Doug Kerr
Well-known member
As a result of my inquiry in another forum about exactly how Evaluative metering works with AI Servo AF on Canon EOS dSLRs , I received a wonderful, thorough discussion of the issues from Chuck Westfall of Canon USA. Since my original inquiry was also posted here, I posted in that thread a link to Chuck's response, and made a few comments.
But I think what I learned (it may be old news to many of you) is so important I thought it deserved its own outing in a separate thread.
Here is Chuck's essay:
http://www.prophotohome.com/forum/ca...tml#post469325
I was startled to learn that, even if several AF points "light up", this merely indicates which points on the scene lie within the depth of field surrounding the distance at the one point that is used as the basis of AF (I of course realized that only one AF detector actually controlled focus), and it is only the area around that "principal" point (and we are not shown which one that is) that is given "emphasis" (whatever that is) by the Evaluative metering system. (I had always thought that the areas around all "lit" AF points were given "emphasis".)
I am quite certain that in my 300D days, this was not the impression I was given by Canon's explanations of the Evaluative metering system (although the details were never completely clear). Of course I may have then jumped to unwarranted conclusions.
The inference I drew was that the AF system, based on distance consideration and other things, decided what the "extent" of the subject was, and arranged for emphasis to be given to the areas around all AF points lying in that extent - a "subject-wide" intelligent metering algorithm.
Perhaps the doctrine has changed (although Chuck suggests not).
I also think I see in Chuck's remarks (as well as by the language in the 40D manual) that Canon today has a more modest view of when Evaluative metering is most appropriate than at an earlier time (when it seemed to be the answer to everybody's prayer for exposure metering).
The pivotal point is that if we have Evaluative metering in effect with Automatic selection of AF points ("all AF" points available), we cannot be sure which AF point (of all those that are lit) is the one (and the only one) whose surrounding area is given "emphasis" by the Evaluative metering system (and of course we still just barely know what that really means).
Now, here's the way this impacts my original reason for inquiring. I had been feeling sad that, if I preselect a single AF point, in order to get predictable focus, I was somehow missing the "intelligence" of the Evaluative metering system in discerning which AF points lay in areas warranting "emphasis" in metering and considering them all in making an exposure decision.
Well, it turns out that, had I used Automatic AF point selection, but managed (perhaps by good luck) to get the "principal" AF point to be on the subject, only the area around that AF point would have been given emphasis in metering anyway.
And I guess now I see why so many people have said, "Evaluative? Don't bother. I use Partial or AWB." Oldies but goodies.
But I think what I learned (it may be old news to many of you) is so important I thought it deserved its own outing in a separate thread.
Here is Chuck's essay:
http://www.prophotohome.com/forum/ca...tml#post469325
I was startled to learn that, even if several AF points "light up", this merely indicates which points on the scene lie within the depth of field surrounding the distance at the one point that is used as the basis of AF (I of course realized that only one AF detector actually controlled focus), and it is only the area around that "principal" point (and we are not shown which one that is) that is given "emphasis" (whatever that is) by the Evaluative metering system. (I had always thought that the areas around all "lit" AF points were given "emphasis".)
I am quite certain that in my 300D days, this was not the impression I was given by Canon's explanations of the Evaluative metering system (although the details were never completely clear). Of course I may have then jumped to unwarranted conclusions.
The inference I drew was that the AF system, based on distance consideration and other things, decided what the "extent" of the subject was, and arranged for emphasis to be given to the areas around all AF points lying in that extent - a "subject-wide" intelligent metering algorithm.
Perhaps the doctrine has changed (although Chuck suggests not).
I also think I see in Chuck's remarks (as well as by the language in the 40D manual) that Canon today has a more modest view of when Evaluative metering is most appropriate than at an earlier time (when it seemed to be the answer to everybody's prayer for exposure metering).
The pivotal point is that if we have Evaluative metering in effect with Automatic selection of AF points ("all AF" points available), we cannot be sure which AF point (of all those that are lit) is the one (and the only one) whose surrounding area is given "emphasis" by the Evaluative metering system (and of course we still just barely know what that really means).
Now, here's the way this impacts my original reason for inquiring. I had been feeling sad that, if I preselect a single AF point, in order to get predictable focus, I was somehow missing the "intelligence" of the Evaluative metering system in discerning which AF points lay in areas warranting "emphasis" in metering and considering them all in making an exposure decision.
Well, it turns out that, had I used Automatic AF point selection, but managed (perhaps by good luck) to get the "principal" AF point to be on the subject, only the area around that AF point would have been given emphasis in metering anyway.
And I guess now I see why so many people have said, "Evaluative? Don't bother. I use Partial or AWB." Oldies but goodies.