Previous news story    Next news story

Video tests of the Canon EOS 650D / Rebel T4i's Hybrid autofocus system

By dpreview staff on Aug 15, 2012 at 19:39 GMT

What improvements has Canon's Hybrid AF system brought to the EOS 650D's usability in live view, and what might this mean for the forthcoming EOS-M mirrorless camera? As a precursor to our imminent 650D/Rebel T4i review, we've published two videos showing how Hybrid AF works, compared both to conventional phase-detection AF and to a contemporary mirrorless rival (in this case the Panasonic DMC-G5). It's a chance see how the 650D performs but also gives an idea of what we can expect from the EOS-M, which uses the same technologies.

For these videos we've used a Canon EOS Rebel T4i with the EF-S 18-135mm f/3.5-5.6 IS STM lens designed specifically for use with the camera's Hybrid AF system. A further explanation of the system can be found by clicking here, but the basic idea is that phase-detection elements on the main sensor help determine the distance of the subject and contrast detection then performs a focus fine-tune, at that distance. The 18-135mm STM's lens' design features a light, internal focus group that can be quickly accelerated and decelerated to suit this autofocus behavior, and uses a stepping motor to allow fast, quiet and precise movement. The EOS-M uses the same technologies and similar designs to offer the same functions.

So, how do they perform?

Canon EOS 650D - Quick AF (phase detection) and Hybrid AF compared

Canon EOS 650D and Panasonic DMC-G5 compared

424
I own it
60
I want it
29
I had it
Discuss in the forums

Comments

Total comments: 243
12
riveredger
By riveredger (Aug 15, 2012)

You should include a Sony SLT model when comparing Live View and/or video AF. They are amazingly responsive and fast focusing in Live View AND can continually AF well in live view, unlike other cameras.

Comment edited 1 minute after posting
1 upvote
Richard Butler
By Richard Butler (Aug 15, 2012)

If you look further down the comments, you'll see that we're already planning to do exactly that.

4 upvotes
riveredger
By riveredger (Aug 15, 2012)

Thanks, Richard.

0 upvotes
ET2
By ET2 (Aug 16, 2012)

I would like to see either the new 18-135mm SAM or 18-50mm SSM for that test

0 upvotes
justmeMN
By justmeMN (Aug 15, 2012)

That doesn't look very promising for the EOS M's AF speed. :-(

Oh well, companies rarely get things right the first time. Perhaps the 2013 version will be significantly faster.

0 upvotes
random78
By random78 (Aug 15, 2012)

True but it is still very surprising given how much time canon has had looking at the market. Panasonic G1 was the first m43 camera but it still was much faster AF than what we see here for 650D, despite being a CDAF only system. And then we have Nikon 1 with its hybrid system. So if EOS M has similar AF performance than it would be puzzling. If it was a company like Pentax then sure I could understand. But for a company like Canon with so much technical expertise and resources, it is fairly surprising.

Comment edited 47 seconds after posting
5 upvotes
Gesture
By Gesture (Aug 15, 2012)

Wish these were longer demonstrations, including standing in front of some real scenes, plus, as someone suggested video. Also, the issue could be speed as related to accuracy. Is one mode faster or faster and more accurate.

0 upvotes
Jun2
By Jun2 (Aug 15, 2012)

What is the result of G5 vs 650D phase?

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

I'd expect the G5 to be faster at single-acquisition AF and the 650D to be better for continuous AF.

2 upvotes
photo nuts
By photo nuts (Aug 15, 2012)

You'll notice both the 650D and G5 take about the same amount of time to recover to the original screen after AF is activated, with the G5 being slightly faster for the center subject.

2 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

Yes, but you can get the G5 to shoot again before it has re-set to the original screen (it can shoot while writing).

8 upvotes
photo nuts
By photo nuts (Aug 15, 2012)

Good point. Looks like Canon's hybrid AF is not quite there yet.

2 upvotes
OniMirage
By OniMirage (Aug 15, 2012)

The g5 was near instant in the tests, the 650D took about 4-5 seconds between each shot in hybrid. In the 650D phase vs hybrid the phase detect took just under a second, slower still than the G5.

2 upvotes
spoorthy
By spoorthy (Aug 15, 2012)

I have heard that the liveview autofocus on the pentax k-30 is extremely fast. You guys should do a comparision

4 upvotes
raimaster
By raimaster (Aug 15, 2012)

+ 1 ... and let see whether hybrid AF is only marketing point?

