[pinholecam]

Thanks for sharing this.
This was pretty much what I have seen myself comparing shots from 2 events (wedding and dinner) shot with A700, D90, D7K, D700 and K5.
Despite all the chest thumping from CanNikony on superior AF, I just did not see better AF hit rate from the hundreds of photos taken, which really surprised me.
In fact the K5 had a better hit rate.

[Qwntm]

Thanks for sharing this.
This was pretty much what I have seen myself comparing shots from 2 events (wedding and dinner) shot with A700, D90, D7K, D700 and K5.
Despite all the chest thumping from CanNikony on superior AF, I just did not see better AF hit rate from the hundreds of photos taken, which really surprised me.
In fact the K5 had a better hit rate.

Faster is not better if it's not sharp.

But faster seems to sell more cameras.

People are basically pretty stupid.

[Class A]

Overall, this detailed look as suggested by Class A confirms a slight but significant lead by the K-5 over the big two.

Thanks a lot for doing this analysis, Falk.

Averaging over the non f/1.4 lenses that were used for both K-5 and D7000 yields 92.7% (K-5) vs 88.8% (D7000). The K-5 is still ahead but by no means with the huge gap that the overall table seems to suggest. Questions the approach to use rather different lens sets for each camera, doesn't it?

[froeschle]

in the bar charts, the results of Sigma 50/1.4 and DA*55/1.4 for Pentax seem to be the other way round, i.e., with the Sigma having the better results. So, there seems to be some error in the printed material.

The order of the results is not kept, however the presented data seems imho to be consistent:

Table of results:
best / worst / average [%]
Sigma: 90,5 / 38,5 / 67,3
Pentax: 100 / 62,5 / 83,4

Bar charts:
max / min [%]
Pentax: 100 / about 60
Sigma: about 90 / below 40


The last two sentences in the report state that fast primes with aperture 1.4 and fast zooms with aperture 2.8 gave bad results. They indicate that the small DOF obviously played a role.


If I had to review this paper in a scientific context, I would have asked some serious questions and called for changes - at least considering the conclusions. Nevertheless, if I would be within Pentax marketing, I now would proudly claim in all advertisements that (according to this test) the Pentax K-5 has the best AF performance within its DSLR class.


For us Pentaxians it at least shows that under the tested circumstances Pentax is really competitive. Regarding the differences between contrast and phase AF, I may have to overthink my distaste for mirrorless solutions.

[RonHendriks1966]

Well my K-5 didn't get any better over the past hours. It is funny to read that when you hit 60 % you are good to go. We'r not demanding at all.

If they had also a DA 40mm/f2.8 to the test they had another almost 100% lens in focus. I'm not surprized bey the results on the DA14mm and DA21mm. It is almost impossible to misfocus with a DA14mm.

[dylansalt]

I'm glad my D7000 goes against the grain of the tabled results in the test.

After at least 2000 shots I have taken I can categorically state that I have a hit in focus rate of at least 98%.

From low light to bright, from F1.8 it delivers in focus and razor sharp, time and time again

Lens used 16-85, 70-300, 35 1.8G

I have also shot side by side with K7/K5 users as well in very demanding light conditions (indoor night time rockclimbing comps) where they have consistently battled and I have got the keepers and have my pics being published consistently.

I now take all these "tests" with a pinch of salt.

[asw66]

I'm glad my D7000 goes against the grain of the tabled results in the test.

After at least 2000 shots I have taken I can categorically state that I have a hit in focus rate of at least 98%.

From low light to bright, from F1.8 it delivers in focus and razor sharp, time and time again

Lens used 16-85, 70-300, 35 1.8G

I mentioned earlier in this thread that a buddy of mine has experienced auto-focus problems with his D7000, and has resorted to manual focusing to improve his hit rate. So I'm really curious about this discrepancy, especially if it might result in some insight that helps my friend. Are you using the default settings for the relevant parameters, or have you tweaked something?

