dsmeltz wrote:
Yes. And the time and energy cost of trying to adjust 1/3 of a stop goes against the law of diminishing returns. The benefit is you have "perfect exposure" the cost is you miss the shot.
Lab and other controlled tests are nice, but they are based on the concept of "all other things being equal" when the one thing you can be sure of outside of a lab is tha all other things are NEVER equal.
But the whole point of my post about the "error" is that semi-automatic gives you as much control as possible yet yields the probably higher accuracy. So it's LESS trouble, not more and gives equal or better exposure accuracy. That's even more true when the light is changing a lot or rapidly such as on a breezy day with some clouds.
So, respectfully, I think you have the diminishing returns analysis backwards. I do agree that the sum of the factors is never equal. However, any technical or scientific analysis strives to limit the variables and "control for" others.
I have not yet verified that auto ISO is equally accurate compared to auto shutter but that's on my to-do list. When shooting birds I need high shutter speeds and typically am limited to F 4.0 or even F 8.0 when using long lenses. So with an aperture that is forced to wide open and a fixed shutter speed because it's birds, then auto ISO is my choice. That leaves me free to think about the subject, composition, timing, frame rate and so on. For other kinds of subjects, especially architecture and landscapes, other techniques may be better.
So, now the important issue is to determine how much, if any, exposure correction to pre-set into the camera in order to get the best result when exposing with spot metering on the critter. That's what analysis of the middle gray vs blown-out results gives me. So to re-emphasise the point about auto vs manual, user-set exposure compensation in advance of a session is not possible in manual. You can certainly apply a 1/3 stop correction in manual but my data suggest strongly that A. you don't know in which direction to move and B. in variable lighting such as outdoors on many or most days you will be doing a lot of fussing with dials and readings. There are two corrections at issue here: 1. the error between auto and manual and 2. the adjustment you may want to apply based on your meter's calibration.
It's easy to demonstrate that cameras vary considerably on this. It's also possible, with little effort, to demonstrate that the post processing applications that are both raw developers and picture manipulators vary a great deal in how they interpret the
actual raw values in the file. RawDigger is the only way that I have found to do this. For that, FastRawViewer and (for Mac) RPP64 are the least biased. CaptureOne, Darktable, RawTherapee can be tweaked and used skillfully to get near that goal but the problem with them is that unless you have first used FRV or RD then you don't have a valid starting point. Btw and as an aside, Luminar is not showing you the raw file at all. I accidentally discovered that it shows the raw file as modified per EXIF values. That surprised me. I will do a separate post on it, later. I don't have every popular application so perhaps there is another available that can help with this issue.
BTW, I agree with Scottie that if the conditions are conducive to it, then the Sunny 16 technique is excellent and I've been known to use it myself. I did use it a lot more in the days of film. To me, it's a special case of incident light metering. I always thought that incident light metering was the most correct when it could be used. That and a little bracketing and you are almost guaranteed good results. As many of you have said, with modern gear and computer post processing, getting it exactly right is less important than with, for example, color slides.
Speaking of bracketing, many cameras, mine included, can bracket for you very fast and very accurately. Of course, if you are shooting birds either in flight or merely active, then you have to choose between bracketing and catching that perfect pose with bursts.