What's up with Canon "calibrating" lenses?

OK, I am new to Canon. I read on here all the time about people sending their lenses in to be "calibrated."

Now I understand what that means in terms of getting the lens to focus where the body tells it it should, but what I don't understand is why this appears to be an issue only with the EOS system?

Is it because of the eletronic lens mount? And that the lens computer and the camera computer need to be on the same page?

And conversly, with other systems that have the motor and computers all in the body it's not an issue?

For example I have never ever read about anyone sending a Pentax body or lens in to be "calibrated!?"

What's the story? Thanks.

in actuality, lense and bodies are not matched up together. so sending both in doesn't insure they will match. lenses are calibrated independently from the body.

Qwntm said:

OK, I am new to Canon. I read on here all the time about people
sending their lenses in to be "calibrated."

Now I understand what that means in terms of getting the lens to
focus where the body tells it it should, but what I don't
understand is why this appears to be an issue only with the EOS
system?

Is it because of the eletronic lens mount? And that the lens
computer and the camera computer need to be on the same page?

And conversly, with other systems that have the motor and computers
all in the body it's not an issue?

For example I have never ever read about anyone sending a Pentax
body or lens in to be "calibrated!?"

What's the story? Thanks.

Click to expand...

--
-tim
http://www.pbase.com/pdqgp

Why does this need to occur at all? Canon is(was) a pioneer in AF technology. With passive AF systems, the camera body alone should detect correcct focus, no?

Robert A

Tim Lauro said:

Qwntm said:

OK, I am new to Canon. I read on here all the time about people
sending their lenses in to be "calibrated."

Now I understand what that means in terms of getting the lens to
focus where the body tells it it should, but what I don't
understand is why this appears to be an issue only with the EOS
system?

Is it because of the eletronic lens mount? And that the lens
computer and the camera computer need to be on the same page?

And conversly, with other systems that have the motor and computers
all in the body it's not an issue?

For example I have never ever read about anyone sending a Pentax
body or lens in to be "calibrated!?"

What's the story? Thanks.

Click to expand...

--
-tim
http://www.pbase.com/pdqgp

Click to expand...

Qwntm said:

OK, I am new to Canon. I read on here all the time about people
sending their lenses in to be "calibrated."

Now I understand what that means in terms of getting the lens to
focus where the body tells it it should, but what I don't
understand is why this appears to be an issue only with the EOS
system?

Click to expand...

Canon service centers have an optical bench. An optical bench is a platform on which various optical measurements can be made (and repeated with accuracy). By using various sources of light, and, especially by using a laser the technician can see optical defects in the various lenses and if the alignment or colimation of the lenses or other mechanicals are preventing the lens (as a whole) from doing its job with crispness and clairity. And as long as they have the lens in the 'jig' they can attache a focal point measurement device to see where the lens does focus and if its in cailbration or not.

The big difference between digital and film with respect to lenses is that digital takes no prisoners. The images are so clear that a trained optician can read lens aberations directly from the image. It was much harder in the days of film since days passed between taking the image and seeing the picture, and most films don't have the resolution that digital does. Digital is, in effect, resolving the idiosyncracies of the lens(es).

Qwntm said:

Is it because of the eletronic lens mount? And that the lens
computer and the camera computer need to be on the same page?

Click to expand...

The camera performs the focusing operations, the lens is mearly a slave.

--
Mitch

Robert A said:

Why does this need to occur at all? Canon is(was) a pioneer in AF
technology. With passive AF systems, the camera body alone should
detect correcct focus, no?

Click to expand...

Not really. Rough process:

• camera looks at how strong the contrast is once,
• camera adjusts the lens a little bit,
• camera looks at how strong the contrast is a second time,
• camera decides how far it will need to move lens in which direction,
• camera moves lens that far,
• camera DOES NOT CHECK FOCUS AGAIN AFTER THAT "GUESS"
• shutter opens for picture.

If the camera wanted to actually find the best focus, it would have to do what any manual focus photographer does: slowly move the lens across a range of focal positions, decide where within that range was right, move back past that best position for confirmation, and move forward again to the best position. This takes way too long for grabbing action shots, so the camera makes a calculated guess and does not try to move the lens across anywhere near its whole focus range, nevermind do it twice.

I think the reason for all the calibration stuff is because Canon judges lens tolerances for sharpness at 5"x7" print size, while suddenly there were a ton of new customers who naively judge sharpness at 100% pixel viewing instead. Canon's AF is accurate within 1/3 of the focal depth, but that's huge compared to the expectations of pixel-peeping point-and-shooters.

A smaller factor is just the changes in EOS EF microchips-- older lenses of any manufacture have had to be "rechipped" to be compatible with new specifications for newer EOS EF bodies. This may influence the issues with how the cameras make assumptions about lens focus motor performance.

--
[ e d @ h a l l e y . c c ] http://www.halley.cc/pix/

Ed Halley said:

Robert A said:

Why does this need to occur at all? Canon is(was) a pioneer in AF
technology. With passive AF systems, the camera body alone should
detect correcct focus, no?

Click to expand...

Not really. Rough process:

• camera looks at how strong the contrast is once,
• camera adjusts the lens a little bit,
• camera looks at how strong the contrast is a second time,
• camera decides how far it will need to move lens in which direction,
• camera moves lens that far,
• camera DOES NOT CHECK FOCUS AGAIN AFTER THAT "GUESS"
• shutter opens for picture.

If the camera wanted to actually find the best focus, it would have
to do what any manual focus photographer does: slowly move the
lens across a range of focal positions, decide where within that
range was right, move back past that best position for
confirmation, and move forward again to the best position. This
takes way too long for grabbing action shots, so the camera makes a
calculated guess and does not try to move the lens across anywhere
near its whole focus range, nevermind do it twice.

Click to expand...

From what I've read since 1987, when the photo mags were first publishing descriptions of how the EF lenses work, I believe you're describing the process correctly. A point here is that Canon lenses include a processor to carry out part of this function.

Ed Halley said:

I think the reason for all the calibration stuff is because Canon
judges lens tolerances for sharpness at 5"x7" print size, while
suddenly there were a ton of new customers who naively judge
sharpness at 100% pixel viewing instead. Canon's AF is accurate
within 1/3 of the focal depth, but that's huge compared to the
expectations of pixel-peeping point-and-shooters.

Click to expand...

My datamining tends to confirm this. The 5x7 size is mentioned on the Luminous-Landscape site. That piqued my interest because in the past the DoF standard size enlargement was 8x10. Going to 5x7 would indicate a rather arbitrary decision to loosen the tolerances.

I compared the depth of field scales on my old Canon FD 50mm f1.8 (design circa 1972) and my Canon EF 50mm f1.8 Mk I (design circa 1986), and indeed--the newer lens scale shows a significantly greater DoF at every aperture and distance than the older lens. Canon DID loosen their tolerances with the EF lens design.

In at least two currently publised sources (the EOS FAQ on usa.canon.com and the PDF pamplet on getting the best use out of 1D series cameras), Canon advises users NOT to judge AF accuracy from the monitor (and especially not at 100 percent), but to print out an 8x10 to judge AF accuracy.

When I shot 35mm film, I rarely enlarged over 11x14 and almost never over 16x20. At that, I expected my photographs to be viewed no closer than arms length. Digital users commonly review their images at the equivalent of 36x48 or larger--and then bury their noses into the monitor.

This is my question in all these cases: Does the problem show up in the mode I present my images for public viewing? If not, then it's irrelevant.

Ed Halley said:

I think this is
A smaller factor is just the changes in EOS EF microchips-- older
lenses of any manufacture have had to be "rechipped" to be
compatible with new specifications for newer EOS EF bodies. This
may influence the issues with how the cameras make assumptions
about lens focus motor performance.

Click to expand...

This has not been the case with Canon. The very oldest EF lenses work even better with the newest cameras than they did with their contemporary cameras. However, newer lenses certainly DO focus faster and more accurately than older lenses (according to Chuck Westfall) because of faster processors in them.

The best conclusion I've seen about the problems Sigma had in this regard is that as a battery-saving action, Canon reduced the power output of their cameras (the only change I've seen identified) at the lens mount. Apparently certain Sigma lenses had power-hungry chips that couldn't operate with the lessened power; the fix was to replace those chips with more efficient chips.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'

RDKirk said:

The best conclusion I've seen about the problems Sigma had in this
regard is that as a battery-saving action, Canon reduced the power
output of their cameras (the only change I've seen identified) at
the lens mount. Apparently certain Sigma lenses had power-hungry
chips that couldn't operate with the lessened power; the fix was to
replace those chips with more efficient chips.

