I know this question has been asked in variations a number of times, but it’s still a bit of an unanswered question to me in my specific scenario.
My rolls are usually shot over more than one day/session and therefor contain different scenes with more often than not wildly different conditions. The frames can be a mix of perfectly exposed blue sky daylight exposures, overcast ones, small window natural light indoor ones and somewhat underexposed artifical light ones. Ocassionally there will also be some flash exposures in dark environments in there.
When scanning a roll like this in Vuescan as raw dng, should I lock exposure on the first frame regardless, find a frame that sits roughly in the middle of these conditions and lock exposure for that one or lock exposure per scene that has significantly different conditions than the one before?
I’m scanning with a plustek OpticFilm 8200i and unlocked Vuescan will usually set exposure somewhere between 1.7 and 3.5 with probably 80 % of frames being between 2.0 and 2.5. Conversion in NLP is done with roll analysis for the whole roll after white balance has been picked off one frame border and synced to the whole roll.
I lock exposure on a part of the film that has no image, like between the frames. That way you get consistent lightning, and also lock it for all the roll. If I scan several rolls on one session, I lock exposure on each one.
I’m camera scanning exclusively, but I suppose that the following should be applicable to scanner output as well.
Films that I filled in one go can profit from locked exposure and Roll Analysis.
Films that took several days and locations can profit too, but don’t do by necessity.
NLP will read the negatives and convert whatever you throw at it…with results that might - or might not - suit as decent starting points for further development/customising. My best practices are:
scan with locked exposure unless histograms huddle to the left, then, expose individually.
convert a series of virtual copies with Roll Analysis and one without RA.
pick what is closest to needs and proceed from there on.
My colour negatives are up to 50 years old, slides and B&W are older. They are what they are and NLP does a decent job in many to most cases. And occasionally, I have to search for “good” colors - or fall back to B&W.
Last words: Many conversions don’t need a lot of tinkering, and some need more attention. Sometimes, RA delivers suitable results, sometimes they improve without RA. Some negatives turnout “better” without setting WB before converting etc. Try/test & learn!