Traditional backlit animation question

[TVPUSER]

I have a question for those who know this stuff, I recently saw a youtube video that goes into the process of backlight animation,
there are video examples of the process but I feel like I don't have the whole picture.


Go to 2:03

From what I gather he details the process like this (in this specific example)

The scene is photographed normally with the glowing element painted black
The film is wound back
Then the counter matte is placed and filtered light is shined through the open areas

What puzzles me is that, the example shows the flames are not opaque, the background is still showing behind the flames...
If the scene was filmed normally with the glowing element in black, then there would be no way to see through it.

1. Is this scene just filmed normally (without flame)
wound back
the counter matte is placed and filtered light is shined through the open areas

or

2. The scene is photographed normally with the glowing element painted black
The film is wound back
the counter matte is placed and filtered light is shined through the open areas

or

3. Is this scene filmed normally (without black flame matte)
wound back
filming black matte
wound back
filming counter matte

[slowtiger]

Not easy to explain. How familiar are you with traditional analogue filmmaking? Because all the terms stem from that, even if we do all of this purely digitally nowadays.

In cel animation, most or all of this is done in the rostrum camera. (I did this myself in the 90's, spending hours in the dark camera room.) You have the normally painted animation, and you have a mask for the lighting element which is mostly black, leaving a hole where it shall shine, and maybe a counter-mask which paints this hole black, but most scenes are done without because the amount of light through the mask will hide anything recorded on that spot before. Repeat for each glowing elements with a different colour.

First you film the animation, then rewind. 2nd pass is the matte, with a light under the table switched on. Usually you will have a colour filter or gel in front of the camera now. This pass will create some lighting with a sharp edge. Sometimes an additional frosted gel is used for blurring that shine, but that takes some light inside the edge of the matte which may not be desired. In that case a 3rd pass is done with the same matte, but out of focus, or with a slightly frosted gel, and/or some additional effects lens. Have a look at Roger Rabbit which achieved that voluminous look with a combination of these mattes plus a matte in the camerra (bi-pack).

For Tron or Star Wars or anything else live-action the process was different.

First: You only have 1 run of film through the camera, everything else is done on the optical printer. (There are some specialists who are capable of doing a lot of effects just inside the camera, but most prefer the printer because it has more control and you can't spoil the precious negative.) Also in the olden days this involved lots of painted cels, serving as mattes.

(What is an optical printer? Basically it is a camera on one side and a projector on the other side, with an aerial image lens in the middle so you can directly film a film without artifacts from a screen. That's the basic setup, but in the 70's they evolved to 4-head-monsters, having 2 projectors and 2 cameras with a split-beam-mirror on each side, to feed 2 sources at once and create a negative and a matte in the same pass. In between there's space to insert filters and distorting lenses, and you can set the projector out of focus, something which could be animated as well.)

So the actor wears a white bodysuit with the glowing parts painted black, in front of a black backdrop. Black body parts and black backdrop must get seperated first (otherwise the background would glow as well), that's done by cel painters via rotoscoping. (For Tron they had about 450 of them, painting about 200.000 cels.) (I think they could've used the sodium matte process for this, as in Mary Poppins, to automatically create a matte around the actor, but maybe they dismantled the screen for that years ago.)

You now have a negative with white parts where it shall glow (let's call it C for camera) and the rest of the character in shades of gray, and a negative from which we make a positive serving as matte around the body (M1 for matte). (For Tron they repeated this process to isolate the eyes and the face as well.) Both of these are put into the projector side of the optical printer, that's called "bi-packing": you can transport a double layer of film as well as a single one, and this is used for everything which needs perfect registering of elements. On the camera side there's some high-contrast film stock, and with some chemical trickery we get a negative and then a positive where we have only the glowing parts as white holes in a completely black frame, this is MG (matte for glow). Last thing we prepare is an intermediate positive from C, that's P1.

Here comes the fun part. Load the camera side of the printer with some very fine-grain colour negative stock. Load the projector side successively with:
1st pass: P1, with some colour filter to give the b/w character its basic tint.
Rewind negative.
2nd pass: MG, with colour filter, to create glowing parts with sharp edges.
Rewind negative, rewind MG.
3rd pass: MG, with colour filter (same), but out of focus to create a halo.
Additional passes for eyes and face. It's possible to create even more elaborate effects, like a thin line around elements which is achieved by careful overexposure of masks and counter-masks so there's a white rim between them, and this can be used with a differently coloured filter.

Now you have a colour negative with all of these passes which goes to the editing room like everything else.

The guy in the video expresses his personal taste regarding imperfections and film grain, but you can easily emulate most of these effects digitally. I have a video file of real scanned-in film grain which I use often in applying glows and stuff, and since you can animate blur amount and position offset it's possible to create the effect of old footage and worn-off sprocket holes. You just really need to know which result stems from which operation in the real world.


(Source: several articles, and Starlog Special Effects Vol. 4, 1984)

[TVPUSER]

Thanks for the response slowtiger! I only have limited experience with film from disposable cameras and polaroids. I've tried double exposure once so that is where I stand.

So for the cobbler example shown in the video there would have to have been 3 passes then?
Since the BG is visible I can assume a black mask for the flames was not used for the first round of exposures.
after that it is the mask with the holes for the glowing elements being filmed first for the sharp color then refilmed for the halo/blur...
so
1,basic footage
2,Mask only hard light
3,Mask only halo light?

Regarding the optical printer, was it common for animation studios to use them in cel animations for backlit effects?
It sounds like the optical printer is more analogous to modern day compositing but it appears as if it uses more reels of film as a result.

[slowtiger]

Animation commonly didn't use the optical printer since it could achieve all effects with the animation camera. Optical printers were in heavy use for titles, think of all the colourful typography over colour live action footage - that's all done in OP. One of the nicest effects ever was the title for the 2nd Indiana Jones film - where the chorus line performs "Anything goes", the title appears over the dancers, and then one of the dancers breaks out and dances in front of the title - of course that was done by hand-painting mattes for that moment.

As the narrator in the video mentioned it was 8 different flames in that scene, so you had a minimum of 9 passes because each flame had a different colour and exposure, plus a halo pass for each flame which needed one.