So digital offers a better capture of data if the image data represented by the histogram is biased to the right without clipping. This is because of the way the sensor captures data. Each stop in the digital sensor range captures half the amount of data than the previous stop as you work from light to dark, which means an exposure biased toward the right has a lot more processable data than one where the bulk of the data capture is in the shadows.
This means that an image that may look very bright on camera preview should produce a much better image after processing. This should mean that the image file - working in RAW - contains the best quality information for the image across the greatest dynamic range, and in the post processing stage the image can be darkened to produce a photograph with the maximum detail and minimum noise, therefore having the richest colours. The critical thing is to avoid loosing data by blowing out.
Using ETTR needs a lot of care; constant monitoring of the histogram to avoid blowing out, and to make sure that the whole dynamic range available has data. Where available you can also turn on the 'blinkies' on the image preview in camera which will give an instant visual warning of blowing when reviewing the image.
The idea of ETTR is at odds with the generally accepted principle of the ideal bell shaped histogram; it suggests that the histogram is more a curve from the right.
It is far easier to recover shadow detail in post processing where the image has captured the data than it is to salvage a blown out image. If the exposure has gone to pure white in places there is no recoverable data to use, and so it's quality can be compromised. Small changes in the light conditions can lead to this happening when using this technique so you need to be constantly checking the histogram to prevent this from happening.
A frequent question is why should you do this, as opposed to making an exposure in the normal manner. A lot of argument goes on about the merit of ETTR and the circumstances where it should be used.
The simplest answer is where the amount of processable detail is important in preparing the image so most typically if print is intended and large size print where the effects of noise are readily apparent. Many images used on a computer or the internet will not be viewed at the sort of quality level where this technique will have a visible impact.
So if you are someone who likes to have the option of printing images this technique offers a means of getting the highest quality output from the shot and the risks and time required to manage it are worthwhile. There are some types of photography where ETTR is easier to use, such as landscape work, and some where it is very difficult, the most obvious being action type shots especially when the light is constantly changing.
The technique itself relies on using the histogram and RAW files; the histogram to indicate being blown out, and RAW files for post processing as they contain far more processable data than JPGs. There is however a problem with this and that is the in camera generated preview image which forms the basis of the in camera histogram comes from a JPG preview of the image generated by the camera. This is usually quite a conservative JPG, which will show blowing out prior to the RAW data file actually blowing. Experience will tell you what the factor is of this conservatism; with some cameras it can be as much as half a stop. So images showing a small amount of blown values in preview will in fact be fully recoverable in post processing the RAW data.
Working with ETTR means usually shooting in manual exposure mode and using spot metering rather than matrix / evaluative mode. It is a technique that is best implemented after you have worked out what the leeway is between the in camera histogram and the RAW data in post, which can be established by taking a series of test shots at different exposure settings from a good histogram as a base reference until the RAW data shows blown.
It is also a good idea to work with multiple exposure images so you have several different copies of the same image with slightly different exposure settings.
This technique is easiest to use in areas like landscape photography where you have time to be able to study the images as you take them and to make the corrections necessitated by ambient light changes, or in circumstances where you are fully controlling the light such as in studio.
It requires practice and can take some time to become comfortable with but, if your use of your photos requires the highest quality, this is a technique that can help take you some way toward achieving this.
All Photography and Video Copyright David Hammant.