What's Next for Canon?

[Sella174]

I believe that Canon has too much invested in their current technology and is thus unwilling to discard that investment, even though much of it has been made defunct.

For example their IS system. Lens-based IS was the only option in the days of film cameras. Canon invested heavily in research and then digital arrived. IMHO, sensor-based stabilisation is much better, especially for video as proven by Olympus's IBIS. Yet for Canon their research must have a return on investment (RoI), so they now bring basically every lens out with IS, irrespective of whether it needs it or not, e.g. the new 24mm, 28mm and 35mm lenses.

The same can be said of their focus-system (the one in the base of the camera, not the on-chip implementation). A relic from heavy investment during the film days, which Canon will continue using (with the required mirror) until such time as the research has paid for itself.

It took guts in 1987 to drop the FD system in favour of the unproven EF system and it was a make-or-break decision. Canon, IMHO, must now again make such a decision and take the loss on all that "old" technology that is basically keeping them back. Mirrorless is, with regards to the entry- and mid-level segment, the future.

But the joke is that Canon need not change the EOS/EF/EF-S system in any way. Nothing in the laws of physics dictate that mirrorless must be small and with a miniscule flange distance. Keep the size of the cameras, keep the current flange distance ... just make the dang things mirrorless.

[AvTvM]

But the joke is that Canon need not change the EOS/EF/EF-S system in any way. Nothing in the laws of physics dictate that mirrorless must be small and with a miniscule flange distance. Keep the size of the cameras, keep the current flange distance ... just make the dang things mirrorless.

I agree on most of your post. And yes, mirrorless cameras could also be built with the current EF/EF-S system's flange distance. 

BUT ... I want to get the entire "digital dividend" = 100% electronic cameras with capabilitis that exceed anything possible with a mirror-slapping DSLR. PLUS smaller body AND lower price. Why should electronic devices be any larger than they absolutely need to be? Smartphones do not contain large pockets of thin air either. :-)

Canon may not have to go all the way down to only 18mm flange distance like Sony did with its E-mount ...  but something around 21mm [=approx. half of FF diagonal] might be the ideal choice. Less restrictions in designing lenses [telecentricity, corner IQ], but still very slim camera bodies.

Also, Canon sholud take the plunge and move to a new lens mount soon, as they did in 1987. This time the pain will be far less, sicne flange distance  will be shorter and  all EF-lenses will remain usable via simple optics-free adapters. And over time - and spurred along by decent marketing and discounting - the entire user-base will purchase a new set of lenses (again), in addition to buying new mirrorless camera bodies - sooner than they might have replaced their existing DSLRs with yet another uninspiring, only marginally better iteration of DSLR.

Now, what's not to like in this scenario - from Canon' s business perspective? Plus: invested customers could not "escape" by switching to another brand either, because that would require at least the same investment.   

[Sella174]

... all EF-lenses will remain usable via simple optics-free adapters.

Yes, and Canon should sell those adapters for $20 even if it cost them $200 to manufacture.

[Lawliet]

For example their IS system. Lens-based IS was the only option in the days of film cameras. Canon invested heavily in research and then digital arrived. IMHO, sensor-based stabilisation is much better, especially for video as proven by Olympus's IBIS.

Yet Sony is going back to optical stabilizers. Canon just didn't take the detour.
Also for AF - you need sensor cells that deliver a good signal at short exposure times.  A dedicated module has a massive headstart here.

[AmbientLight]

For example their IS system. Lens-based IS was the only option in the days of film cameras. Canon invested heavily in research and then digital arrived. IMHO, sensor-based stabilisation is much better, especially for video as proven by Olympus's IBIS.

Yet Sony is going back to optical stabilizers. Canon just didn't take the detour.
Also for AF - you need sensor cells that deliver a good signal at short exposure times.  A dedicated module has a massive headstart here.

You are certainly correct regarding optical stabilizers. Having a stabilizer in the camera body will obviously lead to at least some optimization limitations as the vendor cannot fully optimize for specialized lenses (for example for big whites) and in addition it kills off any opportunity to come out with something like the 100mm L macro's IS system.

[AvTvM]

g
You are certainly correct regarding optical stabilizers. Having a stabilizer in the camera body will obviously lead to at least some optimization limitations as the vendor cannot fully optimize for specialized lenses (for example for big whites) and in addition it kills off any opportunity to come out with something like the 100mm L macro's IS system.

