QUOTE (Stephenaweiss @ Jun 6 2006, 11:53 PM)
John, I think I am getting stuck on language. I think we are saying the same. I will bracket by taking a light reading, then taking a second shot with the lens open one larger F stop, ie: 4 instead of 5.6. By doing this I am giving up some detail ( I believe clipping) in the hi-lights, and hopefully gaining them in the shadow and mid-tone.
Do I have that right? I guess I still think in film terms.
That's about right, except that there may actually be no clipping by opening to 4, or even 2.8, depending on the contrast and key of the scene.
I don't know how true it is with the Nikons, but I know that with the more recent Canons, going to a higher ISO while keeping the f-stop at 5.6 will work, too. For probably all digitals that do RAW, exposing as high as possible at a given ISO without clipping gives the least noise, because you are increasing the signal-to-noise ratio. You must have a workflow, however, that preserves the highlights well, to take advantage. RAW is a big plus for this. Depending on the camera, you may or may not get a higher *absolute* signal-to-noise ratio at higher ISOs, like the recents Canons tend to do. In other words, If I set my 20D to ISO 1600, and expose a scene at +2 EC in manual mode, and then take successive shots with the ISO halving for each shot (1600 -> 800 -> ... -> 100), and then use EC in a RAW converter to make them all equally bright, the ISO 1600 image will have less noise, but more potential for clipping. The ISO 100 image will have the greatest highlight range, but will be the noisiest (not subtly, either; quite dramatically). With a camera that behaves like this, if you are working in a paradigm where f-stop and shutter speed are your primary considerations, then the highest ISO that doesn't clip desired highlights will give the least noise. Noise is only proportional to ISO in auto-modes, or in manual mode where constant metering offsets are used, as the ISO then varies the *absolute* sensor exposure.