1 upvote
Freestyler
By Freestyler (Aug 15, 2012)

I can't quite work out why the K-30 wasn't reviewed first, as it was announced a month earlier, was previewed earlier and unlike the T4i, hasn't been recalled.

But am also interested in how they compare on video.

3 upvotes
fisherman_lol
By fisherman_lol (Aug 16, 2012)

What did you expect. this is DPR and K-30 is not a Canon nor Nikon.

2 upvotes
MarkInSF
By MarkInSF (Aug 16, 2012)

Maybe because the Pentax is an evolution of existing technologies and the Canon something new, at least to Canon. Not to mention that it will sell in massively higher numbers. Oh, and that the technology is also to be used in the EOS-D, a camera many are interested in. Oh, wait, they actually mentioned that last one. Silly me to bring it up again.

0 upvotes
Alizarine
By Alizarine (Aug 16, 2012)

I agree with the above. This post is anyway about a "new" type of AF system so that would make it "more interesting" anyway, compared to the "huge improvement" Pentax did with their SAFOX IX AF system.

BTW, the K-30 is mid-level, not entry-level...

0 upvotes
fisherman_lol
By fisherman_lol (Aug 16, 2012)

What ever you call it, still kicks a$$ for less price.

Comment edited 3 minutes after posting
0 upvotes
revio
By revio (Aug 16, 2012)

Yes, of course it´s more interesting for us buyers that this or that camera actually is faster or better in xxx factors, rathre than we would care for with what technology this preformance is achieved.

@MarkInSF:
That a certain camera will sell more than another, better, camera, is no reason to cover it more in tests/news etc. If a camera actually have a performance advantage in fex Liveview focusing, like the K30, then naturally THAT would be covered, as it is more in the interests of the potential camera buyer.

OK, Canon is way bigger than Pentax, but if Pentax do have a superb product why not see that as a value in itself? Now, knowing K30 have better liveview focusing than EOS 650D then that is more interesting than that the 650D uses some new tech. Seemingly that new tech is outperformed by some supposedly "old tech"....talk about matters of priorities...

0 upvotes
OneGuy
By OneGuy (Aug 15, 2012)

In the second focus shootout the winner Panasonic is not even shown. I bet the G5 is also better looking.

1 upvote
zonoskar
By zonoskar (Aug 15, 2012)

I would be interested in Hybrid AF performance with a fast USM lens like the 17-40L.

1 upvote
ET2
By ET2 (Aug 15, 2012)

Probably worse with USM lens. That's the reason Canon came up wit STM.

2 upvotes
ageha
By ageha (Aug 16, 2012)

Yes, it should be even slower!

0 upvotes
balios
By balios (Aug 16, 2012)

Canon came up with STM for smooth/silent focus during video, not for improved focus speed. Traditional USM lenses are fast but jerky, which isn't good for video. I'd also like to see the results with USM lenses.

3 upvotes
zonoskar
By zonoskar (Aug 16, 2012)

@balio: That was my understanding too. USM lenses are designed to be used with phase-detect AF as this is what is used in normal DSLRs. So this hybrid AF should be able to use USM lenses, but it might not be as smooth to a get focus-lock. And for normal photography, this is not a problem. Taking 2 seconds to focus is...

0 upvotes
Ruy Penalva
By Ruy Penalva (Aug 15, 2012)

I think a better comparison would made in video mode or moving stills.

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

That's something we'll be doing separately. Stay tuned.

2 upvotes
mfj197
By mfj197 (Aug 15, 2012)

On the contrary, I would expect the majority of auto focus usage around the world in these sorts of cameras is for single shot AF. That is exactly the scenario assessed here.

0 upvotes
CNY_AP
By CNY_AP (Aug 15, 2012)

People use the viewfinder for shooting stills on a SLR, which is instantly fast, not the LCD. If I understand the videos correctly, he used the LCD, which required the mirror to flip a couple of times?

0 upvotes
Richard Butler
By Richard Butler (Aug 16, 2012)

The comparison between the 650D and the G5 is shot in live view, using the LCD and only requires the mirror to be flipped down when the photo is being taken.

The other video shows 'Quick AF,' which is viewfinder AF with one additional mirror drop. It is near instant (though the G5 and Olympus OM-D will match or beat it when using kit zooms).

