[SIZE=3][B]DSLR AF is better but CSC[SIZE=3] AF[SIZE=3] is[/SIZE][/SIZE] faster and more accurate -how can that be?[/B][/SIZE]

For many the key weakness of mirrorless compact system cameras compared to DSLRs is autofocus capability. After reading that you may be surprised to learn that Both Panasonic and Olympus claim their CSCs have the fastest autofocus of any commercial cameras, including DSLRs.

Mirrorless cameras use the main imaging sensor to determine focus using a system called contrast-detection. It's accurate because the sensor itself is the focus measuring device and it's fast because lenses for mirrorless cameras have been highly optimised with low-inertia lens elements and mechanisms to enable fast adjustment of the focus point. Single-action autofocus on static subjects is indeed very fast and blows DSLRs into the weeds. But where DSLRs still reign supreme with their phase-detection systems using separate sensors is in continuous autofocus.

Phase-detection systems can estimate correct focus and even the direction the focus needs to be adjusted right at the start of a focus action, tricks contrast detect AF can't emulate. This also makes it easier to focus on subjects like birds in flight and other subjects that are moving around all the time.

The problem for mirrorless cameras is that you need a mirror to divert the light to be checked for focus to the focus sensor. Olympus currently has a big challenge on its hands because it has promised its Four Thirds DSLR system users a camera that will work without AF compromise with its renowned Four Thirds lenses - which simply don't focus too well on mirrorless Micro Four Thirds camera bodies. A sign of this concern is the unearthing of an interesting patent filed by Olympus last year, ironically on the same day of the Japanese Tsunami disaster. The patent was spotted by the Japanese blog called Egami. You can look at the blog via [URL="http://translate.google.com/translate?hl=en&sl=ja&tl=en&u=http%3A%2F%2Fegami.blog.so-net.ne.jp%2F2012-10-10"]Google Translate[/URL] but our correspondent from Japan, Yoshi, has very kindly translated the page for us, including some of his comments:

[IMG]http://dpnow.com/files/blog/Olympus_variable-sized_microlenses_patent.png[/IMG]

[I]Olympus filed a patent for detecting focal points of a lens "in front of the subject" and "behind the subject"

Olympus has filed a patent that improves the focusing speed of contrast-detect AF. By using micro lenses with different heights A,
B and C as are shown in the diagram, it enables to detect "just-in-focus", "behind the focus" and "in front of the focus".
That in turn makes it possible to have features that until now only phase difference AF could have.
(it must be a little bit slower than the phase difference AF - the writer's personal comment)
When they succeed in establishing the production technology required for placing micro lenses with different height ( on the sensor), we can expect contrast AF with even faster AF speed.

- The patent number: 2012-189874
publication date: 4th Oct., 2012
application date: 11th March, 2011 (Oh, this is the day we had the heavy earthquake, Oly was applying a patent on that date! - my remarks)

- for one pixel each, one micro lens with four light reception areas: RGbGrB
- the distance (or height) between light reception plane and micro lens is differentiated;

* 3 kinds of distance/height
* this gives 3 kinds of focal points - "just in focus", "in front of" and "behind" the subject are detected.

The writer thinks that this patent will not only improves the simple AF speed of contrast detect AF but also improves the capability of focusing fast moving subjects.
This in turn makes it possible to use "normal" 4/3rd lenses without frustrations from slower AF.
It is however not known if they succeeded in developing the actual production technology. Even if they did, the device can only be used in the center of the lens, as the light reception efficiency at peripheral zone of lenses are not ideal.

So (the writer thinks) the normal 4/3rd lenses that can be used on this new sensor will be limited.

The superiority of phase difference detect AF over contrast detect AF is now limited to tracking capability of fast moving subjects.
By using image processing technology, the speed difference between the two AFs will be minimized in future.

So (the writer thinks) the normal 4/3rd lenses that can be used on this new sensor will be limited.

The superiority of phase difference detect AF over contrast detect AF is now limited to tracking capability of fast moving subjects.
By using image processing technology, the speed difference between the two AFs will be minimized in future.[/I]

A big thank you to Yoshi for this translation!

So to summarise the idea is that if you can strategically position three different heights of microlens on the sensor you can pinpoint where the focus is just in front of and behind the correct focus position. This could optimise the performance of contrast detect AF, bringing it closer to the capability of phase detect AF. How close is not yet known and of course this is just a patent - like most patents, the idea may not actually be achievable in a commercially viable form.