New focus mechanism for Canon 70D

[MoreOrLess]

If/when they can do on-sensor PDAF that is as fast, accurate and well-tracking as the dedicated sensor PDAF, what do we need the flipping mirror for?

Depends on your preference for an EVF/OVF of course but even with this new system it seems low light AF performance won't be as good as tradisional  AF, Canon claim down to -3 EV for the 6D's centre point for example.

It is interesting that Canon has released this tech on an DSLR first rather than a mirrorless system, perhaps looking to protect their core business a bit?

[hjulenissen]

2) For birding and the like, carrying the bigger, heavier, more expensive, slower frame rate D800 and then cropping from the same focal length  to get a bit less reach (15MP in place of the 20MP) sounds like an imperfect substitute! Ignoring all those other factors and addressing only the reach (which is still a bit worse worse) misses the point. Of course, if you already have a D800, you can make do quite adequately for such tasks, but let us not pretend that there is no advantage to using a tool better suited to that task.

I don't think that anyone recommended the D800 for birding. AF, framerate etc was never a part of the discussion. The post that started the discussion:
"APS-C is still relevant for focal length limited uses such as bird photography. " (boldness by me)

My point was that the APS-C format does not carry an inherent "reach advantage". It only does so as long as it offers higher sensel density and lenses capable of feeding those extra sensels with image information.

-h

[hjulenissen]

It is interesting that Canon has released this tech on an DSLR first rather than a mirrorless system, perhaps looking to protect their core business a bit?

Yes. They must have known that the EOS-M would get a beating over AF, and presumably the 70D focusing system have been under development for years internally. So does this make it more or less likely that we will see a "prosumer" EOS-M series camera anytime soon?

The recent EF-S STM lenses seems to be targeted at contrast AF and moving the lens many times prior to locking focus in the shortest amount of time. So what is their role in a world where (presumably/hopefully) all Canon DSLRs will have blistering fast/accurate on-sensor PDAF that only needs a single (accurate) movement of the lens, like the classic EF-S USM lenses?

Being big makes it hard to change.

-h

[Paulo Bizarro]

Well, the folks at Imaging Resource have tried it, and the say that "By integrating accurate, fully-capable phase-detect autofocus over a majority of the image sensor's surface, Canon is fundamentally rewriting the book on autofocus." You can read the whole thing here:

http://www.imaging-resource.com/PRODS/canon-70d/canon-70dA.HTM

If this is not innovative, I don´t know what else is... the 70D will be the camera with the best live view AF bar none, while having the 7D AF module for "normal" operation. Yes, the sensor is probably not as good as Sony ones, but for sure it is good enough for 99.9% of users.

As for the 70D being the successor of the 7D, I think people assuming that are wrong. The 70D is clearly the 60D replacement, even Canon says so. There will be a 7DMKII; is it taking longer than expected? Well, I look at it from the other perspective, the 7D is still so good, that is enjoying an excellent market life.

[fike]

I have a 7D and I don't think it is holding up where IQ is concerned.  Where focus, frame rate, and buffer size is concerned, it is excellent.  As for high-ISO, and dynamic range, compared to newer cameras it is rubbish.  the Nikon D7100 is much, much better, though focus and frame rate aren't quite as good and its buffer is pathetic.  Unless the subject I am shooting is incredibly novel (copulating eagles in front of a rainbow at sunset) I stop shooting my 7D at around ISO 1000.  Above that, I just can't get the tack sharp quality I want in wildlife photos.  

The 7Ds quality is soo far behind the curve that I have moved to an OM-D for everything that is not wildlife or studio flash work (because I have two canon strobes).  

I don't know what inside knowledge people have that says they will absolutely make a 7DII.  I am losing confidence that it will happen.  The feature set of the 70D seems to mostly overtake those of the 7D.  Some of the key differentiating features between them are now moved to lower 70D: micro focus adjustment, weather sealing,  higher frame rate (though not quite as good), decent frame buffer.  One thing about the 70D that remains inferior to the 7D is the body is plastic instead of magnesium.

