7D AF Sensitivity Setting with 500mm plus 1.xTC for BIF

[Arthur Morris]

So you own both the 7D and MkIV? When does a person that owns the MkIV use his 7D?

BTW, my 7D balances perfectly on my Wimberley Sidekick, with and without the TC and with no battery pack. (I chimp very little and don't use live view a lot, so I'm easy on my batteries.)

I never take my battery pack off of the 7D (or any pro-sumer camera that I have used in the past). It gives them a pro-body feel.

I own two MIV bodies but like to use my 7D on occasion on sunny days always with one of the 70-200 IS zooms.

See Roger's comment on diffraction with the 7D here.

[Arthur Morris]

... The only difference from your suggested settings and what I'm doing is that when I add the TC and I'm shooting BIF, I speed up the AI servo. I feel that that helps me and I'm certain that it's causing no harm.

If the engineers in Japan would provide us quantitative data we might be able to validate my conclusion. It'd be nice to know exactly what happens with AI servo speed when you add a 1.4x TC on a 500mm lens. Does it stay exactly the same or, like the focus acquisition, does it also slow some.

Why can't they say exactly what "fast" and "slow" means? Now that we've been discussing it a week (thanks to all the contributors), I've reexamined the Canon video and descriptions and reread all the attempts to explain it in the camera guides, I'm thinking that it's really the frequency at which the AI servo compares the actual focus to the predicted focus and adjusts to the new target. "Slow" would mean x-times per second and "Fast" might mean 5x-times per second, for instance. To work at all, it has to compare actual results to its predicted results at some frequency.

I'm not expecting science fiction AF, at least not yet. I just want to understand how it works. I don't think that anyone will argue that Canon's done a poor job of explaining it.

Hi David,

When you say ", I speed up the AI servo" please understand that you are not speeding up AI Servo.

I asked Canon's Chuck Westfall this:

"Could the setting for tracking sensitivity--SLOW or FAST say, have any effect at all on AI Servo tracking accuracy with birds in flight?"

Here is his reply:

"Strictly speaking, the answer is no. AI Servo tracking sensitivity affects the speed at which the AF system shifts from tracking the original subject to a new one that takes its place at the selected focusing point. In other words, tracking sensitivity is independent of tracking accuracy."

With regards to this:

"It'd be nice to know exactly what happens with AI servo speed when you add a 1.4x TC on a 500mm lens. Does it stay exactly the same or, like the focus acquisition, does it also slow some."

What do you mean by "AI Servo speed"??? As far as I can figure, there are two considerations: the speed of initial focusing acquisition and the tracking accuracy.....

Lastly, I think that the meaning of your last sentence might be construed as the opposite of what you meant....

[David Stephens]

I never take my battery pack off of the 7D (or any pro-sumer camera that I have used in the past). It gives them a pro-body feel.

Hah, exactly what I don't want. The "pro" bodies feel clunky and overly large in my hands. I know that some people feel there's "status" in that look and feel, but I'm seeking smaller cameras, not larger. Of course, to each his on.

[David Stephens]

Lastly, I think that the meaning of your last sentence might be construed as the opposite of what you meant....

Hi Art,

That sentence is aimed at reader with 8th grade reading level, so, yeah, some will not understand it.

[David Stephens]

I asked Canon's Chuck Westfall this:

"Could the setting for tracking sensitivity--SLOW or FAST say, have any effect at all on AI Servo tracking accuracy with birds in flight?"

Here is his reply:

"Strictly speaking, the answer is no. AI Servo tracking sensitivity affects the speed at which the AF system shifts from tracking the original subject to a new one that takes its place at the selected focusing point. In other words, tracking sensitivity is independent of tracking accuracy."
..

All the more reason that it's sad that Canon doesn't provide meaningful specs. I'd like to see how the predictive AF reacts in mille-seconds as the subject moves toward and away from the lens outside the predicted parameters. There has to be a feedback loop in order for predictive AF to work at all. In some specified light, how many cycles does the loop go through per second? If you add a TC how does that change? If the predicted subject position is out of tolerance from the predicted position, is the AF corrected and a new prediction formulated in a single-cycle or is there some number of cycle to spread the correction over. All this is knowable.

