Swatch wrote:
Anyone know of a good link for an
explanation on interpreting
histograms in PS or digital cameras?
It’s an interesting question. The conventional wisdom is that particular objects in the image appear as bumps in the histogram, and the goal is to select an exposure and contrast setting that distributes the bumps horizontally, more or less evenly. At the same time, the wisdom says, you must keep any of the bumps from being compressed against the black end of the histogram (underexposure), or the white end (overexposure). This works perfectly for images such as scanned text or line art, since there are only two well defined bumps, one for white paper, and a smaller one for the black ink.
But this is an imaginary, idealized histogram. For most photographs, the relation between an object and the resulting histogram is more complex than this. Certain objects such as sky, and flat surfaces do indeed generate recognizeable bumps on the histogram, but the majority of 3d objects cast a U shaped, or multi-peaked histogram. Add multiple objects, some with multiple colors, and the histogram is very complex indeed. It may be impossible, even for an experienced person, to distinguish individual objects in the histogram. Step outside of RGB, and the issues change as well, some simpler and others more complex.
In the end, it is always better to base your decisions on the image instead of the histogram. Often the darkest or lightest portion of an image is unimportant. A common example of this is a blown out sky, or less commonly a black background that contains no important detail.
Relying on the histogram too heavily turns you into a human exposure meter, and may cause you to miss out on extra dynamic range that would make the image look better, and that, after all, is what we are all after. —
Mike Russell
www.curvemeister.com
www.geigy.2y.net
