#### Doug Kerr

##### Well-known member
Especially in discussions of exposure metering, we often run into mention of an exposure result described as "mid gray". Just what exposure result is that, and why does it deserve that name?

One folk phrase in which the term appears is this:

In a proper metered exposure of a frame-filling target with a uniform reluctance of 18%, the exposure result across the frame will be mid gray.​

First I have to point out a problem with that statement, which can be very misleading. In fact, in a proper metered exposure of a frame-filling target with any uniform reflectance (of course as long as its luminance falls in the range acceptable to the metering system), the exposure result across the frame will be the same uniform value, regardless of the target reflectance. This is how reflected light exposure metering works.

Now, what is that exposure result? Well, if we consider the "classical" exposure metering algorithm, using as its exposure index the ISO speed of the sensor, then theoretically that exposure result will imply a luminance of about 12.8% of the "saturation" luminance.

But if we move into the modern era, with the exposure metering algorithm using as its exposure index the ISO Standard Output Sensitivity (ISO SOS) of the sensor, then theoretically that exposure result will imply a luminance of about 18.1% of the "saturation" luminance.

Now does one of these implied relative luminances deserve the name "mid gray"? Well, we might at first think that an implied relative luminance of 50% would deserve to be considered "mid gray".

That would in fact, in graphic arts terminology, be called "50% gray, which sounds sort of "mid-range".
But be careful here. In that notation, an implied relative luminance of 70% is called "30% gray"; after all, is is "less dark", "less gray". And an implied relative luminance of 20% would be called "80% gray" (it is "very dark", "very gray").​

But that doesn't fit our implied relative luminance of 12.8% or 18.1%.

Now if we consider the La*b* color space, it turns out that when L is 50% (the middle of the range of that coordinate), the implied relative luminance is 18.1%. Aha!

So maybe "mid-scale" on the La*b* coordinate "L" is what people have seized upon as a numerical definition of "mid gray".

In any case, I would discourage the use of descriptions of exposure metering matters (or other exposure planning matters) that involve the ambiguous term "mid gray".

Best regards,

Doug

Last edited: