Original Effects Cleaned up with the help of AI...????

[Metryq]

Then cut to the original effect at the last moment so the fade is preserved.

Ugh! I don't think so. I find it disruptive enough watching the dissolves in older movies where there's a "pop" in the image just as the dissolve begins. (This was done to shorten the number of frames needed for the optical effect, rather than running both shots, before and after the dissolve, in full length. The "pop" is some minor mismatch in grain, exposure, etc. The lab may have done everything exactly by the numbers, but even minor variations in every batch of film stock can make the difference. They ran into a lot of that during the making of TRON due to the high contrast processing.)

If the original footage is still available... we're still talking about a lot of processing work for what potential economic gain? TOS re-re-re-release, the Dissolves Edition!

 

[Tosk]

The "pop" is some minor mismatch in grain, exposure, etc.

There was no way around it in the old days. A dissolve was an optical effect, so the two shots that actually dissolve into each other were one generation down from the film on either side.

 

[Maurice]

Yep. Part of the reason you can spot the difference just before a dissolve is that the dissolve is an optical printer effect that's spliced in, and the difference in grain, opacity, etc. in those frames snap in and out at the head and tail of the dissolve.

 

[Metryq]

There was no way around it in the old days.

Yes, there was, but it was expensive. If the "A" and "B" shots were carried full-length on either side of the dissolve, there would be no "pop." The discontinuity occurred because the few frames of the dissolve were processed separately from the shots on either side of the dissolve.

Once editing moved to video, and those EDLs (edit decision lists) were fed to the film cutter for the final master print, it was much easier to arrange such custom-fitted opticals, like dissolves.

 

[Tosk]

Yes, there was, but it was expensive. If the "A" and "B" shots were carried full-length on either side of the dissolve, there would be no "pop." The discontinuity occurred because the few frames of the dissolve were processed separately from the shots on either side of the dissolve.

That removed the "pop", but it did not remove the quality drop in the two shots that dissolved into each other.

 

[Metryq]

There'd be no discernible drop in quality, as the master would all be struck at the same time. Including the "duplication" (basically a latent composite) of the dissolve.

It's also possible that the "pop" was not as visible in the cinema, and therefore not considered worth the cost and effort. When film is converted to video—at least analog video like NTSC—there is a gamma compression curve. Film had greater latitude (steps between black and white). So a minor offset barely noticeable in cinema (like garbage mattes) suddenly jumps out and tries to strangle you in video.

 

[Tosk]

There'd be no discernible drop in quality, as the master would all be struck at the same time. Including the "duplication" (basically a latent composite) of the dissolve.

What is a latent composite?

 

[Metryq]

Latent composite: multiple exposures of the same piece of film before development.

Using the TMA-1 scene from 2001: A Space Odyssey as an example, the live-action set was shot first. A box-like excavation in the lunar surface with ramps leading down to the "monolith" in the pit. The view from the rim of the pit was shot, and then stored without development for many months. Later, models or matte paintings (in multiple exposures) were added to show the distant lunar surface and horizon, with stars and Earth in the sky. Naturally, portions of the frame for each exposure must be "matted" or masked out to prevent ghost-like see-through double exposures of the same part of the frame.

The usual method would be to shoot the live-action scene, and develop it. A projection of that scene would then guide the artist in painting the rest of the scene on glass (a "matte" painting). Thus, the live-action would go through one generation loss. Re-shooting film means you are also taking in the grain of that first recording, rather than shooting a "grain-less" real scene. By taking the risky approach of latent exposures, Kubrick got a much cleaner shot.

Part of the impact of this approach is the "latitude" (dark-to-light scale) of the film vs reality. For example, David Lynch shot his desert scenes for Dune with a special light box attached to the camera. The box had a variable light source with a colored gel to bounce light into the lens. By using an amber gel, he was able to "fill in" the "blacks" of the scene to give an unearthly light. By doing this realtime on the real scene coming through the lens, the color "grading" was much more subtle than if it had been added in post production with the much more limited "latitude" of the film. Doing the effect in post would tend to wash out the scene.

(Such subtle color grading is now very easy with digital processing. In fact, digital compositing has been a godsend to VFX artists.)