[falconeye]

O my .... Let's hope they didn't switch the 7D with the K-5

cf. below. Results weren't swapped, my bad. Sorry.

The order of the results is not kept, however the presented data seems imho to be consistent

You are correct.
Thanks for the double check. I corrected my post above.

They indicate that the small DOF obviously played a role.

If I had to review this paper in a scientific context, I would have asked some serious questions and called for changes

Me too as I wrote above.
They make the same mistake in their SR tests: start from a relative drop in resolution.

You may argue that this is the relevant measure. Because maximum resolution normally is close to the camera's advertized resolution and the tests then measure a camera's ability "to deliver".

But a high pixel resolution (or sharp lens) is good anyway. Less Bayer & AA artefacts, often less noise (per area), higher MTF50 values overall etc. No reason to disadvantage a camera only because it uses a smaller pixel pitch (for a given sensor size) or a sharper lens.

It is about factorization of causes. A standard procedure in science: keep as much parameters constant as possible before drawing conclusions.

Regarding the differences between contrast and phase AF, I may have to overthink my distaste for mirrorless solutions.

Funny. For me, it has an opposite effect:

I am convinced (and always was) about the superiority of the contrast AF method. But I would have guessed that their accuracy lead would have been much clearer.

As it turns out, phase AF with a well calibrated camera/lens combo can outperform contrast AF in accuracy. But contrast AF needs no calibration which makes contrast AF the accuracy winner in the long term.

[Rondec]

I'm glad my D7000 goes against the grain of the tabled results in the test.

After at least 2000 shots I have taken I can categorically state that I have a hit in focus rate of at least 98%.

From low light to bright, from F1.8 it delivers in focus and razor sharp, time and time again

Lens used 16-85, 70-300, 35 1.8G

I have also shot side by side with K7/K5 users as well in very demanding light conditions (indoor night time rockclimbing comps) where they have consistently battled and I have got the keepers and have my pics being published consistently.

I now take all these "tests" with a pinch of salt.

I think your expectations are not nearly as demanding as those in this test. A 98 percent "keeper" rate means that you are accepting some photos where your focus may be adequate, but not perfect.

Not saying that the K5 is really much better than the D7000, but I think they are closer than your comment seems to indicate.

[falconeye]

After at least 2000 shots I have taken I can categorically state that I have a hit in focus rate of at least 98%.
...
Lens used 16-85, 70-300, 35 1.8G

They didn't publish any results for these 3 lenses. Closest maybe is their D7000 Nikkor 70-200/2.8 test. It achieves a consistent performance which is between 80 and 90% (83% average) of possible peak resolution. For all but severe measurbators, 100% of shots would be perceived as "in focus".

I agree to take these tests with a grain of salt. But the remaining solid message is that the K-5 competes well against the big two (for AF.S performance).

Or in other words:

Between the current top APS-C bodies (K-5, D7000, 7D), AF.S performance seems to be more a matter of lens than body.

[philbaum]

I think your expectations are not nearly as demanding as those in this test. A 98 percent "keeper" rate means that you are accepting some photos where your focus may be adequate, but not perfect.

Not saying that the K5 is really much better than the D7000, but I think they are closer than your comment seems to indicate.

Right on - I shot a recent theatre play during rehearsal with a friend who has a D7000, i with the K5. It was my job to go thru the resulting images and pick out the best ones to display in the lobby. I went thru about 400 D7000 images and about 1000 of mine. I didn't see any difference in focus between the K5's and the D7000 images. The selection all came down to composition, color and dramatic impact.

There are now 6 photographers who work with this theatre group, 2 with K5's, 2 with D7000, 1 with D300s, 1 with Canon 50D. They all produce attractive photos.

[pingflood]

Someone hide this thread from garyk

He's recently spent lots of money getting a 7D and very expensive Canon mount super-telephotos due to what he felt were the inadequacies of the K-5's AF.

