«

I'll believe it when I see it.

http://www.43rumors.com/ft5-olympus-...slr-autofocus/

On Jun 15, 8:31*am, RichA <rander3...@gmail.com> wrote:
> I'll believe it when I see it.
>
> http://www.43rumors.com/ft5-olympus-...stem-pen-autof...

It sounds like they've changed something in the lenses
to help. If it's just a way to drive the elements around
more efficiently, it could also be applied to phase-
detect AF, and would presumably give a similar
improvement in performance.

I hope it's a lot more than that; re-buying lenses isn't
high on my list of pleasant ways to spend my money.

RichA <> wrote:
>I'll believe it when I see it.
>http://www.43rumors.com/ft5-olympus-...slr-autofocus/


Olympus hasn't "invented" anything here. The new AF system is the
same one that Panasonic and Ricoh have licensed. It represents a
major step change in the speed of contrast-detect AF systems.

The Panasonic GH2 and G3 have very fast and accurate AF when used on
stationary or slow-moving subjects. We have found that their AF is
less decisive when faced with subjects that move more quickly, such as
vehicles and children. For fast moving subjects, a DSLR is still a
better choice.

However, one thing that the new contrast-detect system offers is
outstanding accuracy. Even the top pro DSLRs cannot beat it.

On Jun 15, 12:10*pm, bugbear <bugbear@trim_papermule.co.uk_trim>
wrote:
> Bruce wrote:
>
> > However, one thing that the new contrast-detect system offers is
> > outstanding accuracy. *Even the top pro DSLRs cannot beat it.
>
> Surely it's unbeatable for accuracy, since it's literally
> checking the focus at the sensor that will record
> the final image. All other systems (AFAIK) need
> to keep the focus system "in sync" with the final
> image sensor, which must always carry some error,
> no matter how small.
>
> I would welcome knowledge one way or the other.
>
> * BugBear

The more plasticy the camera, the worse the phase focus accuracy
because thermal expansion and contraction of the housings causes
shifts in the focus distance from the lens to the sensor used for
focusing. Mirrors and phase focusing will be gone soon.

bugbear <bugbear@trim_papermule.co.uk_trim> wrote:
>Bruce wrote:
>> However, one thing that the new contrast-detect system offers is
>> outstanding accuracy. Even the top pro DSLRs cannot beat it.
>>
>
>Surely it's unbeatable for accuracy, since it's literally
>checking the focus at the sensor that will record
>the final image. All other systems (AFAIK) need
>to keep the focus system "in sync" with the final
>image sensor, which must always carry some error,
>no matter how small.
>
>I would welcome knowledge one way or the other.


In theory, a contrast-detect AF system should always give pinpoint
focus because, as you say, it is making its measurements on the sensor
itself. But in practice, previous contrast-detect AF systems have
been slow and hesitant, and the result is that they failed to focus on
a moving subject.

The new system is much faster, meaning that in most situations,
results are much more decisive. But the system still falls short when
the subject is moving quickly.

Phase-detect systems tend to be quicker because they can sample more
quickly. Therefore there is less scope for error. However, you are
quite right to say that phase-detect systems include systematic
errors, some more than others. Canon's system appears to cause more
problems than Nikon's, but with careful calibration they will both
produce similar results.

So to sum up, contrast-detect systems have the potential for zero
error but are still not as quick and decisive as phase detect systems,
although the latest improvements close the gap or can even eliminate
it for static and slow moving subjects. Phase-detect systems are fast
and decisive but may include systematic errors that can only be
eliminated by careful calibration of body and lens together.

Personally, I'm still a fan of manual focus.

On Jun 15, 11:10*am, bugbear <bugbear@trim_papermule.co.uk_trim>
wrote:
> Bruce wrote:
>
> > However, one thing that the new contrast-detect system offers is
> > outstanding accuracy. *Even the top pro DSLRs cannot beat it.
>
> Surely it's unbeatable for accuracy, since it's literally
> checking the focus at the sensor that will record
> the final image. All other systems (AFAIK) need
> to keep the focus system "in sync" with the final
> image sensor, which must always carry some error,
> no matter how small.
>
> I would welcome knowledge one way or the other.

