We’ve been intrigued of late by two questions: (1) What is the brightness (reflectance) of an average scene? and (2) How many scenes qualify as average? The obvious answer to the first question is that average scenes are 18% reflective—after all, Kodak’s ubiquitous 18% gray card has served as proxy subject matter for decades. But the card has an odd feature: Sometimes the instructions that accompany it tell a photographer to use camera settings directly as read from the card. Sometimes, however, they say to note those settings, then open up half a stop. That curious proviso, when it is included in the instructions, suggests the 18% gray card is not dark enough. If it were a half-stop darker, which would make it 13% gray, there would be no need to open up that half stop; the reading could be used directly. So perhaps an average scene is only 13% reflective, not 18%? If you Google “average scene reflectance,” you are rapidly immersed in a debate about this very question.
To satisfy our curiosity we turned to Photoshop. Every pixel in an 8-bit digital image has a brightness value (BV) between 0 and 255 assigned to it. Photoshop reports the average BV for all the pixels in an image under the heading “Mean” on the expanded view of its histogram palette. We opened 150 digital photos we had taken outdoors in the Galapagos Islands and noted the average BV for each image. Brighter, more reflective scenes give rise to larger average brightness values and vice- versa—but how can one convert between BV in the image and reflectance (R) in the scene?
Doing this required us to construct what is essentially a characteristic curve for the camera used to take all the pictures. This curve, which is camera specific, plots BV of a scene element, as determined when viewing the scene on screen, to the logarithm of that same element’s reflectance, Log R, in the original scene. It is the digital equivalent of the D-Log E curves familiar from the world of traditional photography. We may discuss in a future issue how to make such a curve and what can be learned from it. The important point for present purposes is that it lets us convert the average BV of a scene to an average reflectance for the scene.
The bar chart presented here details the distribution of R values for all 150 images. It displays the percentage of scenes having various reflectance val- ues. The very low-key scenes, such as those featuring a lot of heavy shade or cut in tight on darker foreground mate- rial, have R values ranging from .05 to .07, indicating they reflected, on aver- age, only 5% to 7% of the available light. The very high-key scenes with broad expanses of white sand, clear water, and bright skies, have R values in excess of .30—more than 30% reflective. But the grand average of R for all of the scenes is 0.124, or 12.4%—very nearly 13%, and decid- edly not 18%. Such is our answer to the first question—how reflective is an average scene.
And the bar chart sheds light on the second question of how many scenes qualify as average. Had we used readings directly from a yet-to-be-invented 13% gray card as the sole basis for exposing all 150 scenes, more than half of them would have been correctly exposed, give or take a third of a stop. Had we used readings directly (no half- stop correction) from an 18% card, only one-quarter would have been within one-third stop of the ideal exposure.
And the remaining scenes that even a 13% card would have missed by more than a third stop? These underscore the need for application of some gray matter as well as a gray card. The content of most such scenes, either heavily shaded or rather dazzling, was unusual enough that even a relatively novice photographer would have recognized the need to apply some Kentucky windage to the gray card recommendations.
So what do we take away from this? There is a good reason for Kodak’s 18% gray card instructions to include the “and then open up half a stop” statement. We were pleased to hear it is again incorporated in the most recent edition of the product.
And if any readers with access to a densitometer want to make their own 13% gray card, it has a reflection density of 0.89, rather than the 0.74 density of an 18% card.