EOS users: don't let this happen to you ...

Bob Steele over on NPN pointed me an article containing the following:

--------------------------------------------------

E-TTL flash metering patterns:

The camera uses its evaluative metering system to meter the flash output, based upon the preflash. When in autofocus mode most EOS bodies bias flash metering toward the currently selected AF point, but always in an evaluative mode pattern - they don’t use spot or partial metering patterns. When in manual focus mode it appears that at least some EOS bodies switch to centre-weighted averaging.

Note, however, that this biasing of E-TTL metering to the active point is potentially problematic, since it means that the flash metering is done in almost a spot-metering fashion. Many user complaints regarding flash metering problems in E-TTL mode appear to be linked to this issue. If the camera happens to be over a dark object, for example, flash metering can be considerably overexposed, and vice-versa. The standard answer to this problem is to use FEL and meter off something mid-toned, but this is clearly not a solution for rapid-shooting situations such as weddings and sports. Another approach is to set the camera lens to manual focus, since the body apparently switches to centre-weighted average metering in that mode, but that’s obviously not a useful answer much of the time either.

Users of the digital D30 and D60 have been particularly unhappy with E-TTL flash metering. The 10D apparently reduces this problem by defaulting to a centre-weighted averaging metering pattern in E-TTL, even when the lens is set to autofocus.

Note that since I’ve been unable to find definitive published statements from Canon on this topic it isn’t as authoritative as it could be. Please contact me if you have further information about E-TTL flash metering.

------------------------------------------------------

This is no doubt the phenomenon that Phil Shaw was referring to, and I must admit shakes what little faith I had left in Canon's flash metering schemes. So many variables have now been thrown into this equation that I'll probably never know what was really going on that night ...

Paul Skoczylas wrote:I'm confused by this. If there is a prominent subject, shouldn't the TTL metering shut the flash off when the subject is exposed properly, and not wait for the background to be exposed to 18%? The definition of a "prominent subject" might be different from one manufacturer to another.

If its really prominent yes but birds in flight are rarely that prominent unless you are clipping part of the body or wings.

EJ:

Thanks for your explanation of my overflashing experience at Bosque last year. It happened on my last night. I was just starting to think about now as I have Bosque on my mind:)