7DMKII Focus keeper rate ideas?

[DominoDude]

Hi Domino Dude.
I was aware of that, what I was referring to was how many 0.001mm manufacturing errors cause 1 AFMA unit of correction to be needed.

Cheers, Graham.

To answer the question about how much is 1 unit on the AFMA scale: It is 1/8th of the current lens DoF at its widest aperture.

Edit: Adding source to the above. Rudy W @ Canon, as described in this article -> http://www.learn.usa.canon.com/resources/articles/2011/af_microadjustment_article.shtml

Yupp, I didn't doubt that, Graham. In order to keep it together I quoted you so the information came in a proper context.

 

[DominoDude]

To answer the question about how much is 1 unit on the AFMA scale: It is 1/8th of the current lens DoF at its widest aperture.

Edit: Adding source to the above. Rudy W @ Canon, as described in this article -> http://www.learn.usa.canon.com/resources/articles/2011/af_microadjustment_article.shtml

Actually, Rudy Winston is incorrect. It's not 1/8th the depth of field, but 1/8th the depth of focus. The latter is the sensor-side equivalent of DoF, measured in microns. DoF changes with subject distance, depth of focus is relatively unaffected by subject distance. That distinction means the magnitude of the DoF shift for a given AFMA value changes with subject distance, which is the basis for the recommendation to calibrate at 25-50x the focal length unless you're consistently using the lens at a closer distance.

Ok, good clarification!
I'm not the one to go against Rudy, or you for that matter, on this subject.

 

[dgatwood]

I read posts like this and realise you either have bottomless pockets stuffed with money or lack a clear understanding of manufacturing tolerances.

Either you have way more patience than I do or you're unfamiliar with the concept of testing and binning. Most modern electronics go through an extensive, fully automated testing and binning process, during which the manufacturer burns out fuses, programs PROMs, or whatever to compensate for manufacturing variation. For example, for many years, Intel would burn out defective cores and/or cache on their CPUs and change the product IDs so that they could sell CPUs that were partially defective. One core of that four-core CPU doesn't work? Burn out two cores and sell it as a two-core CPU. Cache on a Pentium chip doesn't work? Burn out the cache and sell it as a Celeron. And so on.

Similarly, when it comes to cameras, the tolerance isn't the main problem. The fact that the user has to calibrate it straight from the factory is the problem. In a sane world, Canon would connect the camera to a test rig using a fixed-focus lens and a target at a fixed distance. The camera would then compute its own base AFMA for each individual focus point in the camera, and would store that data permanently in an EPROM or flash part.

Similarly, for a lens, in an ideal world, Canon would attach it to a test rig consisting of a permanently mounted camera and target, and that modified camera would compute AFMA values for the lens at various zoom settings, and would then program the lens with a series of focus offsets, so that the lens would always focus correctly with zero AFMA (assuming the body was calibrated properly).

When you're talking about a lens that costs thousands of dollars, there is simply no excuse for having to manually calibrate the lens to compensate for significant copy variation. When a lens or camera arrives from the factory, it should have an AFMA adjustment of zero. It should "just work", and no lens/body combination should ever require more than about ±1 AFMA. Anything less than that level of consistency is just plain sloppy, IMO, and reflects a lack of adequate burn-in testing and factory calibration.

 

[Marsu42]

For example, for many years, Intel would burn out defective cores and/or cache on their CPUs and change the product IDs so that they could sell CPUs that were partially defective. One core of that four-core CPU doesn't work? Burn out two cores and sell it as a two-core CPU. Cache on a Pentium chip doesn't work? Burn out the cache and sell it as a Celeron. And so on.

Afaik Intel & AMD also deactivate perfectly working cores/caches if the demand for cheaper cpus is higher than their production ... that's why there are always rumors of "soft-unlocking" schemes with number key you can purchase from the manufacturer. Maybe a business model for Canon - unlock your 18mp camera to 24mp or 11ev to 13ev dynamic range?

 

[Valvebounce]

Hi dgatwood.

dgatwood link
Either you have way more patience than I do or you're unfamiliar with the concept of testing and binning. Most modern electronics go through an extensive, fully automated testing and binning process.
This! :-[ :-X
Similarly, when it comes to cameras, the tolerance isn't the main problem. The fact that the user has to calibrate it straight from the factory is the problem. In a sane world, Canon would connect the camera to a test rig using a fixed-focus lens and a target at a fixed distance. The camera would then compute its own base AFMA for each individual focus point in the camera, and would store that data permanently in an EPROM or flash part.

