Optical compositing

Can anybody explain me what is the exact process of optical compositing or at least tell me if I understand it correctly?

Let’s say it’s the pre-digital age and we want to shoot a shot of a model in space.

I get we have to film the “background plate” - the space, and we shoot the model in front of a blue screen. So what next? How exactly would we proceed?

I imagine it went something like this :

1. We film the background
2. We film the model in front of a blue (later green) screen
3. We filter out the blue screen so that it becomes black - how?
4. We now have a starship with incorrect colors on black background. We make a black and white copy of it (from positive to positive? P -> N -> P ?)
5. We have a shot where the ship is white and the background is black. Correct? Now we do a negative copy of it.
6. Now we have a new shot where the foreground is black and the background is white. What now?

After point 6 we should have
a) background shot
b) bluescreen foreground shot
c) black and white shot with white foreground and black background
d) black and white film with black foreground and white background

What are a and b? I guess you don’t want to touch the original negatives, so are we working with interpositives made from the original negatives?

Now how do we combine it? If I understand it correctly, the black is the matte and it basically means that the black parts of the film will not be exposed. Correct? So what’s the white part for? I know that we need to put the 4 film strips into an optical printer and “somehow” it will produce the correctly combined shot.

So the next logical question is in which order should the film strips be “layered” in the printer? That’s the part I don’t quite get - so are the white parts of c) and d) basically transparent and the black parts opaque?

And the final question - how many generations from the original negatives would the final composite shot be?

Thanks!

A and B are YCM duplicates on Estar base for maximum alignment. In short, 3-strip positives are what goes into the optical printer.

White parts of C and D are effectively transparent, but of course the black parts aren’t perfectly opaque (or we wouldn’t be seeing garbage matte artifacts all over the OT, now would we?).

Ordering the elements is carefully planned as the optical work-- essentially a very fancy double-exposure process-- can take up to 10 hours (maybe longer) depending on the complexity of the shot. Using just one background plate and foreground element, it might go like this:

1. The B&W matte and inverse matte (just calling it that for the purposes of this post) are put into the printer to calibrate alignment.
2. The YCM background element plus inverse matte (where the foreground object is black) is photographed in the printer.
3. Film is rewound. The YCM foreground object plus matte (where everything around it is black) fills in the blank, unexposed part of Step 2.

For the OT, plates were filmed in 8-perf 35mm (VistaVision) and the final composite would be directly photographed onto 4-perf anamorphic 35mm in the optical printer. This is considered one dupe generation with the only steps being Negative(plate)-YCM-OCN(composite).

Thanks! I have a similar question, hopefully you’ll be able to answer 😃 How about rotoscoping? I think that both Star Wars and Star Trek used this, Wars for lightsabers and Trek for phaser fire, how was it done? I honestly doubt they painted over a 35mm film cell, so I imagine they first “developed” the needed frames to some bigger “prints” - big enough for them to be able to draw over them, and then photographed them again ?