How Important Is The Histogram . . . Really?

[obsolescence]

A subject with a relatively flat, even surface and limited range of tonal values, such as a wall, is the least challenging to any camera's metering system and to the histogram usefulness. Metering options are designed to be accurate in certain commonly encountered lighting conditions, and they can work quite well if you follow the manufacturer's guidelines. But often when you shoot in "real world" complex lighting situations, metering inadequacies become pronounced, in terms of errors in predicting whether you're capturing the highlight values you want to have detail in, up to the point of clipping. The histogram is easily fooled by strong backlight, for example. A better test would be to shoot a wide open area with strong backlight (sunlight), to see whether the camera's histogram differs from the Raw histogram from the capture.

The problem is that the SPP histogram is small and the camera histogram is tiny. Real-scene highlights are usually only a very small part of the image, and the right hand end of the histogram trails off into a trickle of values that are almost impossible to see. Raw Digger and Image Analyser both have better histograms.

Very true. One of the main situations where highlights cause misleading cam histograms is when light sources like reflections of sunlight off shiny objects cause a spike (as seen in RD) that might lead one to underexpose, when in fact we may want those specular highlights to blow out anyway. Preserving highlight tones too far into the brightness range of the subject leads to overall dark tones in the rest of the range, which brings excessive noise when corrected in PP. A large proportion of the Sigma Raw files I've worked with (shot by others) have this problem. It is often a trade-off to be determined by metering technique.

I suggest a good subject for testing daylight exposure is a landscape scene with front-lit bright white clouds, to witness how the camera's histogram should look when detail in the clouds is optimally captured -- as much exposure as possible without any blown-out spots. This is an example of where spot metering can help.