Hahaha ..  Really funny! Canon is 2 axes and hybrid af in 100mm L IS macro stabilizes movement in 4 axes. Olympus IBIS is 5 axes. ;-)

I am looking forward to a mirrorless FF-body - maybe sony a9r - with 5 axes in-body is, so i dint't have to buy  in-lens is in a 24/2.8 or 28/2.8. and when a big white with in-lens is is attached, the ibis will automatically be switched off. In one word: Best of all worlds. :-)

[AmbientLight]

g
You are certainly correct regarding optical stabilizers. Having a stabilizer in the camera body will obviously lead to at least some optimization limitations as the vendor cannot fully optimize for specialized lenses (for example for big whites) and in addition it kills off any opportunity to come out with something like the 100mm L macro's IS system.

Hahaha ..  Really funny! Canon is 2 axes and hybrid af in 100mm L IS macro stabilizes movement in 4 axes. Olympus IBIS is 5 axes. ;-)

I am looking forward to a mirrorless FF-body - maybe sony a9r - with 5 axes in-body is, so i dint't have to buy  in-lens is in a 24/2.8 or 28/2.8. and when a big white with in-lens is is attached, the ibis will automatically be switched off. In one word: Best of all worlds. :-)

I don't see axis counting as any feasible way of comparison here.  It must be obvious that from a development standpoint a vendor will run into limitations trying to cater to a broad selection of lens usage scenarios or do you really believe that at Olympus they work miracles instead of engineering? 

[AvTvM]

I don't see axis counting as any feasible way of comparison here.  It must be obvious that from a development standpoint a vendor will run into limitations trying to cater to a broad selection of lens usage scenarios or do you really believe that at Olympus they work miracles instead of engineering? 

I think in-Body stabilization is preferrable for any lens up to 100mm focal length. Beyond that, in-lens-IS is preferable in DSLRs with OVFs as the viewfinder image is also stabilized. As mirrorless cams with EVF will replace DSLRs, sensor-stabilization will be the way to go, since it makes lenses somewhat sharper, lighter, more robust and less expensive and the viewfinder image will also be stabilized then.

In-body sensor stabilization for 6 "axes" - or rather "lateral shift along" plus "rotation around" all 3 axes (x, y, z) - will probably come soon - i.e. including back/forward movement of sensor, which is currently still missing in any manufacturers implementation. Compared to that, Canon's 2-axis "angle-movement" IS looks rather "last century".   

Heck, along with 6-axes Sensor-IS we may even get user-controllable, ultra-high-precision sensor shift and rotation in all 3 axes = tilt/shift with any lens. At least to some degree. ;-)

And yes, Olympus and to a lesser extent also Sony engineers do have a headstart over Canon (and Nikon) when it comes to in-body sensor-shift IS. :-)

[Ruined]

Interesting how people continue to say mirrorless will replace DSLR, while sales trends indicate that statement is the opposite of reality.

[neuroanatomist]

In-body sensor stabilization for 6 "axes" - or rather "lateral shift along" plus "rotation around" all 3 axes (x, y, z) - will probably come soon - i.e. including back/forward movement of sensor, which is currently still missing in any manufacturers implementation. Compared to that, Canon's 2-axis "angle-movement" IS looks rather "last century".   

More like building a better mousetrap, when the traditional design works just fine.  At non-macro distances, the effects of translational shift (in the plane of the sensor) are essentially irrelevant.  I'm not sure how much roll matters, either...but I'd guess not much at all (what would be nice is automatic leveling, but gyroscopes for stabilization aren't able to provide absolute accuracy, only detect and allow compensation for relative motion).  So Canon's 'last century' system compensates for the types of motion that account for the vast majority of 'shake' except at macro distances, and for the latter scenario, they have the 'this century' Hybrid IS than compensates for translational motion in addition to angular.

[AvTvM]

So Canon's 'last century' system compensates for the types of motion that account for the vast majority of 'shake' except at macro distances, and for the latter scenario, they have the 'this century' Hybrid IS than compensates for translational motion in addition to angular.

yes. As I said.  Canon has "2-axis" stabilizer (angular moves) and "4-axis" stabilizer in 100mm Macro. Oly has 5-axis stabilizer (but only) in their latest camera bodies (OMD5 and 1).  I'd imaginge that (micro-) camera shake can really occur in any of the 6 possible axes. Translational along the z-axis (optical axis of the system) is  probably really only relevant in very shallow DOF situations and macro.

Short overview of the 5-axis in Oly IBIS: OLYMPUS OM-D E-M1 - 5-AXIS IMAGE STABILIZATION VIDEO (English)
 
Especially in mirrorless ILCs with short-flange distance, in-body IS is a major asset, since it will also work with any non-IS lens mounted via adapter. Unfortunately Sony has not built their IBIS ("2-axis") into the A7/R - where it would matter most [exacting 36 MP sensor and so far only the 2 native zooms [28-70, 24-70) will have in-lens OSS, but not the 800 Euro 35/2.8 nor the 1000 USD  55/1.8, not to mention any of the other Sony A and E lenses. This is one of the reasons, why I'll probably pass on the A7/R.