Someone mentioned shooting a D800 for birds. It is a poor substitute for the added reach of APS-C.  A few problems:  tracking is difficult because its full-frame viewfinder makes the subject too small to easily track and focus; frame rate is bad; even with all that resolution it doesn't match the magnified resolution that the best current APS-C affords (again the D7100) with the same lenses.

Finally (can you tell I am a canon watcher who has been looking for a new APS-C camera) I don't expect to see any revolutionary improvement in dynamic range or high-ISO performance.  Why, you ask.  Because none of the marketing materials even mention image quality improvements.  Do you think for a second that if Canon had finally made a leap in image quality that the Canon marketing people would hesitate for a second to shout it from the rooftops.  Of course they would, but they've got nothing, so they are marketing their features like WiFi, swivel screen, and live preview focus performance--an achievement of dubious real-world value. Those extra focusing pixels, according to everything I have seen, have nothing to do with image capture, so they won't improve the image in any way, except to possibly improve their face-priority focusing (maybe not a worthless feature).

[BJL]

Paulo,

    Thanks for that link, and I agree that this new approach to AF is potentially a big step forward. I just have a few cautions, mostly extracted from that Imaging Resource article:

1) It seems that all the photodiode pairs are in the same orientation, side-by-side, which makes sense when you think about how CMOS sensors are wired in columns. If so, they are sensitive only to horizontal detail: no cross-style AF sensors. My hope is that this could be fixed in the future by having a 2x2 array of four even smaller photodiodes under each micro-lens. (Or maybe with millions of AF sensors, there will almost always be some horizontal detail somewhere in the desired focus region?)

2) Working at light levels down to EV0 sounds like a big improvement over previous in-sensor PDAF, but is still several stops behind the best SLR PDAF mechanisms, which works at down to EV -2 or -3.

3) That article is mostly describing Canon's description and promotional sample videos, with no independent performance testing yet. Their favorable comparisons of the prototype they handled are to the EOS-M, which sets a low bar, and Canon's own PR rather persistently talks of performance improvements compared only to Canon's own previous Live View AF, which lags behind the competition.


All-in-all, I doubt that "dual-diode AF" is ready to replace high end SLR AF for action photography, but it might be good enough for many of us, now or with some second generation refinements.


P. S. With all this emphasis on video, isn't having to always use the rear screen for video composition getting to be annoying? Is Canon (or Nikon) going to offer accessory EVFs for these video-oriented DSLR's some day?

[BJL]

I don't think that anyone recommended the D800 for birding. AF, framerate etc was never a part of the discussion. The post that started the discussion:
"APS-C is still relevant for focal length limited uses such as bird photography. " (boldness by me)

My point was that the APS-C format does not carry an inherent "reach advantage". It only does so as long as it offers higher sensel density and lenses capable of feeding those extra sensels with image information.

-h

I agree that reach is the main issue; I only discussed the D800 because you had offered it as an example. But by the way, I do think that weight and frame rate are also relevant to some of us when choosing cameras for "focal length limited uses", which can include hours of camera carrying and fast changing subjects where burst shooting is useful.

But on the core topic of reach: the persistent fact is that smaller formats do in practice offer more reach with a given focal length, through offering smaller photosite pitch and thus higher absolute resolution. Nancy's sentence that you quote uses the words "is still" and so is about the present state of camera options, which is that:
(a) the D800 gives only 16MP over the same sensor area (1.5x crop) where Nikon DX bodies give 24MP with the same lens
(b) other Nikon and Sony DSLRs give only about 10MP in 1.5x crop.
(c) the best that any Canon 35mm format DSLR gives in the 1.6x crop of Canon EF-S bodies is 9MP, vs 20MP in the 70D.

So I would agree that, putting aside speculation about future trends in sensor resolution,
"APS-C is still relevant for focal length limited uses such as bird photography. " (boldness by me).


