From: Paul Furman on
David J Taylor wrote:
> Paul Furman wrote
>
>> It's not clear his camera gives a range, just one beep. And he got
>> consistent dead-on results from the front. And consistently front
>> focused coming from the rear so a fuzz factor band appears not to
>> explain it. My Nikons give little bracket arrows when approaching
>> acceptable focus, one side or the other or both, then a green dot in
>> the middle when it's optimal.
>
> What you describe, Paul, is exactly what I suggested might happen - that
> the focus stops with a non-zero, but "acceptable" error (i.e. within the
> depth-of-focus), and that error is a different error sign according to
> which direction focus is approached.
>
> It sounds as if "his camera" has a slight front focus offset, and a
> dead-band of about twice that offset. Approach from infinity, and you
> arrive at:
>
> offset - dead-band = near-perfect focus
>
> and approach from close-up and you arrive at:
>
> offset + dead-band = noticeable focus error.
>
> If I've understood you correctly, that is.

Except you got the starting points reversed.

--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam
From: David J Taylor on

"Paul Furman" <paul-@-edgehill.net> wrote in message
news:hivlsv$cvs$1(a)news.eternal-september.org...
> David J Taylor wrote:
>> Paul Furman wrote
>>> It's not clear his camera gives a range, just one beep. And he got
>>> consistent dead-on results from the front. And consistently front
>>> focused coming from the rear so a fuzz factor band appears not to
>>> explain it. My Nikons give little bracket arrows when approaching
>>> acceptable focus, one side or the other or both, then a green dot in
>>> the middle when it's optimal.
>>
>> What you describe, Paul, is exactly what I suggested might happen -
>> that the focus stops with a non-zero, but "acceptable" error (i.e.
>> within the depth-of-focus), and that error is a different error sign
>> according to which direction focus is approached.
>>
>> It sounds as if "his camera" has a slight front focus offset, and a
>> dead-band of about twice that offset. Approach from infinity, and you
>> arrive at:
>>
>> offset - dead-band = near-perfect focus
>>
>> and approach from close-up and you arrive at:
>>
>> offset + dead-band = noticeable focus error.
>>
>> If I've understood you correctly, that is.
>
> Except you got the starting points reversed.
>
> --
> Paul Furman

I'll take your word for that!

Cheers,
David

From: Wilba on
Paul Furman wrote::
> Wilba wrote:

>> |< 430mm >| sensor to subject distance for optimal
>> focus
>
> 430mm from sensor to subject when focused.

No, the POF is always 430mm from the sensor in the beep test. If the subject
is there too, it will be in focus.

>> If I start with the POF on the farside of the subject (subject less than
>> 430mm from the sensor) -
>> C S<------ >>DOF ------>POF
>> and increase the sensor to subject distance (so the POF approaches the
>> subject from the farside), I get a beep (focus confirmation) at 433mm,
>> i.e. front-focus -
>
> 433mm is more than 430mm so the text on the line above describes back
> focus.

When the subject is 433mm from the sensor, and the POF is 430mm from the
sensor, that's front-focus.

> The diagram immediately following below shows front focus with the subject
> 3mm further away from the plane of focus than it was before.

Exactly! That's what I'm depicting. The POF is always 430mm from the sensor
in this scenario.

> It looks like you turned the focus ring.

No, the sensor to subject distance increased while the POF remained 430mm
from the sensor.

> Didn't you say the subject was 430mm away from the sensor
> when focus is perfect?

No, I said if the subject is coincident with the POF (430mm from the
sensor), it will be in focus.

> This scenario still has the plane of focus at 430mm but the subject
> isn't in focus any more.

You got it!

> Does not compute. Illogical.

The whole scenario is - sensor to POF distance is constant, sensor to
subject distance changes. Computes perfectly.

> The DOF symbols with various numbers of angle brackets are just
> theoretical and make the diagrams confusing. I deleted them in my version.
> You could discuss them later but they aren't part of the measured data.

Correct.

>> C POF<-------- >DOF/2 -------->S
>> |< 430mm >|< 3.0mm >| "farside
>> focus" case
>> This is the counter-intuitive "cross-over" bit - you'd expect this case
>> to put the subject 427mm from the sensor for a 3mm beep band.

This is the bit I screwed up. See previous message.

>> This outcome -
>> C S POF
>> | | |
>> never occurs.
>
> It does if optimal focus is 430mm and the camera's AF system confirms
> focus at 433mm. That's back focus.
>
> C S POF
> | | |
> 0mm 430mm 433mm

Incorrect. C to POF is always 430mm, so back-focus would look like this -

C S POF
| | |
0mm 427mm 430mm

which never happens.

