>>4450154Histograms show you the amount of pixels that fit into that specific tonal value. If I take a photo of a blackboard, the histogram would be a big lump at the left side because everything is nearly black (technically a whole bunch of shades of gray). Actual photos are more complex and using the histogram can tell you how bright or dark your image is being recorded as, without relying on ambigious things like your LCD screen or your settings. All the pixels that are on the left most edge (and only the edge itself) are completely black, and everything on the far right edge is completely white. Normally that means you've crushed the blacks and cooked the highlights until they have no actual detail; just black and white. It is generally good to avoid this. This is why the "perfect exposure" has a wide lump focused reasonably well in the center of the histogram, and nothing touches the edges.
But, if you have true black in your scene (say, a night shot with fuck all lighting) or light sources (the sun, a lamp, reflections off shiny surfaces) you *should* have some part of the histogram showing some pixels in either of those values.
The center of a histogram is 18% middle gray, which if you use evaluative or center weighted metering, is what the camera is going to try and aim for. The reason we sometimes want spot metering is that the camera will try to expose for the whole image, which is sometimes not possible (limit of DR) or not what we want (see
>>4450166).
Spot metering lets you meter for a very small region at the center of your shot, which you can aim at different areas and get given different EV values. So you can feel free to focus on the important parts of the image you wish to nail exposure on. This is a very handy technique combined with the 'Zone System', where you can roughly tell what EV value a given surface should have and test your exposure by pointing at different parts of the image instead of trusting the camera software.
>pic rel 
