DxOMark tests Canon EOS 70D Dual Pixel AF against the Sony SLT-A77

DxOMark has tested the Canon EOS 70D’s live view autofocus system in comparison to the Sony SLT-A77, looking at focus speeds and accuracy in both movie and stills modes. The two cameras offer an intriguing contrast in technologies; the 70D uses Canon’s latest ‘Dual Pixel AF’ on-sensor phase detection, while the A77 employs a separate phase detection AF sensor which receives light via a semi-transparent mirror. Click the link below to see how the two cameras fare in DxOMark’s head-to-head testing. 

As always, it’s best to take such tests as useful data points that tell part of the story, rather than a definitive final word. DxOMark used the Canon EF 40mm f/2.8 STM on the EOS 70D and the Sony DT 16-50mm f/2.8 OSS SSM on the A-77, which employ very different autofocus systems. The former focuses by moving the whole optical unit back-and-forwards with a linear stepper motor, and we remarked on its slightly pedestrian AF speeds in our recent review. Meanwhile the Sony lens uses an ultrasonic-type motor to drive an internal focus element, which is normally a speedier solution. It’s also worth bearing in mind that Canon specifically says it gears the focusing speed down during movie recording to achieve smooth transitions between subjects, as rapid focus jumps can be disconcerting for the viewer.  

DxOMark has also completed its three-part article with lens recommendations for the EOS 70D, covering fully 130 options from both Canon itself, and third-party lens makes such as Sigma, Tamron, Tokina and Samyang. Other tests published last week include a sensor review of Pentax’s budget K-500 SLR, and reviews of two of Canon’s EF-M lenses for its mirrorless EOS M. Click the links below for more:

Canon EOS 70D review gains test scene samples DxOMark tests Canon EOS 70D sensor and lenses User experience: In-depth look at Canon EOS 70D's Dual Pixel AF system Canon EOS 70D preview updated with studio & real-world samples
Canon EOS 70D


AD in KC

What I’d really like to see is an objective, head to head comparison between a Capresso Infinity coffee grinder and a tuna sandwich. I’ve got my own suspicions which is best, but I’d like to see a scientific DxO evaluation before I decide.


They could have at least tested with a halfway decent lens. That 40mm is not a speed demon by all accounts. I have a hard time believing they did not have access to better lenses.


It seems that dedicated PDAF sensors (via mirror and AF chip) is still far better than on-sensor PDAF sensors.

The Nikon 1 series on-sensor PDAF is fast.. but, it uses tiny lenses and its technology has not been proven on larger sensor cameras with larger lenses.


I anyway find it quite striking that most reviews and comparisons so far have completely ignored Sony’s SLTs, preferring to praise this new model over previous Canons or comparable Nikons instead.

Imaging Resource remarkably manages to do its entire review without even once mentioning Sony (!), while CameraLabs only compares to the Nikon 7100, of which nobody ever claimed that its live view AF was stellar.

But after all the hype we wouldn’t want this brand new Canon be made to look bad by a two-year-old Sony, would we now?


Dear Dpreview, i kmow, Amazon sold a lot Cannons, but you have used A77?
In video mode, A77 too “gears the focusing speed down during movie recording to achieve smooth transitions between subjects, as rapid focus jumps can be disconcerting for the viewer”.
You tell of this only about cannon, is incorrect, dear Dpreview……


I used an A57 for a week and enjoyed the 60p video, with great tracking ( the kids running around the pool). It seems like the gap of photo quality and video performance is narrowing, but we can still consider one better suited than the other. I wonder each manufacturer’s patent protection is limiting a convergence of both technologies?

Also…wonderful to see immediate image testing by just announced cameras. Well done indeed.



Personally, I’m having a hard time “Disregarding the better color and exposure accuracy of the Sony SLT-A77.”

The A-77 results just looked so much better to me.


Fixing color issues in videos is a lot more time consuming as you cannot apply the same adjustment to the whole video. Proper exposure, WB, color, etc out of the box is critical for the consumer market.


Well, A77 wins easily, especially in low light where the difference is enormous.

If “slow and steady” is good – than A77 got a “Slow AF” mode that you can turn on in options, without taking your eye off the viewfinder. So… win-win for Sony’s SLT.
Only real issue is AF tracking, though in real life, where subjects don’t run like crazy back and forth – A77 can handle it quite nicely too :).
I guess Sony owners are very happy after seeing that test.

On-sensor PDAF got a VERY long way to go.


“Only real issue is AF tracking, though in real life, where subjects don’t run like crazy back and forth”

Try to shoot kids, or sports, like floor during a gymnastics competition in a poorly lit gym, for example.


I was thinking exactly that.

Sony models (my A700 had this, too) have a ‘slow AF mode’ that allows for slower but more accurate AF, and with the current video options this also provides smoother refocusing. Win/win.


I guess you don’t have little kids then.


The A77 autofocus and tracking are much faster than canon 70D. The better color and exposure accuracy of the Sony SLT-A77 is obvious too. At least for me, the Sony SLT model is a clear winner and more attractive proposition for video users.


One of the most poorly conceived and executed tests I’ve seen from any site and I shoot Canon. This test makes me wonder if the DXO naysayers might have a point.

Dave Luttmann

I presume if they found the opposite in their test, it would proclaimed as confirming Canon’s success.


That test was a joke! Real subject matter would be nice!


AF tests are made on flat, artificial targets for a reason.

KW Phua

Slow and steady. Good for video. Hope one day fast enough to replace the mirror.


The AF comparison test is the most interesting camera test I have seen in quite some time.
Shooting stationary resolution charts is totally useless for the last 10 years or so.


Hmmm… something about that picture….


It’s DPR’s way of testing the astuteness of its followers.


What..all 700 of them.. ? 🙂



Source Article from http://www.dpreview.com/news/2013/09/18/dxomark-tests-canon-eos-70d-dual-pixel-af