logo Sign In

Post #567356

Author
TServo2049
Parent topic
Star Wars Colortiming & Cinematography (was What changes was done to STAR WARS in '93?)
Link to post in topic
https://originaltrilogy.com/post/id/567356/action/topic#567356
Date created
29-Feb-2012, 2:56 PM

-1 and cinch have a theory that the early-80s transfers have the most accurate colors. I was never sure about that - judging by stuff like the lack of the Tantive burn marks, my theory was that a new low-contrast IP was made from the untimed negative, one that would transfer better to 1" analog tape on early-80s telecine equipment.

Now, I'm not so sure. Obviously, tweaking was done so it would look better in the analog video realm, and the image is often way too bright with very flattened contrast. But now I'm wondering if the film source for the pre-1992 releases did in fact have the original timing.

As a test, I took the Binary Sunset image LexX posted back in page 1, which he captured from a "1983 rental video" (possibly PAL, judging by the slighty squished image, indicative of an uncorrected 720x576 PAL capture). I tweaked it in (of all things) Microsoft Office Picture Manager. All I did was turn down the midtones and shadows, turn up the highlights, and turn down the saturation a little.

First the original screenshot, then my version:

Then, as another test, I tweaked the Japanese Special Collection image msycamore posted, also back in page 1. This is from (I think) the Dark_Sega capture, the red channel is bleeding to the right, but look what I was able to get out of it. This time, I turned up the midtones and highlights, turned down the shadows, turned down the contrast (to compensate for the extremely high-contrast result), then increased the amount of red (hue 0) by 10 points and brought the saturation down a couple points.

Original image:

My version:

And the results for both are very close to what Mike Verta says the scene looked like in the IB Technicolor print screening. I was stunned that I could adjust these old 80s video transfers and get a result that approximated the '77 timing. The colors seem to match; it's only the brightness, contrast, and gamma that are completely different.

UPDATE: I decided to tweak the other pan-and-scan images LexX put up

I know that the stars are gone; this is meant to show the color accuracy, even if I have to crush the detail to do it

Again, ignore the crushed/posterized/solarized details on Leia's face and elsewhere. I'm not concerned if the contrast looks weird or if details are crushed out. I'm trying to bring out the colors that the transfer's flat contrast and gamma boosting obscured.