0 upvotes
Ruy Penalva
By Ruy Penalva (Aug 16, 2012)

Does the smaller G5 sensor play a role in that focus readiness advantage? Could the combination of phase and contrast detected be responsible for that lag since it primarily aim at video capabilities than at still and is a plus in the still mode? A comparison with Canon new mirrorless should clarify part of this.

0 upvotes
eye999
By eye999 (Aug 15, 2012)

can any 1 propose which camera is better canon eos 650d or 60d? dpreview.com when you gona publish the full review of 650d ?

0 upvotes
mpgxsvcd
By mpgxsvcd (Aug 15, 2012)

Can I just say great job dpreview. You guys are bringing us the content we want to see. Thank you.

16 upvotes
mpgxsvcd
By mpgxsvcd (Aug 15, 2012)

Great comparison. Do you have any videos showing the difference when they record movies instead of stills? Great job. Keep up the great work.

6 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

That's part of an article coming in the next few weeks.

6 upvotes
David Naylor
By David Naylor (Aug 15, 2012)

It would be interesting to see how the movie/live view AF compares between the 650D and the 600D. Does the new "hybrid" AF make any difference?

0 upvotes
facedodge
By facedodge (Aug 15, 2012)

Make sure to compare it to contract detection on a T3i in the final version of this review... That is where the improvement is made. You've got apples and oranges here.

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

It's an improvement, but it's hard to imagine it being slower than the T3i performing CDAF.

Since the G5 and T4i are targeted at similar buyers, I'd disagree that this is an apples to oranges comparison. At worst it's a green apples to red apples comparison (they appeal to slightly different tastes, but your expectations of them aren't wildly dissimilar).

11 upvotes
facedodge
By facedodge (Aug 15, 2012)

You've replied to almost every post. Well done.

We appreciate the work you guys do. Keep it up!

8 upvotes
random78
By random78 (Aug 15, 2012)

I totally agree with Richard that comparison to G5 makes a lot of sense sense. I am not that crazy about comparison with PDAF. In my mind hybrid AF is a technology to improve upon CDAF-only systems so it will be more interesting to see how much improvement we have over the CDAF in 600D. But still comparison with PDAF is interesting and thanks for the work that you put in!

0 upvotes
Alizarine
By Alizarine (Aug 16, 2012)

The AF systems are comparable.. but the cameras...hmm well as I've never tried m4/3s before, maybe I should.

0 upvotes
Ross the Fidller
By Ross the Fidller (Aug 16, 2012)

If you want small & fast AF, then yes you should try a M4/3's camera. The AF with the kit lens, 12-50mm on the Olympus OM-D E-M5 is instant but when regular 4/3's lenses are attached, then they are slower to focus.

Comment edited 32 seconds after posting
0 upvotes
mfj197
By mfj197 (Aug 15, 2012)

What lens was used on the G5

1 upvote
Richard Butler
By Richard Butler (Aug 15, 2012)

The Lumix Vario X 14-42mm retractable kit lens - a relatively comparable step-up kit lens (the standard 14-42mm is similarly fast).

1 upvote
mfj197
By mfj197 (Aug 15, 2012)

Thanks Richard. It's certainly a fast focusing lens too, by that demonstration. Good comparison - thanks for putting it up.

1 upvote
Eleson
By Eleson (Aug 15, 2012)

Why not compare with any sony slt also.
Publishing a separate article with a "test" of "Canon Quick AF" makes it look like it is something new and revolutionary.
Nonetheless, this is a really great reward for being the last into the game. :)

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

As it says in the text, this is part of the forthcoming Canon review, it's not intended as a comprehensive look at all available focus systems.

You may be interested to know that we shot video from several different forms of AF (including the SLT's phase-detection) yesterday and hope to publish a full article on that shortly.

10 upvotes
random78
By random78 (Aug 15, 2012)

Thanks! Looking forward to that

0 upvotes
Francis Carver
By Francis Carver (Aug 15, 2012)

Hybrid AF, wow!!! Finally, Canon is experimenting with cutting-edge technology that others (Sony, Panasonic) have been using for a couple of years now. :-))

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

Sony and Panasonic?

We've seen on-sensor PDAF + CDAF from Fujifilm and Canon but I don't remember any Panasonic or Sony implementations. Ricoh also made a Hybrid AF system, though that used a secondary sensor.