Similarly, for a lens, in an ideal world, Canon would attach it to a test rig consisting of a permanently mounted camera and target, and that modified camera would compute AFMA values for the lens at various zoom settings, and would then program the lens with a series of focus offsets, so that the lens would always focus correctly with zero AFMA (assuming the body was calibrated properly).

When you're talking about a lens that costs thousands of dollars, there is simply no excuse for having to manually calibrate the lens to compensate for significant copy variation. When a lens or camera arrives from the factory, it should have an AFMA adjustment of zero. It should "just work", and no lens/body combination should ever require more than about ±1 AFMA. Anything less than that level of consistency is just plain sloppy, IMO, and reflects a lack of adequate burn-in testing and factory calibration.

I don't know what the scope is for automating this process, would it require quite a high human interaction to mount and dismount each item, thinking lens release button?
I would imagine that each machine would be costly, plus use a lot of floor space, thinking test an 800mm lens, plus there would be a requirement for more than one of each, routine calibration etc, I still see the price climbing, though possibly by a much smaller margin.
Again I find I am educated by the diverse range of knowledge here. Learning a little every day helps us know we are still alive!

Cheers, Graham.

 

[Sella174]

I don't know what the scope is for automating this process, would it require quite a high human interaction to mount and dismount each item, thinking lens release button?

I am a big fan of job-creation.

I would imagine that each machine would be costly, plus use a lot of floor space, thinking test an 800mm lens, plus there would be a requirement for more than one of each, routine calibration etc, I still see the price climbing, though possibly by a much smaller margin.

Canon either already has dedicated assembly lines for each and every lens in production; or Canon produces lenses in batches, adjusting the assembly line each time. (I remember reading an article on how Henschel built locomotives in the 1930's ... )

 

[Loswr]

I don't know what the scope is for automating this process, would it require quite a high human interaction to mount and dismount each item, thinking lens release button?

I am a big fan of job-creation.

Most companies are not big fans of increased labor costs. Those costs are passed on to the consumer. So, I guess you're also a big fan of higher prices.

 

[Sella174]

I am a big fan of job-creation.

Most companies are not big fans of increased labor costs. Those costs are passed on to the consumer. So, I guess you're also a big fan of higher prices.

I am also a fan of lower crime rates, which would lead to lower/zero insurance premiums ... thereby reducing the total cost of ownership.

 

[dgatwood]

I don't know what the scope is for automating this process, would it require quite a high human interaction to mount and dismount each item, thinking lens release button?

I'd imagine a human would have to mount and dismount each item. I'd expect that to happen as part of basic QC testing anyway, though. I mean, they do at least connect each lens to a controller and perform an electronics test, right?

 

[BLFPhoto]

Sella is unrealistic in his expectations and lacks perspective on design and manufacturing of high precision equipment. Only part of the AFMA equation is electronic. But the whole point is to adjust for minor mechanical tolerance deltas. The magnitude of those deltas is so minuscule as to make further refinement inordinately expensive. It is an acceptable level of risk that minor adjustments have to be made. Further, mechanical systems wear over time, even the best ones. AFMA allows Canon users to account for differences in their older gear as wear, or even minor dings or knocks that shift parts. Mechanical systems change over time. AFMA gives us a simple, painless way to account for that.

And yes, real professionals in all fields dealing with tight tolerances have to calibrate their gear from time to time. I'm sure Neuro can talk long about calibration of scopes and scanning equipment....even multi-million dollar machines.

Calibration is not indication of inherent lack of quality. I think we're lucky to have user calibration available. In the film days, we had to send our gear back to Canon, Nikon, etc. to fix minor focus problems. It was a PITA because it often meant sending bodies with the lenses.

 

[Loswr]

Sella is unrealistic in his expectations and lacks perspective...

Yeah, my favorite from this thread was the expectation that a pro renting some gear should not have to calibrate it. I guess rental houses are supposed to have the same calibration gear Canon should have, then have all rentals hand-delivered by white-glove couriers lest a careless UPS handler drop a box. Oh, and they should do all that without passing the costs along to the customer.

Ok, that's actually my second favorite. Top honors go to lower crime rates leading to zero insurance premiums. : :

 

[Sella174]

Sella is unrealistic in his expectations ...

Not really, given the current trend in consumer protection legislation.

... and lacks perspective on design and manufacturing of high precision equipment.

When you allow a large tolerance variation in your products, you are no longer manufacturing high precision equipment.

AFMA allows Canon users to account for differences in their older gear as wear, or even minor dings or knocks that shift parts. ... AFMA gives us a simple, painless way to account for that.

