· 3 min read
Center-weighted and matrix metering patterns
How camera meters average a scene with center-weighted and multi-zone matrix patterns, where each fails, and when an exposure override is warranted.
Written in by Simon Lehmann Editor
The rear-screen histogram is the most reliable exposure feedback a digital camera offers, more trustworthy than the brightness of the preview image, which shifts with ambient light and screen settings. But the histogram is a specific kind of map, and reading it well depends on knowing what its two axes actually represent and where the displayed data comes from.
A histogram is a bar chart of tonal distribution. The horizontal axis runs from black on the left to white on the right; the vertical axis counts the number of pixels at each tonal value. In an 8-bit rendering that axis spans 256 levels, from 0 to 255. The height of the plot at any point is simply how many pixels carry that brightness.
The horizontal axis is not linear with respect to scene luminance. The displayed values are gamma-encoded, which redistributes tonal levels to approximate how human vision perceives brightness, a relationship that itself follows roughly a power law near luminance raised to 0.45. This is why midtones occupy a generous portion of the graph while the brightest stops are compressed toward the right edge.
Two failures are read directly from the ends of the graph. When highlight values pile up against the right wall, those pixels have reached maximum value and recorded no detail: they are clipped, and no processing recovers what was never captured. A spike hard against the left wall indicates blocked shadows, tonal information crushed to pure black.
A distribution that simply leans left or right is not in itself an error. The shape that suits a low-key subject differs entirely from one that suits a snow scene. What matters is whether tones the image needs have been pushed off either edge.
The histogram on the screen is not computed from the raw sensor data. It is derived from the embedded JPEG preview the camera builds from current picture settings, including contrast, saturation, and white balance. That preview has already been tone-mapped and white-balanced, so its histogram clips before the underlying raw file does.
The practical effect, as Jim Kasson notes in his work on in-camera histograms, is that the display is pessimistic about highlight headroom: avoiding clipping in the JPEG histogram usually leaves a margin of real, recoverable detail in the raw file. How large that margin is depends on the light’s spectrum and the white balance applied. Raw shooters who expose to the right are working against a histogram that consistently understates how far right they can go.
· 3 min read
How camera meters average a scene with center-weighted and multi-zone matrix patterns, where each fails, and when an exposure override is warranted.
· 3 min read
Why removing the color filter array raises a digital sensor's resolution and sensitivity compared with desaturating a Bayer color file to grayscale.
· 3 min read
How and when to bracket exposures by full and fractional stops, how to set the spread for film versus digital, and when brackets serve as insurance or as blending source frames.
The grainmag companion app
Meter and place your tones without a signal. No account, no internet required — just you, the light, and the grain.