6 upvotes
Andy Crowe
By Andy Crowe (Aug 15, 2012)

And the Nikon 1, don't forget :)

4 upvotes
chlamchowder
By chlamchowder (Aug 16, 2012)

By hybrid, he might be talking about the secondary sensor implementations used by Olympus, Sony, and Panasonic for a while.

But to answer that, the secondary sensor approach is completely different - you're still using the main PDAF sensors, and subject to their limitations (mainly accuracy problems)

0 upvotes
fastlass
By fastlass (Aug 15, 2012)

it is obvious that both the testing methods and the review products are flawed. But after correcting for these errors, I believe the same results will be produced.

2 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

Would you care to point out these 'obvious' flaws?

This behaviour is consistent with our real-world usage (with two independently-sourced units), or we wouldn't have included it.

Comment edited 3 minutes after posting
10 upvotes
cs hauser
By cs hauser (Aug 15, 2012)

@ R Butler - The shallower depth of field will result in slower CDAF. That is the problem with the test.

In the video shown above, the 650D had significantly shallower depth of field than the G5 because (1) subject magnification was much bigger, and (2) aperture used in exposure was bigger, and (3) sensor is bigger.

For a proper comparison of CDAF, depth of field should remain constant. That means keeping the same subject magnification, and it also means using a faster lens on the mFT. It's NOT enough to stop down the mFT lens, because auto focus is conducted with the lens wide open, regardless of exposure settings. A proper test would have the 650D with 18-135 STM @ 18mm (f/3.5) compared to the G5 with Panasonic 14mm (f/2.5) pancake.

That would yield near-identical magnification and depth of field.

Comment edited 4 minutes after posting
4 upvotes
Richard Butler
By Richard Butler (Aug 16, 2012)

In principle, yes, but it doesn't make a huge difference - these videos are indicative of the performance you will experience when you shoot with the cameras.

I haven't tested that exact combination, but the 14mm is extremely fast at focusing, while the 18-135mm isn't. Also, to give you some idea of how much difference it might make, we found a less than 1/30th of a second difference between using the 18-135mm and the 40mm F2.8, despite the pancake's larger aperture.

3 upvotes
Jman13
By Jman13 (Aug 16, 2012)

Trust me, it doesn't make a difference. Use the G5 or Olympus OM-D with the Olympus 75mm f/1.8 and you'll have much shallower depth of field on the micro 4/3 camera, and it will still destroy it. Why? The 75/1.8 focuses faster than the 14-42.

The new Olympus and Panasonic bodies focus faster than essentially any DSLR in single shot. I know it absolutely trounces my 1Ds Mark II, and I can't imagine that the improvements in more recent bodies are enough to make up the difference, especially in lower light where the difference is absolutely enormous. In good light with a fast USM lens, they are very close.

0 upvotes
Baczek
By Baczek (Aug 15, 2012)

what this video doesn't show is focus accuracy. i don't care how fast PDAF focuses if the picture comes out all soft. hybrid AF is MUCH more accurate with older lenses like the 50 1.8.

4 upvotes
PeterTom
By PeterTom (Aug 15, 2012)

That's exactly my point, too.

I would be glad to see a comparison of accuracy PDAF vs. CDAF in low light.

2 upvotes
WT21
By WT21 (Aug 15, 2012)

CDAF will still yield more accurate focusing than PDAF in low light IF it gets focus. CDAF is not prone to front or back focusing. But, CDAF can struggle in low light. Then again, Canon Rebels have been terrible in low light, and for years they didn't (and perhaps still don't?) come with a simple AF assist light!! When both use AF assist lights, CDAF will still be more accurate with center point focus. CAF is the only place where PDAF rules now.

1 upvote
random78
By random78 (Aug 15, 2012)

Right, but I think the important comparison here is with G5 and with 600D CDAF (which they didn't show). Hybrid AF is expected to be an improvement over CDAF-only systems because it can use the PDAF to quickly get in the zone and then use CDAF for final accuracy check - so basically best of both worlds. Right now it seems somewhat faster than the 600D CDAF but way behind the speed of dedicated CDAF cameras like G5. And G5 being a CDAF only system would also be at least as accurate as 650D. 650D is tested with a lens specifically designed for this technology so we can no longer blame it on PDAF-based lenses not working well for CDAF. If the EOS-M has similar performance than it will be fairly disappointing considering that other mirrorless cameras are doing much better using only CDAF. The video AF is of course an unknown right now.