Click to expand...

Could this really be all that the rechipping process is about? That's quite interesting as it would wipe the floor with all those theories about conflicting lens communication protocols.

Roy.

RDKirk said:

My datamining tends to confirm this. The 5x7 size is mentioned on
the Luminous-Landscape site. That piqued my interest because in
the past the DoF standard size enlargement was 8x10. Going to 5x7
would indicate a rather arbitrary decision to loosen the tolerances.

Click to expand...

My understanding is that the APS sensor size AF algorithms in the Canon DSLRs were not changed from FF ones, which assume a CoC of 0.035. If visual tolerance is correct at 8x10 and 0.035 for FF, it would be about correct for APS at 5x7 using the same CoC (Zeiss' formulas can be worked in any number of ways, since he made a number of key assumptions, such as smallest detail an average eye can resolve, "standard" print size, and "standard" viewing distance).

--
Thom Hogan
author, Nikon Field Guide & Nikon Flash Guide
author, Complete Guides to the Nikon D70, D100, D1 series, and Fujifilm S2 Pro
http://www.bythom.com

Thom Hogan said:

My understanding is that the APS sensor size AF algorithms in the
Canon DSLRs were not changed from FF ones, which assume a CoC of
0.035. If visual tolerance is correct at 8x10 and 0.035 for FF, it
would be about correct for APS at 5x7 using the same CoC (Zeiss'
formulas can be worked in any number of ways, since he made a
number of key assumptions, such as smallest detail an average eye
can resolve, "standard" print size, and "standard" viewing
distance).

Click to expand...

I'm sure this is true if the intention is to keep the DoF assumptions on which lens scales (and AF systems accuracy specs?) are based accurate for both formats that those same lenses might be mounted upon...

This doesn't explain, though, the difference in the scales on the FD and EF lenses, because the EF lenses were also designed for full frame. But they WERE designed for autofocusing...did Canon assume a larger CoC specficially for autofocusing? That makes the AF accuracy problem a heck of a lot less of a problem.

Figuring out Canon is like being a Western intelligence analyst trying to figure out Kim Jong Il. It's not as if Canon couldn't write a few white papers now and then.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'

Roy van der Woning said:

Roy van der Woning said:

The best conclusion I've seen about the problems Sigma had in this
regard is that as a battery-saving action, Canon reduced the power
output of their cameras (the only change I've seen identified) at
the lens mount. Apparently certain Sigma lenses had power-hungry
chips that couldn't operate with the lessened power; the fix was to
replace those chips with more efficient chips.

Click to expand...

Could this really be all that the rechipping process is about?
That's quite interesting as it would wipe the floor with all those
theories about conflicting lens communication protocols.

Click to expand...

Canon has always maintained that they've never changed the comms protocols, and, indeed, the oldest EF lenses do work perfectly on the newest cameras. Yet, we know Sigma lenses had a problem (of course, Canon denies any knowledge of even Sigma's existance).

I've seen one claim that Canon did lower the power output, and that's the only change I've ever read having been identifed by anyone.

Merely lowering the power output of the camera is a change that's not actually a change in protocol (so if we look at it obliquely, Canon didn't actually lie about not changing the protocol). It would mean those 3rd party makers who happened to use power-sipping chips from the start (undoubtely more expensive chips) got lucky.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'

just about all AF system components from all camera manufacturers require calibration during manufacture. So your question is really "Why is recalibration so prominent with Canon systems" and the answer to that is simply poor quality control during manufacture. This applies to both their bodies and their lenses. If you buy a Canon body then be prepared to check it out nine ways to Sunday before the exchange period expires. The more "entry level the body" then the more likely the AF may be off. Same applies to lenses and the cheaper the lens...well you get the picture.

Qwntm said:

OK, I am new to Canon. I read on here all the time about people
sending their lenses in to be "calibrated."

Now I understand what that means in terms of getting the lens to
focus where the body tells it it should, but what I don't
understand is why this appears to be an issue only with the EOS
system?

