Ghost in the Shell

[MattMahdi]

Fascinating and disturbing reading. I first saw GITS many years ago and honestly had forgotten the details.

Then came the live-action remake and I actually enjoyed it. It wasn't great, and actually had some significant problems, but it got me interested again in the original.

After I found and rewatched the anime (thank you, Walmart), I was truly surprised at the number of shots that were remade.

And now I find out that about all I can be thankful for is that I didn't purchase the 2.0 version of the anime. I'll look forward to seeing what you people come up with for a more definitive version, and I'll keep reading from curiosity.

[BronzeTitan]

Here's some info from the GITS 2.0 Blu-ray . .

GITS 2.0 Blu Explorer files listing

GITS 2.0 Blu VLC codec listing

Note that all the video files are 24 FPS [yellow highlighted], except for one [blue highlighted] at 30 fps.

The continuous animated menu for the Bonus Features (right-click, then "view image" for it's full size) . .

And a screenshot from the "GITS ORIGINAL MOVIE" to compare it's cropping against the other releasess (right-click, then "view image" for it's full size) . .

[BronzeTitan]

Looking closer into the "interlace mess", here's a revision of my earlier observation about the "tracking graphics" opening shot. Even though the interlacing is messed up (more below), I'm now thinking the tracking shot was a deliberate, film-time extension technique to save expense. They used the interlacing concept to generate more screen time with less labor. If the multiple ghost images were the result of true telecine, the ghosted figures would show on alternate scanlines. Instead, the apparent scanlines are actually part of the graphics, not real scanlines, and the tracking graphic ghosts always follow one-another on the same "scanlines".

Regardless, the interlace is messed up, but I've seen it before (more on that later). Normally, it should be a clean process when telecining 4 frames of actual film at 24fps into 5 frames of video at 30fps . .

This should be the present state of the GITS Original Movie. It should have clean alternate scanlines that correspond to the frames of the above chart. When inversion-telecined to correspond to the chart below (note: duplicate fields generated to create the extra frame above,  are discarded below), the source image should return to it's clean self, as if it had never been altered . .

So the first thing to do was examine a sequence of 5 frames. According the the chart, the 1st, 2nd, and 5th frames-fields should be the same as in the original frames, while the 3rd and 4th frames-fields are composed of various fields. BTW, the organization of the pictures below is a thumbnail of the interlaced frame, followed by a thumbnail of the even field and it's cropped but full-size self, followed by the same thing for the odd field ... for all 5 frames.) . .

GITS Original Movie - frame 1 :

GITS Original Movie - frame 2 :

GITS Original Movie - frame 3 :

GITS Original Movie - frame 4 :

GITS Original Movie - frame 5 :

As you can see, only the fields of frames 3 and 4 (see the edges of Major Kusanagi's white vest) have double images from the camera-tilt-down. I've seen this sort of thing before and guessed it was from older optical telecine equipment. As the film ran through it, the scan caught the fields out of sync (from timing adjustments?). On that example, I saw that only half the field was cross-contaminated by another field image, to give it that double exposure. This must happen when scanning film to professional video tape, indicating that the DVD (interlaced, 30fps) was made from such tape.

There is a fix for it. Way back when, I worked up a proof-of-concept for that problem and left it at that. It was simplicity itself -- inverting and superimposing only that half of the contaminating field onto the field to be fixed and done! Oh, it did require restoring the brightness/contrast of the finished field to compensate for the flattening effect of superimposition.

[spoRv]

There is a fix for it. It was simplicity itself -- inverting and superimposing only that half of the contaminating field onto the field to be fixed and done! Oh, it did require restoring the brightness/contrast of the finished field to compensate for the flattening effect of superimposition.

Eagerly waiting for the fix!

[Colek]

Would love to see your fix, there are plenty of such releases sadly (John Woo's The Killer US Bluray for example).

[zoidberg]

'The Original Movie' bonus feature on the 2.0 disc is a hard telecined master (from what looks like an analogue source). Unfortunately it's also heavily/poorly compressed so artifacts are everywhere, even with a great IVTC like TFM you will get frames bleeding into one another. It does have 2.0 Japanese/English PCM tracks and the subs may be different (I haven't compared).

[BronzeTitan]

... fix ...

... fix ...

Hints?     OK, ok.  I'll work up (fingers crossed) the proof-of-concept again ... "with circles and arrows and a paragraph ... explaining what each one was".¹


¹ "Alice's Restaurant" [or "The Alice's Restaurant Massacree"] by Arlo Guthrie
    YouTube: Alice's Restaurant - Original 1967 Recording

[BronzeTitan]

'The Original Movie' ... is a hard telecined master ...

So that means there is no progressive master to go back to (excepting to reconstruct the entire movie from all the original elements, if they still exist)?  

Hmmm ...      ... maybe I should patent this very complex and time intensive procedure that takes specially constructed and delicate scientific equipment to process in an atomic-bomb-proof shelter to keep the gamma rays from throwing off the sensitive readings ...

Well, guys, I gotta go and become a millionaire! See ya!   $$ $$

[zoidberg]

BronzeTitan,

I'm referring to the version that's on the 2.0 disc as a bonus feature.

[BronzeTitan]

Right. So the 25th Anniversary release (24fps) then is from the original, progressive (24fps) source?