0 upvotes
Ross the Fidller
By Ross the Fidller (Aug 16, 2012)

As a 4/3's & now also M4/3's user, I would like the concept of on sensor PD-AF/Hybrid to use with existing 4/3's lenses, but the newer M4/3's lenses that are optimised with internal focussing for CD-AF, are very fast & that is the challege for Canon to meet with. If they can achieve a fast Hybrid AF, then they will do well, but they haven't normally been quick to embrace features from other manufacturers (especially if it started with Olympus). It seems they do so with reluctance.

Comment edited 2 times, last edit 1 minute after posting
0 upvotes
AluKd
By AluKd (Aug 15, 2012)

Let's see how they perform: slow as molasses.

Comment edited 33 seconds after posting
5 upvotes
jj74e
By jj74e (Aug 15, 2012)

so....basically the hybrid af system does nothing and is just a marketing point (for single shot, non-continuous AF anyway)

Comment edited 54 seconds after posting
8 upvotes
chlamchowder
By chlamchowder (Aug 15, 2012)

That would also be a very interesting comparison: Does the hybrid system offer an improvement in single shot live view AF speeds when compared to older cameras like the T2i?

And how does the hybrid system compare to Nikon and Pentax's CDAF speeds? It would be nice if Canon's system is a little faster than current DSLR CDAF technology, while retaining the same degree of accuracy.

2 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

The 650D is certainly faster than the CDAF system on the 600D. We've not compared against a Nikon DSLR setup, but they'll be similar - a long way off the pace of a designed-for-CDAF mirrorless lens/body combination.

We'll be looking at its continuous AF and movie performance soon.

Comment edited 6 minutes after posting
6 upvotes
raimaster
By raimaster (Aug 15, 2012)

how about comparing against CDAF on Pentax K-30? it must be interesting since both camera have same target? anyway we still waiting full review on this camera (k30) as DPR promised

0 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

This is from the forthcoming 650D review, so we compared it to one of its peers we knew to do this well (to show how it compared to the level of performance that you can get for the same money).

If we include such a video in the K-30 review (and I'm not going to promise anything), it would probably also be against the G5.

3 upvotes
davids8560
By davids8560 (Aug 15, 2012)

Hmmm. Point well taken!

1 upvote
chlamchowder
By chlamchowder (Aug 15, 2012)

G5's CDAF looks really, really fast - about as fast as traditional PDAF systems. Of course, the 4/3 sensor does give more DOF with the same aperture and field of view.

What's more interesting to me, though, is how the systems compare when dealing with moving subjects.

3 upvotes
Richard Butler
By Richard Butler (Aug 15, 2012)

We'll be covering that in an article in the coming weeks.

5 upvotes
BJN
By BJN (Aug 15, 2012)

More depth of field would actually mean less contrast difference for the G5 to work with when seeking accurate focus. Unless you're suggesting that Panasonic's speed comes from intentionally delivering only "ballpark" accuracy. That would be a new tactic and I doubt it would be accepted in pixel peeping forum discussions.

0 upvotes
CNY_AP
By CNY_AP (Aug 15, 2012)

Having a greater DOF hides focusing errors. If the focus is off by 1 foot, but your DOF is 10 ft, you will never be able to notice the error.

1 upvote
random78
By random78 (Aug 16, 2012)

Do you realize that the DOF difference between Canon and m43 is only 1.2x? So if the DOF for canon shot is 1ft than the DOF for equivalent m43 shot would be 1.2ft. Hardly enough to make any meaningful difference in AF performance.

0 upvotes
Jman13
By Jman13 (Aug 16, 2012)

And yet, the G5 and other m4/3 cams manage to have dead on accurate focus with any lens, even the 45 Macro at macro distances where DOF is 1mm or so, or with fast lenses like the 45/1.8 and 75/1.8...and they are more accurate and still nearly instant to focus. Go handle a G5, GX1 or OM-D with a fast lens and tell me that it's not faster and more accurate than a DSLR's phase detect.

Now, continuous focus is a different story. They've gotten better with recent bodies on CDAF continuous focus, but it's still nowhere near what a good PDAF system will do.

0 upvotes
Total comments: 243
12