You have two types of histogram: an overall luminance histogram, and a colour channel histogram. More about that later.
The axis are:
X axis = black at the left hand side, through to shadows, mid tones and highlights, then pure white on the right hand side. The extreme left represents pure black, and the extreme right is pure white. The scale is 0 to 255.
Y axis = The proportion of the image that is that bright, according to the X axis. I don't know the exact units, but a larger peak = more of that type of light/colour (more later).
A histogram is telling you how much of each type of light is in your image. It's probably the single most useful thing when it comes to reviewing an image you just took. Colours, white balance etc can be changed later, but a badly exposed image is much harder to correct. Zack Arias said he shot a whole wedding with 80% of the screen taped over and only the histograms showing to check exposure.
You can work out a lot about an image depending on the histogram.
An image where the histogram is larger towards the left contains a lot of black and shadow. It may be underexposed, or it may be correct, depending on the subject. A night time scene will have mostly shadows, with spikes of highlights (from streetlights, the moon etc)
And obviously an image with the majority of the histogram piled to the right contains a lot of mid tone and highlight. It may be overexposed, or correct.. a white wedding dress against a window won't have many shadows, but that isn't a bad thing.
You can also see the amount of contrast in an image. Something with a U shape has a lot of black shadow, less mid-tones and lots of highlights (i.e. high contrast). Something a ^ shape has a lot of mid-tone but less black/highlight. It is low contrast - again, not a bad thing necessarily.
If anything is touching the edges of the histogram, it is pure black or pure white. This is *generally* a bad thing.. these areas contain no photographic detail, but again it can be fine depending on your photograph or taste.
Colour histograms are a little more complex - you can split up the luminance histogram into blue, green and red channels. In sum, they total up to the luminance histogram. You can have a luminance histogram which looks fine, but you can blow out detail in colour channels, thus ruining the colour of an image - i.e. you have an area with pure green, red or blue channel. A summer field will have bright blues, darker greens etc, but the luminance histogram may average out in between and look low contrast. You have low *luminance* contrast, but still have colour contrast between green and blue.
A good guide of reading colour histograms is here:
http://www.kenrockwell.com/tech/yrgb.htm
Also, you can find a good guide to standard histograms here:
http://www.kenrockwell.com/tech/histograms.htm
Yes it's rockwell, but this is a straightforward guide with no bullshit.
Rockwell is only wrong about one thing. In modern digital cameras it is better to "expose to the right" - i.e. if you have a high dynamic range image (black shadows between trees but a really bright sky and white fence posts), it's better to slightly overexpose, so you can get some detail into those shadows. Then you can recover highlights in software. I can get 2 stops of recovery from my camera when in raw mode. Pikel, I think you have a d7k, which has the same sensor as me, so you should get two stops of recovery too. Older cameras can recover less.
Of course you can bring up black to get detail, but you will generate a lot of shadow noise.
Hope that was useful.. I'm sure cfk will chip in too, and we will overlap in some places and get some new info too
variety.