If this cycle is interupted by focusing on a new subject with the focus point, how long is the delay before refocusing when set on FAST or SLOW. How does the AI servo differentiate between an out of tolerance predicted position vs. a new subject? Is it based on the amount of change in the focus difference or something else?

Once again, if you re-read the manual and/or view Canon's video on this matter, no where is
Westfall's interpretation made clear.

[Arthur Morris]

Hi Art,

That sentence is aimed at reader with 8th grade reading level, so, yeah, some will not understand it.

Well, I do not understand what you wrote above either but let's go back to what you wrote initially:

"I don't think that anyone will argue that Canon's done a poor job of explaining it."

Whatever grade you are in or whatever your reading level or comprehension, what you wrote means that everyone would would agree that Canon has done a good job of explaining it.

Is that what you meant?

[Arthur Morris]

Hah, exactly what I don't want. The "pro" bodies feel clunky and overly large in my hands. I know that some people feel there's "status" in that look and feel, but I'm seeking smaller cameras, not larger. Of course, to each his on.

What works for you is of course best. For me, the pro-sumer bodies feel like toys and do not balance well with the super-telephoto lenses.

[Arthur Morris]

All the more reason that it's sad that Canon doesn't provide meaningful specs. I'd like to see how the predictive AF reacts in mille-seconds as the subject moves toward and away from the lens outside the predicted parameters. There has to be a feedback loop in order for predictive AF to work at all. In some specified light, how many cycles does the loop go through per second? If you add a TC how does that change? If the predicted subject position is out of tolerance from the predicted position, is the AF corrected and a new prediction formulated in a single-cycle or is there some number of cycle to spread the correction over. All this is knowable.

If this cycle is interupted by focusing on a new subject with the focus point, how long is the delay before refocusing when set on FAST or SLOW. How does the AI servo differentiate between an out of tolerance predicted position vs. a new subject? Is it based on the amount of change in the focus difference or something else?

Once again, if you re-read the manual and/or view Canon's video on this matter, no where is
Westfall's interpretation made clear.

David,

Sad is a choice. (Byron Katie: www.theWORK.com).

Thanks for the explanation above. I would not be holding my breath for those answers.

And my gut feeling is that if you knew all of the answers it would not make one iota of difference when photographing birds in flight.

Respectfully.

[David Stephens]

Is that what you meant?

Of course not, I meant what I said.

[David Stephens]

David,

Sad is a choice. (Byron Katie: www.theWORK.com).

Thanks for the explanation above. I would not be holding my breath for those answers.

And my gut feeling is that if you knew all of the answers it would not make one iota of difference when photographing birds in flight.

Respectfully.

Yes, agreed, Canon is making a choice here and I wouldn't suggest that anyone hold their breath waiting for hard, or soft, data.

However, I AM confident that if Canon would provide the data that Roger would explain it to us very clearly. :)

[Arthur Morris]

Of course not, I meant what I said.

Then I will pass your positive comment along to Canon.

[David Stephens]

Then I will pass your positive comment along to Canon.

That's fine, just use my words.

[Arthur Morris]

Sir. With all due respect you are in error. Your words clearly do not mean what you intended to write which would likely have been something like this:

"I think that everyone would agree that Canon's done a poor job of explaining it."

[Arthur Morris]

More from Chuck Westfall:


Hi, Arthur:

Thanks for passing along David Stephens' comments. He raises some good questions, but the reality of the matter is that there are too many variables involved to answer his questions at the level of detail he has requested. For instance, differences in overall light level, subject contrast, subject speed, subject distance, lens focal length, other aspects of lens performance, camera performance, etc. can all have a bearing on AI Servo AF tracking sensitivity. Even if Canon released detailed specifications according to the tracking sensitivity setting on the number of milliseconds the AF system holds focus after losing the subject, it would still be incumbent on photographers to choose the setting that best suits the type of photography they are doing, with the cameras and lenses they are using, based on their personal judgment and physical shutter-timing capabilities.
Best Regards, Chuck Westfall