Re: spot, averaging, or incident metering?

thanks, this is a very good explanation.

I tried an experiment like this last night at dusk, and the meters
were all within half a stop of each other. One thing I realized is
that the "ideal conditions" for metering with averaging and incident
meters are actually kind of rare for me.

--shannon

>Shannon wrote:
>
>> Lately I've been experimenting with a little averaging/incident meter
>> that is very small and light. I used it as an incident meter for a
>> few weeks and just developed some of the negatives. A lot of them
>> are either over or underexposed. I was metering in the shade, with
>> the meter set at twice the speed I normally shoot the film at
>>
>> But since the results were rather spotty, I have been using the meter
>> in averaging mode recently. This seems to yield slightly better
>> results, but still not as good as the spot meter.
>>
>> Am I doomed to carry around my heavy spot meter everywhere, even when
>> using small cameras? It seems that when I meter a scene with all
>> three meters, I rarely get the same reading from even two out of
>> three. I believe the spot meter, as I said, because I get very
>> consistent results when I use it. But the incident and averaging
>> meters only agree with it under rather special circumstances, when
>> the light is perfectly even everywhere and the scene is not too
>> contrasty. Like under overcast skies.
>
>The first thing to do is see if the meters all agree. Here's one easy way
>to do this: take a Kodak 18% grey card to a location in wide open shade
>(i.e., not sunlit but exposed to a large amount of sky) on a cloudless day.
>Read the grey card with the spot meter and the averaging meter, and the
>incident light with the incident meter. (Make sure the spot and averaging
>meters are close enough so that the card fills their field of view.) The
>readings should all be the same. If not, the meters are not calibrated
>with each other and you need to address this before trying any further
>comparisons.
>
>A spot meter is generally used to read a uniform area of scene luminance,
>and allows the photographer to decide how to expose that scene area. (In
>practice, one usually meters several areas and may decide to adjust
>processing [N-2, N-1, N+1, N+2, etc.] and choose a print material to
>"place" those subject luminance values closer to a desired print reflection
>density.)
>
>An averaging meter looks at a much larger area of the scene, which (barring
>test conditions such as above) almost certainly contains a wide range of
>luminance values. It does not average based on the relative importance of
>each luminance value in the scene, however, but on the relative sizes
>(areas) of the different luminances in its field of view. For example, if
>you point an averaging meter at the horizon, roughly half of its field of
>view will be taken up by the relatively bright sky, and the resulting
>exposure will generally be too low. Thus, the averaging meter makes the
>assumption, "The areas of all of the different important luminances in my
>field of view are approximately equal." If this is true, the meter will
>give good exposure recommendations. If not, it will provide poor
>recommendations. (Note that you can often point the meter differently to
>compensate for this -- for example, by pointing the meter down in the
>previous example you might include more land that has luminance values
>close to the land just below the horizon, which is a primary
>compositionally important area.)
>
>The incident meter makes the assumption, "I am in the same light that is
>falling on the important objects in the scene, and the reflectance values
>of the important objects in the scene are clustered symmetrically and
>rather closely around 18% grey." Again, if this is true (for example, for
>the mythical "average subject" on an overcast day), the meter will give
>useful readings; if not, it won't. And again, you can do some degree of
>adjustment by intentionally placing the meter in *different* light than
>the important objects in the scene are in, but this is at best very
>approximate.
>
>Best regards,
>
>etienne
Received on Sat Oct 1 08:07:20 2005