Okay, I see...
So my personal rating of a film stock is just for the calibrations of my light meter / exposure tools?
Yes, really. In pre-production you do exposure tests, shooting the same test scene(s) (representing what's the most important in your film, like skin, costume, maybe snow or foliage) with normal exposure based on EI written on the can, then with +1/3, -1/3, +2/3 stop and so on. Then you process and scan on the exact same system you'll use in post. Project and choose a base exposure which gives you the best image and see how much you can deviate from that (what's you usable latitude).
For example, you found normally processed 5203 looks best on Arriscan exposed 2/3 stops over (real-world example), so you rate it EI32.
That's the "proper" way of finding the EI but considering the huge latitude of modern neg stock you'll never be way off with a recommended EI. And with basic sensitometery knowledge and some experience you'll pretty precisely predict what over-under exposure and push-pull will to to your image.
So in other words, me telling a lab to push 1 stop means they will simply aim for another density in my negative and nothing else?
They'll simply run their machine slower, so that neg will ideally get 0.15 points denser.
0.15 because exposure difference of 1 stop translates to optical density difference of 0.3D and with a coarsely 0.5 gamma (which means 2x tone compression) of a negative it equals 0.15D.
Sometimes you desperately need that density.
Say, you shot the leading lady on 5219 rated 500T at you Zeiss Standarts' maximum T2.1 but there wasn't enough light and her face read T1.4 on a spot meter. You don't want her face whole 2 stops under because it'll be like 0.4 density (it's murky mess which gets you fired).
So you either request a push-1 and get a 0,55D with reasonable grain/contrast or do a push-2 and get "perfect" 0,7D, albeit very contrasty and grainy.
When you go for a "pushed" look, it's not about the density but contrast and grain.
BTW, pull-process is used much less and mostly for "artistic" reasons rather than technical. Overexposure doesn't really ruin your images unless you hit 4-5 stops over.
What if a roll has different densities across?
As long as all the shots print/scan well and yield a quality image, it's OK.
If not, and you're stuck with a mixture of over- & underexposed footage on a roll, you decide how much UNDERexposure you can tolerate on the underexposed scenes and PUSH accordingly. Don't worry about overexposing other stuff on the roll unless there are important details shot 4 stops over (say, window views). But why did you do that if they're important?
Usually you keep some densities (like actor's faces) consistent and ensure there's always some black and some highlights in any frame. But everything other changes from one lighting setup to another. It's OK for a subject to render different densities under different light.
Even a face can go from -1 to +2,5 stops relative to middle gray in a really bold movie (but unlikely within a roll). Just make sure the audience sees the eyes (at least some reflection in them) in your portrait CUs/middle shots.
Is that when best-light or scene-by-scene developing comes to play?
There's no best-light processing, only best-light print and telecine.
If you're in real trouble, you can try to splice your roll into parts and process them differently. Theoretically possible given you had a nice complete camera report. Don't expect your lab to do it!
If I were to expose a negative correctly for its native ASA and tell the lab to push 1 stop - my final result would be a somewhat overexposed image?
It'll be 0.15D more dense. Plus grainier and more contrasty, won't have a noticeable "overexposed" look (like overexposed video would).
Edited by Michael Rodin, 06 November 2016 - 09:44 AM.