Matches my understanding -- and experiments.

When focus is critical and time allows (which covers a few
real photographic situations, and all lens-testing
situations), I now try to use Live View in tripod mode on
my D700, which uses contrast-based AF off the primary
sensor, to set the focus.

The very fast phase-detect AF in the D700 is wonderful for
fast-moving subjects, even in quite low light, and I
wouldn't give it up for anything I've seen from
contrast-based AF yet (and "yet" is the important word
there, I think). But absolutely, it's a more complex
system, can get out of alignment, requires a mirror system
and the corresponding thicker body, etc.

John A. <> wrote:
>On Wed, 15 Jun 2011 17:48:32 +0100, Bruce <>
>wrote:
>>bugbear <bugbear@trim_papermule.co.uk_trim> wrote:
>>>Bruce wrote:
>>>> However, one thing that the new contrast-detect system offers is
>>>> outstanding accuracy. Even the top pro DSLRs cannot beat it.
>>>>
>>>
>>>Surely it's unbeatable for accuracy, since it's literally
>>>checking the focus at the sensor that will record
>>>the final image. All other systems (AFAIK) need
>>>to keep the focus system "in sync" with the final
>>>image sensor, which must always carry some error,
>>>no matter how small.
>>>
>>>I would welcome knowledge one way or the other.
>>
>>
>>In theory, a contrast-detect AF system should always give pinpoint
>>focus because, as you say, it is making its measurements on the sensor
>>itself. But in practice, previous contrast-detect AF systems have
>>been slow and hesitant, and the result is that they failed to focus on
>>a moving subject.
>>
>>The new system is much faster, meaning that in most situations,
>>results are much more decisive. But the system still falls short when
>>the subject is moving quickly.
>>
>>Phase-detect systems tend to be quicker because they can sample more
>>quickly. Therefore there is less scope for error. However, you are
>>quite right to say that phase-detect systems include systematic
>>errors, some more than others. Canon's system appears to cause more
>>problems than Nikon's, but with careful calibration they will both
>>produce similar results.
>>
>>So to sum up, contrast-detect systems have the potential for zero
>>error but are still not as quick and decisive as phase detect systems,
>>although the latest improvements close the gap or can even eliminate
>>it for static and slow moving subjects. Phase-detect systems are fast
>>and decisive but may include systematic errors that can only be
>>eliminated by careful calibration of body and lens together.
>>
>>Personally, I'm still a fan of manual focus.
>
>And of course all systems are subject to the limits of how quickly and
>precisely you can move the glass.


Indeed. Going back a few years, that's why Canon's EOS lenses were so
much faster to focus than "screwdriver drive" AF Nikkors, which led
Nikon to pay for a very expensive license to use the USM technology.

Nikon called it "Silent Wave Motor" rather than "Ultra Sonic Motor" as
if it was their own technology, but it was Canon's. At some expense
it transformed the focusing performance and the new AF-S Nikkors were
every bit the equal of Canon USM lenses.

Indeed, they were actually better, because Nikon did not share (at
least, not to anything like the same extent) the systematic focusing
errors that the EOS system suffered from.

In recent years I think the patents may have expired because Nikon is
putting silent wave motors in lenses across the range, including kit
zooms. They would not be doing that if royalties had remained at the
previous level.

Doug Bashford <> wrote:
>Re: Olympus about to surpass DSLR focus speed/accuracy?;
> On Wed, 15 Jun 2011, Bruce wrote:
>
>> In recent years I think the patents may have expired because Nikon is
>> putting silent wave motors in lenses across the range, including kit
>> zooms. They would not be doing that if royalties had remained at the
>> previous level.
>
>
>As usual, when patents and copyrights expire,
>the consumer and the economy wins!


True, but without laws that protected intellectual property and
allowed inventors and developers to profit from their work, a lot of
technological development may never have happened.