>> Can you identify the point at which the diagram is "opposite" to the
>> results, and describe in your words what you're seeing?
>
> Done.

From the above, I believe you have been reading these diagrams depicting the
beep test as if they were depicting AF tests (in which the sensor to subject
distance remains the same and the sensor to POF distances changes).

If you can stand it, we should just work on the AF case, which looks quite
different but shows the crossover phenomenon. Let me know if you have the
will to carry on. :- )


From: Wilba on
Chris Malcolm wrote:
> Wilba wrote:
>> Chris Malcolm wrote:
>>> Wilba wrote:
>>>>
>>>> Okay, let's assume you're right, that there is a miscalibration which
>>>> results in a consistent offset in the focus (I've accommodated that
>>>> possibility in my other reply today). How does what we've been
>>>> talking about account for the difference between the two focuses?
>>>> One is consistently _here_, and other is consistently _there_.
>>>> That's the significant thing. How do you account for the width
>>>> of the beep band?
>>>
>>> In many modern cameras and lenses which communicate electronically the
>>> width of the "beep band", i.e., the change in focus over which the
>>> subject is deemed to be in focus by the AF system, is set by an
>>> acceptable focus error parameter for that lens. In other words focus
>>> is considered to be ok not when the focus error is at its smallest,
>>> but when it's less than this acceptable error parameter. Very sharp
>>> lenses will have a smaller value than softer lenses. The lens chip
>>> holds the value, supplied to it at lens calibration time, and it
>>> supplies this value to the camera along with its other parameters such
>>> as max aperture, focal length, etc..
>>>
>>> When you send a lens off to be recalibrated one of the things they may
>>> do is to rewrite the lens focussing parameter table with values which
>>> more accurately reflect the performance of that particular copy of the
>>> lens, and which are more appropriate to your specific camera and
>>> purposes.
>>
>> Right-oh, so there's nothing we can think of so far that directly
>> attributes
>> the width of the beep band to any form of ARFD.
>
> Depends what you mean by "directly".

A first-order effect.

> A lens with AFRD won't be able to be as accurately focused by phase
> detection as one without, so the lens calibraters might wish to set the
> acceptable error (width of beep band) larger.

A second-order effect.


From: Chris Malcolm on
Wilba <usenet(a)cutthisimago.com.au> wrote:
> Chris Malcolm wrote:
>> Wilba wrote:
>>> Chris Malcolm wrote:
>>>> Wilba wrote:
>>>>>
>>>>> Okay, let's assume you're right, that there is a miscalibration which
>>>>> results in a consistent offset in the focus (I've accommodated that
>>>>> possibility in my other reply today). How does what we've been
>>>>> talking about account for the difference between the two focuses?
>>>>> One is consistently _here_, and other is consistently _there_.
>>>>> That's the significant thing. How do you account for the width
>>>>> of the beep band?
>>>>
>>>> In many modern cameras and lenses which communicate electronically the
>>>> width of the "beep band", i.e., the change in focus over which the
>>>> subject is deemed to be in focus by the AF system, is set by an
>>>> acceptable focus error parameter for that lens. In other words focus
>>>> is considered to be ok not when the focus error is at its smallest,
>>>> but when it's less than this acceptable error parameter. Very sharp
>>>> lenses will have a smaller value than softer lenses. The lens chip
>>>> holds the value, supplied to it at lens calibration time, and it
>>>> supplies this value to the camera along with its other parameters such
>>>> as max aperture, focal length, etc..
>>>>
>>>> When you send a lens off to be recalibrated one of the things they may
>>>> do is to rewrite the lens focussing parameter table with values which
>>>> more accurately reflect the performance of that particular copy of the
>>>> lens, and which are more appropriate to your specific camera and
>>>> purposes.
>>>
>>> Right-oh, so there's nothing we can think of so far that directly
>>> attributes
>>> the width of the beep band to any form of ARFD.
>>
>> Depends what you mean by "directly".

> A first-order effect.

>> A lens with AFRD won't be able to be as accurately focused by phase
>> detection as one without, so the lens calibraters might wish to set the
>> acceptable error (width of beep band) larger.

> A second-order effect.

Back in the old days before cameras had computers inside the
performance of such smart features as autofocus was dominated by the
limits of electromechanical design and performance of the system. We
have so much better technology today, plus computers, that a great
deal of the performance is dominated by second order effects. It's
easier and cheaper for the manufacturers to do it that way.

In much of this long discussion you've been trying to find direct
first order effects to explain things which are the results of second
order effects.

--
Chris Malcolm