I do agree that there is an advantage to having more than enough pixels, to allow for "loose framing" and later cropping. Then again, for me 20MP, or even a mere 16MP, allows adequate cropping latitude for wildlife photography.

[Bernard ODonovan]

Here is an ISO 3200 test shot of the EOS 7D (normally the lighting in these is good enough to make the fine pixel cameras like the D800 look stella):

http://www.imaging-resource.com/PRODS/E7D/FULLRES/E7DhSLI03200_NR_2D.HTM

Here is a Canon sample of the new EOS 70D at ISO 3200. Whilst the light is less diffuse, more direct, and possibly assisting in detail modeling, it looks streets a head of the 7D:

http://canon-premium.webcdn.stream.ne.jp/www09/canon-premium/eosd/samples/eos70d/downloads/07.jpg

Unless Canon misprinted 3200 for 100, I have a very good feeling about this new sensor... Could be wrong when ''imaging-resource'' do their equivalent test, but fingers crossed! 

[fike]

Here is an ISO 3200 test shot of the EOS 7D (normally the lighting in these is good enough to make the fine pixel cameras like the D800 look stella):

http://www.imaging-resource.com/PRODS/E7D/FULLRES/E7DhSLI03200_NR_2D.HTM

Here is a Canon sample of the new EOS 70D at ISO 3200. Whilst the light is less diffuse, more direct, and possibly assisting in detail modeling, it looks streets a head of the 7D:

http://canon-premium.webcdn.stream.ne.jp/www09/canon-premium/eosd/samples/eos70d/downloads/07.jpg

Unless Canon misprinted 3200 for 100, I have a very good feeling about this new sensor... Could be wrong when ''imaging-resource'' do their equivalent test, but fingers crossed! 

That shot does look good, but I am not getting excited until a few of the independent testers show some more controlled tests.  I do see some low frequency noise in there--blotchiness. It is probably better than the 7D but by how much. 

[Bernard ODonovan]

Here is another one. The new RX1R at ISO 3200:

http://www.imaging-resource.com/PRODS/sony-rx1r/FULLRES/RX1RhSLI03200NR2D.HTM

Once again the new EOS 70D is holding up very well against this new Full Frame naked sensor. It could just be the lighting but if you look at the fabric in focus on the Canon shot near the models neck, it really is holding clean detail...

I hope this translates into like for like test shots, but still very promising  

[hjulenissen]

I do agree that there is an advantage to having more than enough pixels, to allow for "loose framing" and later cropping. Then again, for me 20MP, or even a mere 16MP, allows adequate cropping latitude for wildlife photography.

There is also the issue that having reasonable sharpness (being mainly sensor-limited) at 18MP APS-C is quite hard (for me and in many scenarios at least). If you are at all limited by camera/subject movement, lens sharpness, focus accuracy etc, then a 15 MP cropped D800 may rapidly have the same effective "reach" as an 18 MP 7D or 24 MP D7100.

I expect BIF/wildlife photographers to be fully aware of those things (and doing their best to combat them).

-h

[Bernard ODonovan]

That shot does look good, but I am not getting excited until a few of the independent testers show some more controlled tests.  I do see some low frequency noise in there--blotchiness. It is probably better than the 7D but by how much. 

Just read through the ''imaging-resource'' preview and they suggest Canon are not claiming better RAW other than keeping standards but with higher MP. What Canon are claiming it seems is much better in camera JPEG processing. Not much interest to those with a RAW workflow on this site I guess 

DPReview recently did some samples of the Sony Cyber-shot DSCRX1R and I preferred the SONY Jpegs to the DPReview RAW samples in the shots that interested me, so Canon may win back a lot of Customers in the main pool of potential users if they up their game in this area for those that want great out of cam shoots.

I am guessing you work your own RAW files so you may not feel there is enough improvement to consider it or potentially the EOS 7D2 for a replacement