Although I don't have a D3, I followed with interest the threads Jono Slack started (A personal thank you to RobSteve and Auto White Balance, Nature and why not to do it!) in response to RobSteve's advice about auto white balance in the D3 Color Calibration settings in ACR thread:
It was clear that Jono's posts offered convincing (photographic) evidence that RobSteve's advice was sound. And yet, although I shot some tests with my D300 which were equally persuasive, for whatever reasons -- stubbornness? stupidity? -- I just couldn't shake the Auto White Balance habit. (In my own defence, I shoot a lot of the time in mixed interior light, for which auto white balance usually delivers excellent results.)
Originally Posted by robsteve
A couple of days ago, when I was taking some pictures in sunlight to contribute to the APO-Lanthar 180/4 thread, I realized that I'd forgotten to switch the white balance from Cloudy to Auto. I made the change and continued shooting. Here are two consecutive images:
White Balance: Cloudy
White Balance: Auto
Neither of these looks or feels right: Cloudy is slightly too warm, Auto is too cool.
I can't help thinking that if I'd set the white balance to Daylight, the result would have been pretty close to what I recall the scene being like.
Interestingly, Lloyd Chambers made a similar suggestion in a recent weblog post titled Nikon D3 white balance:
I’ve had generally excellent or at least pleasing results with the auto white balance of the Nikon D3. But in spite of its large built-in database of scenes, I’ve found that it often cannot grok exposures which are dominated by an overall color...
Its variable results are why I often just shoot on Daylight white balance (sunlight or cloudy or shade) in any kind of outdoor scene; I get much more predictable results frame-to-frame.
I hereby make a public commitment to NOT use Auto white balance whenever I'm shooting outdoors.
Just one question, though: both RobSteve and Lloyd suggest using Daylight WB no matter what the outdoor light. Even though it's slightly more work, would one get even more predictable results by matching the white balance to the prevailing light: sunlight, cloudy, or shade? Or is that more trouble than it's worth?