Why I switched to an OM-D Locked

Started Feb 11, 2013 | Discussions thread
This thread is locked.
PerL Forum Pro • Posts: 13,959
Re: Has nothing to do with reality

Anders W wrote:

PerL wrote:

Anders W wrote:

PerL wrote:

Anders W wrote:

qianp2k wrote:

I am not sure if worth than a Rebel T4i or T3i however especially if I mount EF lenses thru adapter (so saving in weight/size become irrelavent) that no matter how Canon improve CDAF speed, still will be slower than EF lenses on native EF mount.

Just by looking at the results one can see that this test has very little to do with performance in the field. Panasonic GX1 has better AF-C than Nikon D4? Canons entry level 650D has better AF-C than 7D and 5DIII? That is just a few of many unbelievable results.

My own experience with testing AF-C with CDAF vs PDAF is totally different and so is every practical review I have ever read.

If you have made any systematic tests of AF-C performance, please feel free to make a post carefully describing your methods and results.

The keyword for me is practical tests. When I test AF I do it on real life situations, shooting real sports and fast moving objects. For instance I tested the AF-C of the G2H vs a Nikon D300s by shooting a skier coming against the camera many rounds.

In what way is your test setup significantly different from the one used by the FNAC lab? And in what way is your test superior to the one they performed. As far as I can see, the only significant difference is that yours leaves more room for operator mistakes to affect the outcome.

Same run, G2H typical result, struggling to lock focus at all. Hard to compose last frames because of EVF stutter.

Results for the D300S and the GH2 from the FNAC lab were, when tested a couple of years ago:

D300S: 7

GH2: 16

Source:

http://multimedia.fnac.com/multimedia/editorial/labo/reflex_2011.pdf

In short: Your results are in good agreement with theirs. But current MFT cams have results roughly twice as good as that of the GH2.

Not really agreement, 17 vs 2 in sharp shots.

I have done a lot of camera testing of many different bodies for a magazine, especially in regards to sports and action shooting and AF-C and have a good amount of experience in this. I always tries to use situations that comes close to the actual intended use of the cameras.

See my comment above. What, specifically, is wrong with the realism of the FNAC test from this point of view?

As far as I understand they use a test bench to simulate the results. When the results are all over the place in terms of you what would expect you got to be suspicious how well the simulation works.

Take for instance the car driving at 50 km/h shot at 200 mm eqv from 50 meters and closing. The m43 lens they use (14-140) has wide open a DOF of 40+ meters at 50 meters distance - sharp from 35 m to infinity. Whats the point of that?

The more advanced cameras from Nikon and Canon has lots of settings for the AF-C - release priority, focus priority - hybrid focus/release, color tracking etc. How is that accounted for in the tests?

Also, there is so much involved with good AF-C and tracking than a simple linear test.

These settings are so elaborate that Canon has publiced a pdf-guide for different sports and situations. For instance swimming, where the swimmer disappeares shortly beneath the surface, you want a delay before the camera starts hunt all the way back. And when a camera gets a lock, how well does it keep the lock despite a messy background?



How well does it track off center?



Just because AF-C is complicated I think the only way is practical tests. When we did a head-to-head with the Canon 7D and Nikon D300s in AF-performance we shot 2,000 images of motocross, indoor handball and greyhound-racing, switching between the cameras, counting the number of keepers. I believe that way of testing gives a much better image of the performance in the field than the FNAC - which is discredited because the results. The cheapest Canons doesnt have better AF-C than the semipro models, nice graphics or not.

My view is that if the terrain and the map differs, trust the terrain.



The short comings of CDAF in AF-C is so well known that it should not be a topic of discussion - I am sure you know the technological reasons. (AF-S on the other hand is pretty good with CDAF).

The only thing that is known here is that CDAF struggled with speed generally in the past. But time moves on.

The FNAC tests are the best tests of AF-C performance that I am aware of (which is not the same as saying that I think they are perfect). Let me know if you have other/better sources. And I mean tests, not mere opinions, like those you cite below.

Thom Hogan from his otherwise rather positive OM-D review:

Unfortunately, it all goes downhill from there. If your subject is moving, things change. If it's moving quickly towards or away from you, the E-M5 will be slower than most DSLRs, and depending upon the speed of the subject, far less precise...

Do you have a link to the test where Hogan describes his methods and results?

...Go one more step and try totrack focus with continuous shooting and everything completely falls apart.

Ming Thein in his also rather positive OM-D review:

Even though Olympus claims that continuous and tracking AF is greatly improved with the OM-D, frankly, it’s unusable. Continuous autofocus can’t seem to anticipate subject motion; it drops after the first frame, and usually comes close but fails to re-acquire the subject. Tracking AF is a similar story...

Do you have a link to the test where Thein describes his methods and results?

-- hide signature --
Post (hide subjects) Posted by
(unknown member)
(unknown member)
tko
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow