«

Is it all due to connector differences, battery power, or what? Or, is
it possible Olympus has done it, in order to keep people who already
own 4/3rd lenses from using them, forcing the purchase of m4/3rd
lenses?

http://forums.dpreview.com/forums/re...ssage=40734721

On 27/02/2012 6:05 p.m., RichA wrote:
> Is it all due to connector differences, battery power, or what? Or, is
> it possible Olympus has done it, in order to keep people who already
> own 4/3rd lenses from using them, forcing the purchase of m4/3rd
> lenses?
>
> http://forums.dpreview.com/forums/re...ssage=40734721
>
Perhaps there isn't any conspiracy, and perhaps it was already explained
in an interview with a Panasonic executive a year or two ago.
I haven't looked for the link, but IIRC his explanation was that they
designed lenses for PD AF from the ground up with low mass moving
elements for focus, with focus motors optimised for multiple fast
incremental movements, as needed for fast CDAF.
Which is /possibly/ why Canon and Nikon have seemed to be tardy with
implementation of MILC in APS-c and FX formats. Perhaps Nikon are
closer, as the "1" series hybrid AF system might work with their
existing lens ranges in PD mode.
Will be interesting when the big two do make a move, whether they use
existing mounts (with extension tube) or new mounts.

RichA <> wrote:

>Is it all due to connector differences, battery power, or what?


The official explanation is that 4/3 lenses are designed for use with
the phase detect AF systems of E-System DSLRs whereas m4/3 cameras and
lenses use contrast detect AF.


>Or, is
>it possible Olympus has done it, in order to keep people who already
>own 4/3rd lenses from using them, forcing the purchase of m4/3rd
>lenses?


No, there isn't a conspiracy. You're being paranoid as usual.


>http://forums.dpreview.com/forums/re...ssage=40734721


The article linked in that posting would have told you what I wrote
above.

On Feb 27, 7:15*am, Bruce <docnews2...@gmail.com> wrote:
> RichA <rander3...@gmail.com> wrote:
> >Is it all due to connector differences, battery power, or what?
>
> The official explanation is that 4/3 lenses are designed for use with
> the phase detect AF systems of E-System DSLRs whereas m4/3 cameras and
> lenses use contrast detect AF.

That really doesn't specifically answer why the lenses are so
sluggish. Since you mention it, it appears they didn't implement
circuitry in the new cameras that can distinguish between the phase
and contrast focusing "features" of the lenses. This omission is due
to one of two thing: They didn't have room or money for the circuitry
in the new cameras, or they left it off on purpose, as they have focus
confirmation with non m4/3rd lenses, a feature Nikon has had since (I
believe) the D1X/D100.

RichA <> wrote:
>On Feb 27, 7:15*am, Bruce <docnews2...@gmail.com> wrote:
>> RichA <rander3...@gmail.com> wrote:
>> >Is it all due to connector differences, battery power, or what?
>>
>> The official explanation is that 4/3 lenses are designed for use with
>> the phase detect AF systems of E-System DSLRs whereas m4/3 cameras and
>> lenses use contrast detect AF.
>
>That really doesn't specifically answer why the lenses are so
>sluggish.


I didn't say it did. I said that was the official explanation. In
practice, the contrast detect AF system works exceptionally well with
lenses that are designed for it.

However, when you use a 4/3 lens that was designed for phase detect AF
with a m4/3 body that has only contrast detect AF, the AF system
hunts, often wildly, and can take as much several seconds to reach
correct focus. Typically it takes about a second, which is really
slow.

The problem seems to be that, when faced with an out of focus image
from a lens that isn't optimised for contrast detect, the camera's AF
system isn't able to decide which way to go to obtain correct focus.
So it spends valuable time using trial and error instead - hence the
hunting.

Somehow, on the other hand, a lens designed for contrast detect will
work together with the AF system to deliver near-instantaneous focus,
because it not only knows which way to go but can tell approximately
how far it needs to go. Even while it is moving towards correct
focus, it is checking and re-checking the rate of increase of contrast
and re-evaluating how far to go. All this seems to happen in the
blink of an eye, because the AF systems on the latest Panasonic bodies
are unbelievably fast and accurate.

I don't pretend to know the detail of how the electronics in the lens
and camera body work together. All I know is that it works, and at a
speed I have never previously experienced. The G3, GH2 and GX1 all
have this super fast AF, as does the Olympus E-P3 and possibly the
E-PL3 and E-PM1, but I'm not sure about the last two.

Now add a lens that doesn't have the necessary CPU, such as one from a
4/3 DSLR, and you are back to good old-fashioned trial and error,
which takes time. Typically a second, sometimes less, sometimes a lot
more. And that's why people are frustrated and dissatisfied.

But put a 4/3 lens on a 4/3 DSLR and it focuses quickly and accurately
using phase detect AF.


>Since you mention it, it appears they didn't implement
>circuitry in the new cameras that can distinguish between the phase
>and contrast focusing "features" of the lenses. This omission is due
>to one of two thing: They didn't have room or money for the circuitry
>in the new cameras, or they left it off on purpose, as they have focus
>confirmation with non m4/3rd lenses, a feature Nikon has had since (I
>believe) the D1X/D100.


If you are going to put phase detect circuitry in to the cameras, you
need phase detect sensors. Olympus and Panasonic won't be doing that,
as it will add complication and cost that customers won't pay for.
There is also the issue of royalty payments for in-sensor phase detect
AF as used in the Nikon 1 System.

I don't pretend to be an expert here. I am a very satisfied user of
m4/3 and don't have any 4/3 lenses. I am told that we haven't had any
complaints from customers - they seem happy to be able to use their
4/3 lenses on their new m4/3 even if the AF is slow. So I think it is
a problem that exists more online than in the real world, with people
getting aerated about theoretical issues that don't have quite such an
impact as they think on real life shooters.

We have seen this phenomenon on newsgroups for years. People get
extremely worked up about theoretical problems that are barely noticed
in the real world. They obsess about things like the Nyquist
frequency and MTF that may be of some relevance to designers of photo
equipment but have far less relevance (often zero) to people who shoot
images. Unfortunately, they cannot see the wood for the trees, and
translating their claimed knowledge of the technology into capturing
images that please, entertain and move people seems far beyond their
abilities.

Remember when the Hubble telescope was first lauched? It has a very
fine primary mirror, the most accurate figure and polish ever, but it
was the wrong figure so it was spherically aberrated. Then they built
the "Costar" which was an add-on corrective optics set that fixed the
problem. What is surprising is that they literally can't work with
the camera electronics to "adapt" the 4/3rd lens CPU outputs to work
with the new contrast focus cameras. Nikon's V1 apparently works very
well with phase lenses with the adapter they have.

RichA <> wrote:

>Remember when the Hubble telescope was first lauched? It has a very
>fine primary mirror, the most accurate figure and polish ever, but it
>was the wrong figure so it was spherically aberrated. Then they built
>the "Costar" which was an add-on corrective optics set that fixed the
>problem. What is surprising is that they literally can't work with
>the camera electronics to "adapt" the 4/3rd lens CPU outputs to work
>with the new contrast focus cameras. Nikon's V1 apparently works very
>well with phase lenses with the adapter they have.


That's because the Nikon V1 has phase detect AF. It would be very
surprising if phase detect lenses did not work well on it.

http://www.dpreview.com/reviews/nikonv1j1/8

"Thanks to the lightweight elements inside their lenses, the J1 and V1
focus very swiftly in good lighting using phase-detection AF. In poor
light, autofocus switches to contrast-detection, but although still
reasonably responsive, both cameras struggle much more to get an
accurate focus 'lock'."

Rich <> wrote:
>Bruce <> wrote in
>news: :
>> RichA <> wrote:
>>>Remember when the Hubble telescope was first lauched? It has a very
>>>fine primary mirror, the most accurate figure and polish ever, but it
>>>was the wrong figure so it was spherically aberrated. Then they built
>>>the "Costar" which was an add-on corrective optics set that fixed the
>>>problem. What is surprising is that they literally can't work with
>>>the camera electronics to "adapt" the 4/3rd lens CPU outputs to work
>>>with the new contrast focus cameras. Nikon's V1 apparently works very
>>>well with phase lenses with the adapter they have.
>>
>>
>> That's because the Nikon V1 has phase detect AF. It would be very
>> surprising if phase detect lenses did not work well on it.
>
>I wish they'd considered it for the Olympus then, and Panasonic. Though
>the contrast focus is very fast now, it tends to be fast with static
>objects only.


I don't know if Panasonic or Olympus even entertained the idea. The
on-sensor phase detect AF of the Nikon 1 system is (as far as I know)
unique, and will be subject to very high royalties. So it can't have
been an option for m4/3.


>> "Thanks to the lightweight elements inside their lenses, the J1 and V1
>> focus very swiftly in good lighting using phase-detection AF. In poor
>> light, autofocus switches to contrast-detection, but although still
>> reasonably responsive, both cameras struggle much more to get an
>> accurate focus 'lock'."
>
>
> Which isn't really the case with Panasonic and newer Olympus, they do
>well with contrast focus in dim light so having the addition of phase
>would have been a real "exacta win."


True, but for the reasons I stated above, it was never going to
happen. That's probably why Panasonic, Olympus and others have chosen
to develop contrast detect AF to its full potential, and why the
contrast detect AF in m4/3 is so much better than the contrast detect
AF in Nikon's 1 System.

Bruce <> wrote:
> RichA <> wrote:
>>On Feb 27, 7:15Â*am, Bruce <docnews2...@gmail.com> wrote:
>>> RichA <rander3...@gmail.com> wrote:

>>> >Is it all due to connector differences, battery power, or what?

>>> The official explanation is that 4/3 lenses are designed for use with
>>> the phase detect AF systems of E-System DSLRs whereas m4/3 cameras and
>>> lenses use contrast detect AF.

>>That really doesn't specifically answer why the lenses are so
>>sluggish.

> I didn't say it did. I said that was the official explanation. In
> practice, the contrast detect AF system works exceptionally well with
> lenses that are designed for it.

> However, when you use a 4/3 lens that was designed for phase detect AF
> with a m4/3 body that has only contrast detect AF, the AF system
> hunts, often wildly, and can take as much several seconds to reach
> correct focus. Typically it takes about a second, which is really
> slow.

> The problem seems to be that, when faced with an out of focus image
> from a lens that isn't optimised for contrast detect, the camera's AF
> system isn't able to decide which way to go to obtain correct focus.
> So it spends valuable time using trial and error instead - hence the
> hunting.

So far, so fine.

> Somehow, on the other hand, a lens designed for contrast detect will
> work together with the AF system to deliver near-instantaneous focus,
> because it not only knows which way to go but can tell approximately
> how far it needs to go.

That would mean that the lens has an inbuild phase detection
focus unit (or something to that effect). I know there were
some AF lenses that AFfed on their own (without the camera),
so that's not impossible --- just very unlikely in that case.

What I'd assume is a very fast AF motor designed for near-instant
accelleration and stopping of the (as light as possible) focus
group. That would allow the camera to very quick get to the
next test stop and reverse the direction near instantly if they
overshot or went the wrong way. Add some heuristics with
distance guessing, mix with an adapted binary search to
update and confirm the heuristic guesses ...

A phase detection optimized lens doesn't need instant stop
and start, but would need high "long distance" moving speeds
and stopping exactly on the right spot[1] (else a correction is
needed, which takes time). Which also means the minimizing
the focus group weight is not as pressing.

> Even while it is moving towards correct
> focus, it is checking and re-checking the rate of increase of contrast
> and re-evaluating how far to go.

Of course.

> All this seems to happen in the
> blink of an eye, because the AF systems on the latest Panasonic bodies
> are unbelievably fast and accurate.

Hmmm. Accurate --- well, if the focus stops are close enough
to each other, then yes. (Imagine there are only 15 positions
where the lens can/will stop. With a large sensor that's gonna
not be enough.)

Unbelievable fast --- for contrast AF. Not so for phase AF.
Unfortunately. (At least that's what I hear.) But better contrast
AF should "help" the phase AF makers to keep on improving and
think of new ways to faster, more accurate AF.

> If you are going to put phase detect circuitry in to the cameras, you
> need phase detect sensors. Olympus and Panasonic won't be doing that,
> as it will add complication and cost that customers won't pay for.
> There is also the issue of royalty payments for in-sensor phase detect
> AF as used in the Nikon 1 System.

Of course, if contrast AF is 'fast enough', then customers won't
pay. If the alleged speeds (I don't have first hand experience,
so there) are as reported, that will be fast enough for many,
maybe even most use cases (only excluding the real highest AF
speed needs) --- which is fine! (I'll probably stay with
DSLRs for a long time --- and the more camera companies make
money, the more they have for developing better cameras,
inclusive DSLRs.
And maybe I want a better pocketable camera.

> We have seen this phenomenon on newsgroups for years. People get
> extremely worked up about theoretical problems that are barely noticed
> in the real world. They obsess about things like the Nyquist
> frequency and MTF that may be of some relevance to designers of photo
> equipment but have far less relevance (often zero) to people who shoot
> images. Unfortunately, they cannot see the wood for the trees, and
> translating their claimed knowledge of the technology into capturing
> images that please, entertain and move people seems far beyond their
> abilities.

And there are those who ignore well known problems because they
are fanbois of some company, technology or feature, and see only
the real or imagined or theoretical advantages. Like a --- purely
hypothetical --- scene with --- imaginary --- light where only
red pixels in bayer patterns would be triggered by the light,
causing much less resolution, being a clear proof that Foveon
is oh so much better. Or claiming that AA filters are, ah,
not really needed on DSLRs.


-Wolfgang

[1] which is distinct from instantly stopping. Though faster
start and brake times can help, of course.