. You two are the separate programs.

 

If you ignore each other's efforts, it would make a great, comic brief for others -- disaster for the two of you! If you try to co-ordinate by shouting your intentions, what are the odds the car is going to get where its going faster than if just one of you does the driving?

 

Hmm... This last analogy might work as the outline for a new TV reality show.

>> So the tricky part is adding a third program which decides which system is better suited to handle a given input and to communicate that decision to the two systems so that, for example, both of them don't try to correct the same input.

 

 

Why add a third program? Simply activate both. As I see it, you can NOT have overcorrection. Ever. See my example above.

 

Happy shooting,

Yakim.