[AmbientLight]

So Canon's 'last century' system compensates for the types of motion that account for the vast majority of 'shake' except at macro distances, and for the latter scenario, they have the 'this century' Hybrid IS than compensates for translational motion in addition to angular.

yes. As I said.  Canon has "2-axis" stabilizer (angular moves) and "4-axis" stabilizer in 100mm Macro. Oly has 5-axis stabilizer (but only) in their latest camera bodies (OMD5 and 1).  I'd imaginge that (micro-) camera shake can really occur in any of the 6 possible axes. Translational along the z-axis (optical axis of the system) is  probably really only relevant in very shallow DOF situations and macro.

Short overview of the 5-axis in Oly IBIS: OLYMPUS OM-D E-M1 - 5-AXIS IMAGE STABILIZATION VIDEO (English)
 
Especially in mirrorless ILCs with short-flange distance, in-body IS is a major asset, since it will also work with any non-IS lens mounted via adapter. Unfortunately Sony has not built their IBIS ("2-axis") into the A7/R - where it would matter most [exacting 36 MP sensor and so far only the 2 native zooms [28-70, 24-70) will have in-lens OSS, but not the 800 Euro 35/2.8 nor the 1000 USD  55/1.8, not to mention any of the other Sony A and E lenses. This is one of the reasons, why I'll probably pass on the A7/R.

Strange how Sony contradicts the assumptions that in-body IS is preferable by providing the A7/R without that. Wouldn't they jump on the opportunity to use their existing technology, if it were as preferable? It is really essential to understand that axis-counting is no good as a way to determine effectiveness of an IS system. I believe that the A7/R coming without in-body IS just demonstrates that Sony has proceeded through the learning curve far enough to realize that they can achieve better results with in-lens IS, simply because this is optimized for each lens.

Addressing multiple axis for image stabilization does complicate the overall design, so in contrast to what many might expect a vendor keeping their design as simple and as effective as possible will likely end up with the best product. This is not to say that further development will not include additional axis. This trend will likely remain, but we must not overestimate products by simply counting axis. Olympus is not currently selling a product from the year 2200, even if their product requirement specifications do allow for additional functionality.

[Sporgon]

Have any of you guys banging the drum for sensor based stabilisation actually used it in a dslr ?  From my experiences with it it  is not as effective as a lens based system, even with shorter focal lengths.

[Ruined]

Have any of you guys banging the drum for sensor based stabilisation actually used it in a dslr ?  From my experiences with it it  is not as effective as a lens based system, even with shorter focal lengths.

A lot of time people let marketing specs overwhelm them regardless of performance.  I saw people going nuts on the HD forum about some 32" 4K resolution noname TVs available for under $1000.  Then they wondered after buying it why it looked worse than their 720P TV.  Specs don't tell the whole story and can be easily manipulated to make a product sound better than it actually is.

[AvTvM]

... It is really essential to understand that axis-counting is no good as a way to determine effectiveness of an IS system.

con we agree that in practice camera shake can and will occur "in any direction, sideways and rotational - in all three axes of our three-dimensional world (x, y, z) and even in any combination therof? 

Yes, number of axes stabilized does not tell us anything about an IS system's effectiveness (how well it stabilizes ... 0, 1,2,3,4,+ stops - and under what conditions?). But offering stabilization in more/all directions is essential to build a dramatically better IS system than one working "in 2 directions (x,y - angular) only.

And as for in-lens IS "being specialized for each lens" this really is a marketing joke. Of course it has to be "specialized" and "built to order" for each lens, because of the differences in optical design. After all in-lens IS introduces an additional, moveable lens element/group in the light-path, which would otherwise not necessarily be needed to yield the desired imaging capabilities.

At the sensor level however, shake is shake. It has direction/s, frequency and amplitude. Amplitude will be larger, the narrower the FOV of the mounted lens, that's all. For teh system to work well, it does not have to be tailored to each lens. It just has to be effective in quickly moving the sensor exactly in the right direction/s by the required amount of travel. Irrespectively of whether a lens with 10mm lens or 800mm focal length is mounted, and what optical lyout is used in that lens.

The best in-sensor IS systems today are every bit as effective as in-lens IS ... at a fraction pof the cost. Since it is needed only once per body, not once in every single lens. The only inherent advantage of in-lens IS systems is the ability to stabilize the viewfinder image in a DSLR. In mirrorless cameras with EVF, this is not needed, since the viewfinder image comes directly from the sensor and will automatically be stabilized. No matter what lens, as long as the amplitude of the camera shake is within working limits of the IS system. :-)