Just read a fascinating article by a post-production specialist who works with photographers on large advertising projects. He was writing about how to handle images with a large dynamic range and offering some advice on "How to shoot wrong, to get the get the right result"
His basic idea was that for digital raw files you should always shoot with slight overexposure.
The reason for this was that raw files have 1 to 1 1/2 stops of latitude for highlights. He sees the problem in images as being the noise created when shadows are lightened. The theory is that using Photoshop -
1) you expose for the shadows (add 1 to 1 1/2 extra stops of exposure)
2) in raw conversion pull back the exposure to bring the highlights back (don't use recovery)
3) now lighten your shadow areas as required
His contention is that when you now lighten the shadows you will create less noise because these areas were correctly exposed in the first place.
I was pleased to read this since its basically what I've been doing for years.
The common advice is to expose to prevent burnt-out highlights. This seems to be almost universal. There is this determination to preserve highlights at all costs, often to my mind resulting in a poorer image.
I first used this technique of overexposing when I used the Kodak 14n full-frame 14MP DSLR a few years ago. This camera suffered from very bad noise and artefacts, particularly in the shadow areas, and it was essential to expose correctly for these. Fortunately the sensor had incredible leeway in the highlights, up to 3 stops. I've never seen this since however.
This image was taken on the Kodak. A very difficult image to expose for. This was how the raw file looked before processing.
As can be seen I was able to restore the highlights, and lift the shadows in the boat without creating a lot of unpleasant noise. Exposing for the highlights would have resulted in an image like this.
Lightening the bottom of the image would have resulted in a great deal of noise being generated.
The theory does assume that the camera sensor has a good dynamic range. However I've used this successfully on all my cameras including m4/3. If in doubt, bracket the shot if possible, as that will cover you.
The writer calls this technique "one shot HDR" and can be seen as such. Often it just isn't possible to take several shots without anything changing in the image. There is a way to do HDR by creating three different versions of the same file. However the above technique does much the same in one shot.
Finally, a word on histograms. While these can be useful, to my mind many photographers become obsessed by them. You should always use your eyes to rate whether an image looks right.
I sell 1000's of images every year, most are published in magazines or books. Every file that I send out either to a library or a client has the following done to it.
In levels in Photoshop I adjust the RGB output level boxes to 5 and 250, instead of 0 and 255.
This takes out a pure black and a pure white, which is essential for printing. A pure white will result in no ink being sprayed onto the paper, resulting in the surface showing through. A level of 250 means that something will be there, changing a pure white into a very light grey. This means that your file is now printable.
