So let's take a look at this more colourful example to see it working again in practice. Although it's more colourful, the situation is still controlled so we can take our time to get it right. Read the captions for the info...
The flower is evenly lit by natural light and positioned in its vase against a blue background. The camera (again using Evaluative/Matrix metering) has given me an exposure of 1/4sec at f/8 using ISO 100. Remember these are the raw images you are seeing here - I've done nothing with them. Please ignore the inset image on the photos of the LCD panels and only look at the Histogram. Remember, I've had to photograph the back of the camera so only the graph is accurate! Again, with this shot, on a quick LCD review the image looks fine. What does the Histogram below tell us?
Just like on the previous page we can see instantly that there is room to move the tones further to the right. We can also see the interesting split between the RGB channels. There's not much in the Shadows because, barring the hole in the vase, there are very few shadows in the shot. So with minimal exposure change we can lighten the image so it looks like the next shot.
This is more like it. It's not a huge difference from the previous shot but it's enough to make the difference I want. So how much exposure compensation have I added to get a shot that I think is right?
As you can see from the LCD, I've only added a third of a stop of plus exposure and although the Brightness Histogram suggests I can take it a bit further, look at the Red Channel Histogram that is largely representative of the highlights - it's pushed to the far right so, in theory, I don't want to risk adding any more exposure and blowing those highlights. But what happens if I do?
Oh, absolutely nothing untoward happens! The image has lightened fractionally but the highlights are still okay. Take a look at the Histogram (still ignoring the inset image) below for this plus 2/3rds image.
The Brightness Histogram is still showing the potential for another plus 1/3rd stop of exposure. But the red channel Histogram is pushed even further over to the far right so that is has a whole peak lying against the far right wall so I wouldn't push the exposure any more. But it does illustrate that the Brightness Histogram isn't 100% accurate and needs to be used as a guide only.
The good news is that if I chose to push that extra 1/3rd stop in this case and the highlights were blown even further, the chances are I'd be able to bring that exposure back slightly in my post-processing and recover the lost detail. In fact, if I was shooting raw (which I always do) then sometimes I could even get away with blowing the highlights like this...
With two stops over the camera's original metered reading and about 1 stop over the actual correct exposure, you might be forgiven for thinking this image is a write-off. But is it? Let's have a look at it after I've processed it in Lightroom.
All I've done here is pull the exposure slider back to the left by - (minus) 1.65 in Lightroom as you can see from the screengrab on the left. The result is okay, although I'd say the highlights on the tips of the petal are 'borderline'. But because I've gone from overexposure to the correct exposure it hasn't introduced any horrible noise as it would (especially in the blue) if I had been correcting underexposure (too dark).
However, it should be noted this isn't recommended as a general rule and it certainly won't work as well if you are shooting JPEG where the exposure latitude is far less. So this is not an excuse for getting lazy with your exposures either, just a note that if you are shooting raw as you should be, you sometimes have a chance of getting some lost detail back if you 'accidentally' overexpose your shot.