Can't comment on the K5 since I have never owned or used one (though it seems like a very nice camera!), but have had a 7D for a year now and shoot a lot of birds with it. Went on a trip to a local bird lake a few weeks back and shot probably 800 or so pictures. Coming home I could not find more than 8-10 out of focus / poor focus shots that were not to blame on the photographer. The misses were mostly due to me not dropping the focus point on the bird in flight and landing on the background instead. If I got the center point on the bird so the camera could start tracking it locked onto it and stayed put through just about anything. It is every bit as good as my 1Ds II when it comes to AF performance.

As to the results of this test, I just have one thing to say: if I owned a camera that missed 30-40% of shots like the 7D and D7000 did it would have been returned within a week. I must have gotten lucky. (Or more likely, it is because I use very good lenses. The superzooms might have crap AF performance but that is nothing I have experience with and do not care about since the IQ is such garbage on them anyway.)

[Rondec]

Can't comment on the K5 since I have never owned or used one (though it seems like a very nice camera!), but have had a 7D for a year now and shoot a lot of birds with it. Went on a trip to a local bird lake a few weeks back and shot probably 800 or so pictures. Coming home I could not find more than 8-10 out of focus / poor focus shots that were not to blame on the photographer. The misses were mostly due to me not dropping the focus point on the bird in flight and landing on the background instead. If I got the center point on the bird so the camera could start tracking it locked onto it and stayed put through just about anything. It is every bit as good as my 1Ds II when it comes to AF performance.

As to the results of this test, I just have one thing to say: if I owned a camera that missed 30-40% of shots like the 7D and D7000 did it would have been returned within a week. I must have gotten lucky. (Or more likely, it is because I use very good lenses. The superzooms might have crap AF performance but that is nothing I have experience with and do not care about since the IQ is such garbage on them anyway.)

I think he is using the 7D with a 300mm f2.8 lens from Canon and was using the K5 with a lens of the same specs from Sigma (https://www.pentaxforums.com/forums/pentax-k-5-forum/150973-k-5-vs-7-d-328-lenses.html).

Edit: I should say that Gary really liked the 300 f2.8 L lens and found the 7D adequate for his purposes -- mainly lacking in high iso situations as compared to his K5 that was in for repairs.

[dylansalt]

I think your expectations are not nearly as demanding as those in this test. A 98 percent "keeper" rate means that you are accepting some photos where your focus may be adequate, but not perfect.

Not saying that the K5 is really much better than the D7000, but I think they are closer than your comment seems to indicate.

I'm certainly not denying the focus accuracy of the K5 - I found the accuracy of my K10D more than sufficient.

When viewing on a hd led 24" monitor I can certainly see if they are sharp (D7000) which in reality should be more than adequate for any user - the rest is pure measurebating.

But now looking back at my K10D pics my pics with the NIkon do have more critical sharpness - whether that has to do with the lenses I don't know

I was questioning the massive discrepancy in oof shots with the D7000 in the test

[philbaum]

The human mind is quite amazing. Each one of us every day is assessing almost everything in our environment. Not only do we analyze a lot of things ourselves, but we socialize with others and compare our analysis with those observations that others have made.

Drug testers often go to extremes in trying to eliminate bias from test results, such as using "double blind" observations where not only the patient is unaware of which medication type they are taking, but the administering doctor is also unaware. The "placebo" effect and "self-fulfilling prophecy" are examples of the human mind taking actions to justify a long-held belief.

With so many individual annecdotal accounts (which often conflict), where is the truth on camera AF? While no study is "perfect", this German testing counts for something because they worked with an independent testing organization, involved a lot of camera equipment, and it isn't just one individual speaking through their own bias (and we all have biases)

They performed about 1300 measurements with 57 lenses on nine cameras. 10 shots using AF were taken in
wide/standard/tele position with each lens.
Another picture using manual focus with optimal resolution is the reference.
The ratio of the resolutions of the pictures then serves as a measure.
Lenses used on the cameras:

Subject to the limits of how AF was tested, the very least that can be said from this study is that Pentax K5 AF is competitive with top brands. (This study didn't test predictive tracking or low light focusing, as far what others have said)