Is it because of the eletronic lens mount? And that the lens
computer and the camera computer need to be on the same page?

And conversly, with other systems that have the motor and computers
all in the body it's not an issue?

For example I have never ever read about anyone sending a Pentax
body or lens in to be "calibrated!?"

What's the story? Thanks.

Click to expand...

How does your essay pertain to front and backfocus problems? Seems that's a consistent failure of some kind, yes? Isn't that what people are sending their cameras/lenses in for?

--
Eric, Snapshooter
DSixty, GTwo, Four20EX, Fuji 38Hundred
http://www.pbase.com/haglunde

Ed Halley said:

Robert A said:

Why does this need to occur at all? Canon is(was) a pioneer in AF
technology. With passive AF systems, the camera body alone should
detect correcct focus, no?

Click to expand...

Not really. Rough process:

• camera looks at how strong the contrast is once,
• camera adjusts the lens a little bit,
• camera looks at how strong the contrast is a second time,
• camera decides how far it will need to move lens in which direction,
• camera moves lens that far,
• camera DOES NOT CHECK FOCUS AGAIN AFTER THAT "GUESS"
• shutter opens for picture.

If the camera wanted to actually find the best focus, it would have
to do what any manual focus photographer does: slowly move the
lens across a range of focal positions, decide where within that
range was right, move back past that best position for
confirmation, and move forward again to the best position. This
takes way too long for grabbing action shots, so the camera makes a
calculated guess and does not try to move the lens across anywhere
near its whole focus range, nevermind do it twice.

I think the reason for all the calibration stuff is because Canon
judges lens tolerances for sharpness at 5"x7" print size, while
suddenly there were a ton of new customers who naively judge
sharpness at 100% pixel viewing instead. Canon's AF is accurate
within 1/3 of the focal depth, but that's huge compared to the
expectations of pixel-peeping point-and-shooters.

A smaller factor is just the changes in EOS EF microchips-- older
lenses of any manufacture have had to be "rechipped" to be
compatible with new specifications for newer EOS EF bodies. This
may influence the issues with how the cameras make assumptions
about lens focus motor performance.

--
[ e d @ h a l l e y . c c ] http://www.halley.cc/pix/

Click to expand...

First I sent in one body with NO lenses. They said they would check the body against a 50 1.4. THIS time, I sent in another body but this time with the lenses I had issue with.

I was having misfocus on 35 1.4L, 85 1.8 usm and 135 f2 at fstops between 1.4 and f2. Canon did something to one of my bodies because when it came back, it was better. NO question. I kept that one and sent in the other body and those three lenses this time. It's in Irvine as I type this.

So, maybe it is a lens issue, but whatever they did to "body only" it helped. My guess is factory tolerances are not as high as some lenses demand. Only a guess.

Thom Hogan said:

My understanding is that the APS sensor size AF algorithms in the
Canon DSLRs were not changed from FF ones, which assume a CoC of
0.035.

Click to expand...

Thom is correct. Canon uses a CoC of 0.035mm for measuring autofocus accuracy. I've asked Canon about this and have been told that "EF lenses are designed to produce a minimum circle of confusion of 0.035mm."

Chuck Westfall has written that Canon makes "normal-precision" and "high-precision" autofocus systems. Normal-precision autofocus systems (e.g. D30, D60, 10D) are designed to focus within one depth of focus (equivalent to one depth of field) for the lens' maximum aperture, whereas high precision systems focus (e.g. 1D, 1Ds, 20D) within 1/3 of a DOF (See: http://www.bobatkins.com/photography/eosfaq/eos3af.html ). The 20D exhibits "high-precision" autofocus, but only when used with lenses with a maximum aperture of F2.8 or larger (se page 68 of the 20D instruction manual).

So, for example, when mounted with using a Canon 70-200 F4 lens, the 20D will exhibit normal precision focusing.

What does this mean in practical terms? If means that if you try and autofocus on something 100 feet away with this lens at 70mm, the camera may end up actually focusing at anything between about 50 feet and 750 feet, and still be within canon's specifcations. (See the depth of field calculator here:
http://www.tawbaware.com/maxlyons/calc.htm ).

Max