Idaho Airships, Inc.

An imaging target will present a range of of brighter to darker features-it’s “Luminance Range.” That luminance range is dependent upon two factors: The amount of light available (illuminance), and how much of that ambient light is reflected back to the camera (reflectance).

In aerial photography applications we typically meter from reflected light using our camera. An Incident Light Meter may determine the actual level of ambient light but they are terribly awkward to use in the great majority of aerial applications.

Consider the following aerial (Sandy, UT 8/07):

The amount of ambient light is quite constant throughout the frame.

There is a wide range of reflectance from the various subject elements in this image: The white concrete parking lots are not actually very reflective (~20%), while the dark glass facades of the structures in the immediate background are very reflective (~90%). This (by my experience) does not represent an extreme situation for an aerial oblique assignment.

Our challenge is to amalgamate a number of exposure criteria in order to produce the most virtuous image-one without (or with minimal…) unduly clipped highlights, with the most densely populated upper key possible for demosaicing/conversion, with the maximum amount of saturative potential, and without (or with minimal) posterization or blocking in the output shadows. This is a sort of digital-sensitive extension of the classic Adams-Archer-White Zone Metering-it doesn’t simply take into account film curves, it recognizes that our digital capture is linear and must be Gamma-adjusted, contrast-controlled, then converted to an output profile with supplemental black and white point manipulations.

Food For Thought: This is the previous image, blurred (Gaussian) mercilessly. It’s gray: Mean = 88.2, Std. Dev. <3 (the blur was comprehensive). CMY’s K ~30%. What does this indicate about the global exposure?

A subject element’s luminance comprises the result of ambient (or incident) light and reflected light. Since there are rarely any compositions which enjoy a homogenous reflectance, you can sense the possibility for exposure slop…particularly when your camera presumes that everything produces a reflectance of 18%. 18% was chosen as an amount representative of a typical outdoor composition under sun and tends to favor chrome film performance over negative films and digital (which didn’t exist at the time 18% was established as a standard).

Gray cards are manufactured with 18% reflectance. They are also manufactured at 50% tonal value (neutrality)-that’s why they’re gray cards. If they were manufactured with 50% reflectance, they would result in a lower EV being set by the camera because the card would be speciously bright.

So, for aerials and while using reflected light metering, we’ll speak in the 18% reflectance dialect and in this exercise, presume unadulterated spot metering.

Scottsdale, AZ 11/07. This is the prepared version of the following simulations. The red dots will indicate the spot metering locus.

If the metered element has a reflectance of 5% your camera will meter it presuming 18% and your image will be overexposed as the camera adjusts exposure up to compensate for the paucity of light being reflected.

If the metered element has a reflectance of 50% your camera will meter it presuming 18% and your image will be underexposed as the camera adjusts exposure down to facilitate the extra light being reflected.

If the metered element is dark, i.e. an average RGB Channel luminance of 50, your camera will presume it is metering an neutral (midrange) tone and your image will be overexposed as the camera adjusts exposure up to compensate for the paucity of luminance.

If the metered element is bright, i.e. an average RGB Channel luminance of 220, your camera will presume it is metering a neutral (midrange) tone and your image will be underexposed as the camera adjusts exposure down to facilitate the extra luminance.

It’s up to the photographer to extrapolate a “correct” exposure from these situations. To aid in such a diagnosis, consider the following guidelines:

• The midpoint of a 10-level EV continuum is 0EV, representing accurate, neutral exposure. Something that is a neutral gray (K = 128) with 18% reflectivity in the subject will be rendered to neutral gray in the converted raster when exposed at 0EV
• Something that is not a flat or extremely glossy white will have a reflectance of about 90%. That will make it about 2EV above 0EV for reflectance
• Something that is not glossy or flat black will have reflectance of about 5%. That will make it about 2EV below 0EV for reflectance
• Pure black is about 5EV below 0EV for luminance
• Pure white is about 4EV above 0EV for luminance
• In contemporary DSLR’s, you’ll have substantial low key detail down to about -3EV, and substantial high key detail to about +3EV. That’s a 7EV range, incidentally. These figures presume intelligent prep and talented output as well as accurate exposure. This does not represent their entire capture range or marginal (extreme levels of) performance.

Thus, you might set your mind to adjusting +/- 2EV for reflectance variance and +/-4EV for tonal variance, presuming you are spot metering. Yes, it’s a bit of mental gymnastics…and frankly (purists will want my head on a platter for saying this), probably counterproductive for most aerial assignments because the cameras are so good and the adaptive/evaluative metering so sophisticated that professionally acceptable results are available without all the technical voodoo.

Incidentally, the goal in aerial photography is not to completely avoid clipping, but that’s a dissertation in and of itself. Simply remember that you’re output is limited to ink black and paper white (for prints) anyway.

Also, since we’re dealing with aerials, there’s no reason to explore the Zone System in depth-it’s nearly unworkable in 99.9% of aerial tasking.

Finally, CMOS (linear) captures are not biased toward any of the primaries aside from supernumerary G receptors, although the light they capture may certainly be (color temperature). I have found reliable references that support standardized luminance calculations differentiated between G (~60%), R (~30%), and B (~10%), and others that state only the Mean is used (R + G + B)/3 or (R + G + B + G)/4. If anyone can make input on this issue, please forward a reference (I am fresh out of ISO2721:1974).

L