There is no way that AFMA will ever allow you to adjust for a shifted lens element.

Also, your basic statement is only one-third correct. Old'ish Canon cameras and all new "Rebels" do not offer AFMA. Thus, only if you own a camera that does offer AFMA can you use it ... and this is what fraction of Canon DSLR camera owners? (At least Sigma offers users that docking pod thing for calibrating their lenses, irrespective of the Canon camera owned.)

I'm sure Neuro can talk long about calibration of scopes and scanning equipment....even multi-million dollar machines.

I'm sure neuroanatomist can talk long about many other things as well.

Calibration is not indication of inherent lack of quality.

No, calibration is not an indication of a lack quality, rather the extent to which new equipment needs to be calibrated is an indication of the lack of quality control on the part of the manufacturer.

 

[Sella174]

Yeah, my favorite from this thread was the expectation that a pro renting some gear should not have to calibrate it.

I do believe that specifically I wrote (bold added):

What will happen if such a professional photographer rents (or borrows) a lens at the last moment, but must now spend an hour first to calibrate the combination?

Obviously you've never had equipment croak on you at an event and then have some nice person loan you his "backup" gear. Or vice versa.

I guess rental houses are supposed to have the same calibration gear Canon should have, then have all rentals hand-delivered by white-glove couriers lest a careless UPS handler drop a box.

Personally, I fully expect a reputable rental company to check each and every lens for faults upon receiving it back. Would you be amicable if that super-great lens you rented arrives the afternoon before the big event and it cannot focus properly ... no matter how much you AFMA, huh?

Oh, and they should do all that without passing the costs along to the customer.

Why not. As the saying goes: you pay peanuts, you hire monkeys.

Ok, that's actually my second favorite. Top honors go to lower crime rates leading to zero insurance premiums.

Understandably my original statement was rather a difficult concept to grasp, seeing as it combines legal theory and social science, plus borrowing elements from psychology. However, I am glad it at least had some entertainment value ... granted I'm no Bill Nye.

 

[Loswr]

Yeah, my favorite from this thread was the expectation that a pro renting some gear should not have to calibrate it.

I do believe that specifically I wrote (bold added):

What will happen if such a professional photographer rents (or borrows) a lens at the last moment, but must now spend an hour first to calibrate the combination?

Obviously you've never had equipment croak on you at an event and then have some nice person loan you his "backup" gear. Or vice versa.

Failing to plan is planning to fail. If a professional rents something at the last moment, not allowing time to test out the equipment before it's needed, said professional is not very competent. As for having something fail at an event and needing to borrow a replacement on the spot, again that's failing to plan and thus planning to fail. One should always have a backup – already tested and calibrated – for critical equipment needed to carry out the job properly.

I'm not a professional photographer, so I haven't had those issues in that context. I can say that as a scientist running time-critical experiments, I've had equipment failures over the years...but none of them have resulted in failed experiments, because I always ensure there is backup equipment available...pre-tested and calibrated so it can be immediately swapped in if needed. That's just proper planning...something done by good professionals in any field.

I guess rental houses are supposed to have the same calibration gear Canon should have, then have all rentals hand-delivered by white-glove couriers lest a careless UPS handler drop a box.

Personally, I fully expect a reputable rental company to check each and every lens for faults upon receiving it back. Would you be amicable if that super-great lens you rented arrives the afternoon before the big event and it cannot focus properly ... no matter how much you AFMA, huh?

Of course they should check it for proper operation before it's sent out. What you suggested is they they should somehow ensure that the gear they send me functions perfectly with my own gear with which I plan to use the rental(s), and that's beyond their control. For example, I once dropped my 5DII and while the camera was cosmetically and functionally okay after the fall, all of my AFMA values shifted 10 units negative. How is the rental company going to account for that?

As for it arriving the afternoon before the big event, I wouldn't be so foolish as to rent something and have it arrive the day before I needed it. Rather, I'd have it arrive with sufficient time to test it (even if it left the rental house in perfect condition, shipping damage can occur), and still have a replacement shipped out to arrive before the event, if needed. Failing to plan is planning to fail, remember?

Judging by your comments above, you are willing to accept or excuse a lack of proper planning (i.e., incompetence) by yourself or professionals you may choose to hire. It that works for you, fine...it doesn't work for me.

Ok, that's actually my second favorite. Top honors go to lower crime rates leading to zero insurance premiums.

Understandably my original statement was rather a difficult concept to grasp, seeing as it combines legal theory and social science, plus borrowing elements from psychology.

The concept was simple to grasp, the problem is your grasp on reality. Lower crime rates already lead to lower insurance premiums – localized crime rates, along with localized claim filing rates, are the major drivers for geographical variation in insurance premiums. You did suggest lower premiums, but also zero premiums. For-profit insurance companies giving free coverage? Completely unrealistic.

Also, coverage for photography gear (at least in the US) is rarely for theft only. Loss and unintentional damage are also covered by those policies, and low crime rates won't prevent people from dropping their gear.

 

[sanj]

I am a cinematographer working with top end gear. Before start of every movie my 2nd AC visits the rental house with test charts and checks focus distance with tape measure vs eye focus, zoom lens calibration, flare etc to see if all is ok. OCCASIONALLY he has to ask the rental house to change a lens. This is a perfectly normal procedure in the industry.

All good rental houses have a calibration bench and a tech works on it non stop calibrating lenses.

But are normal consumers supposed to do this with their small cameras is the question….

 

[Loswr]

But are normal consumers supposed to do this with their small cameras is the question….

What's a 'normal consumer'? I'd say a typical consumer buys an entry-level dSLR with the kit lens(es). An APS-C sensor with an f/5.6 lens means DoF is deep enough to mask focus errors even if the user is pixel peeping (which many won't).

People who want the best possible performance from high-end gear should expect to put in some effort to achieve that goal.

 

[Bigjezza]

Woah, interesting discussion going on here. I had no idea when I started this thread lol.

I like to think of these L lenses and bodies as suitable for professionals, but not perfect. I'm happy with the prices now without additional QC and cost. What I mean is the average joe with a bit of saving can afford the exact same system as the 'togs use on the sidelines at any pro sports game. Now think about movies. Independent film makers aren't exactly shooting with the same stuff as Hollywood, are they?

So anyway my issue isn't the body. Its my lens...

I discovered I can check the consistency simply by performing AF and looking at the scale on the lens. If I had thought of this a few weeks ago my shutter count would be a lot less :-[ I let it focus in live view or MF and make a mental note on the scale of where it landed.

What's happening is at 50-70mm on the 24-70 its pretty much hitting the same spot on the scale every time it AF's. I lost count of good AF after 50 goes... Cool. Between 24-50(ish)mm 5 shots here, 1 there, 6 here, 1 there, etc... If the regular AF falls on that spot, all good. Which it does above 50mm.

For a minute I thought in horror maybe its my shiny new 7DMKII... So I got the 70-200 f2.8 IS II and performed the same test at 70 and 200. It mis focussed 1 out of about 50 goes. I can live with that, until someone comes along and convinces me that is not acceptable

I have worried about this for about 2 weeks, doing tests ad hoc time permitting. I came to this conclusion about an hour ago after spending a day at the zoo and in the back yard learning that under 50mm was bizarre.

So I can actually trust this body at f2.8! Now to find a good repair place, and a f1.2 lens haha...

 

[racebit]

People who want the best possible performance from high-end gear should expect to put in some effort to achieve that goal.

It is unfortunate that Canon did not use DPAF to assist PDAF calibration. That would be one of the most important features of DPAF. When 70D did not include it, I though it was because it was first iteration of DPAF, but I was confident it would present on following iterations. Very sad it is not.
The disadvantage of DPAF vs PDAF is speed, that is why we still need PDAF, but we do not need speed to calibrate.

So, yes I agree we should put some effort into calibrating the lens, but Canon should put some effort too, and Canon did nothing to assist the procedure and left all effort to the user.

 

[Valvebounce]

Hi racebit.
Wouldn't this have already been possible on anything that has live view focus? Or is DPAF significantly more accurate, I thought DPAF benefit was speed of live view focus?

Cheers, Graham.

People who want the best possible performance from high-end gear should expect to put in some effort to achieve that goal.

It is unfortunate that Canon did not use DPAF to assist PDAF calibration. That would be one of the most important features of DPAF. When 70D did not include it, I though it was because it was first iteration of DPAF, but I was confident it would present on following iterations. Very sad it is not.
The disadvantage of DPAF vs PDAF is speed, that is why we still need PDAF, but we do not need speed to calibrate.

So, yes I agree we should put some effort into calibrating the lens, but Canon should put some effort too, and Canon did nothing to assist the procedure and left all effort to the user.

 

[Bigjezza]

I had one situation the other week that DPAF failed over PDAF. It was a galvinised steel pole sticking up and the DPAF just wouldn't lock on it no matter what, even after magnifying. Oddly enough spot PDAF nailed it. Can the 7DMKII still do contrast detect AF?