Cropping the Original Trilogy : 35mm vs dvd (gout) — Page 3

Yeah, I agree with msycamore. 5% cropping seems like the golden mean. :)

Here's where my estimate of the Senator cropping fits into that diagram.

Good grief, look at how much they took off the right! (I'm willing to bet that's the edge where the worst platter scratches were...)

All in due time. I'll be doing more comparisons when time permits.

My guess is that it's some kind of light spill from the saber glow element - that is, the film frame was just slightly larger than the opaque black Kodalith, so the animation stand backlight leaked around the edges. Then, during optical compositing, this backlight spill would have been burned into the darker, less exposed edges of the live-action element.

I believe this sort of "backlight spill" also occurred in the electricity animation when the Jawas knock R2 out. It was brought up in the thread for DJ's project - here's the picture that was posted there:

Yes, notice that the color of the bright edge garbage is basically the same color as the saber glow. That's why I think it was burned in with the glow itself.

Here's another cropping comparison, this time of the binary sunset wide shot.

Green is 0%, blue is 5%, red is 10% (thanks msycamore); cyan is GOUT, magenta is JSC, and yellow is Technidisc. (Sorry, I forgot the 97SE and '04 DVD/ '11 Blu-ray. There'd be too many colors to keep track of, anyways...)

It's interesting to note how open both the GOUT and JSC are for this shot. They're both around 5%.

The Technidisc, however, crops more than 10% off the sides - but as it's more like 2.25:1 than 2.35:1-2.40:1, it has more on the top. A LOT more. It seems to come very close to the edge of the frame.

Not sure how much this reflects the cropping of the transfers as a whole. We already know that some scenes in the GOUT are dramatically "zoomboxed."

Maybe I'll try one of those lightsaber shots next.

A bit off topic, but speaking of weird things appearing in the frames, here's 3 frames from the battle of the death star, a closeup of Biggs in his cockpit. Does anyone know if this was in the original prints as well?

What exactly are we supposed to be looking at?

I do see that in the second frame, the Death Star footage doesn't look like it was fully exposed in the composite; it gets darker and more blue as it gets closer to Biggs.

I know about blue spill, where the glow from the blue backing spills around stuff in the foreground, but I can't recall ever seeing blue from the bluescreen footage spilling onto the background element. (Sort of a reverse blue spill?)

There's also a smudge out the window near Biggs' nose. The way it looks semi-transparent makes me think it's on the DS background element, and shows up semi-transparent because it's in that weird blue-spill zone.

I checked Puggo Grande, and the background "blinks out" on that frame there, too. Even with the fading, you can make it out:

And here it is in the Moth3r boot. You can faintly see that smudge, too:

Well, I'm no expert on optical compositing, but I've read the 80s ILM coffee-table book, Christopher Finch's tome "Special Effects: Creating Movie Magic", and Ray Morton's book on Close Encounters, so I have a basic understanding. (Which is why I also could also come up with a theory on that weird Biggs frame You_Too found.)

Well, it probably doesn't apply to every scene. As I find more screens of raw frames with sufficient detail to match up the edges of the various transfers, I'll do more.

Thanks for those frames TServo, that confirms it's an FX "error" or whatever to call it.

OK, here's one of the lightsaber frames -1 posted earlier (albeit brightened by me). Again, I combed together previous screenshots from the forum as my references - thankfully, I could find plenty of images of this shot.

Again, I didn't include '97 or '04/'11, as the image is pretty small and extra lines would just clutter everything. (And also because the scene was recomposited in the SE anyway.) Same color coding as before - blue is 5%, red is 10%, cyan is GOUT, magenta is JSC, yellow is Technidisc.

The aspect ratio of -1's screenshot isn't 2.35-2.40:1, it's closer to 2.20:1, but I can't tell if it's cropped a bit, or vertically stretched a bit. I included the 5% and 10% anyway, even though the actual 5% and 10% cropping may be wider.

Again, the horizontal cropping of the GOUT and JSC fall in the 5% zone; Technidisc seems to be halfway between the 5% and 10% lines. The vertical cropping is a different story: GOUT's is near the 10% line (I wonder if they were trying to hide the aforementioned bright areas on the edges?), while Technidisc is close to 5%, and JSC is even less (which is why the glows are so visible on those, JSC in particular - the gamma boosting didn't help either).

This time, it's the JSC that has the least cropping - probably too little cropping on the top and bottom...

I'm not sure if the cropping is *completely* random. I will warn that the 0%-5%-10% boxes in my earlier comparisons may be inaccurate, since some of these screens were cropped or stretched by -1, and may be the wrong aspect ratio or not show the 100% complete frame area. Also, I had to do some guesswork with the binary sunset comparison, since there are no "landmarks" at the top of the screen.

I have your Technidisc rip plus the JSC V8 Project burned, but thanks for the offer.

I'll do the Han chase scene if -1 posts uncropped LPP frames where the whole border is visible. I don't think I can work with the faded frames, they seem to be crooked or bowed, and the gate edges are cropped out.

In the meantime, here's Luke and Ben. Same color coding as before. This time I sourced it from an image with soundtrack and sprocket holes (I cropped off the left sprocket holes). I lined it up with the SMPTE framing chart scan I used to make the 0%-5%-10% diagram, including the blank soundtrack space, so the boxes should line up perfectly:

Again, GOUT is pretty close to 5% cropping, but it's shifted to the right. The JSC is cropped on the sides a little more than the GOUT, and the Technidisc is cropped even more, but both have more on the top and bottom. The JSC basically touches the top of the used frame area (you can even see a bit of the rounded edge in the top right-hand corner), while the Technidisc comes very close to the bottom.

I will point out that both the JSC and Technidisc transfers share a trait of earlier widescreen transfers to crop more on the sides than, so the image would read better on smaller TVs. GOUT is cropped on all sides, but in general it has more on the sides than the other two even if it has less on the top and bottom (and as I demonstrated before, sometimes this is a good thing).

Here are some more 0%-5%-10% comparisons from this part of the film - I'll do the GOUT/JSC/Technidisc comparisons in the future.

Notice, by the way, that some shots occupy the entire usable frame area instead of the standard Panavision aperture. Don't ask me why, I have no idea. For those scenes, I'm guessing that the picture outside of the green border never showed up in any theater.

Here's the 2x stretched SMPTE RP-40 framing chart I use for reference (reduced, the full resolution is much higher), complete with the blank space where the soundtrack is located. (You can see my 0%-5%-10% boxes. I drew in the dark green "target" lines when I was making it, so that I could center the 5% and 10% boxes.)

The full usable picture area is everything to the right of the soundtrack space. The reason I don't think the area to the right of the 0% box ever showed up in theaters is because it falls outside of the SMPTE framing area (the checkerboards), and the projectionist would have lined up the framing so that the image on the screen fell inside the checkerboard.

I'm not sure how many more scenes in the film use the entire frame, since -1 hasn't shared many complete frames including the soundtrack and sprocket holes.

Hey -1, could you do more screenshots like these, just full-frame scans with soundtrack and sprocket holes, and 2x horizontal stretch?

negative1 said:

very cool pictures and comparisons.

by the way, my aspect ratio is off on a lot of pictures,

but that doesn't change the overall percentages i posted,

which is just for comparison. when i do crop the sides, you

are seeing the entire picture unless i mention it.

I was talking about my percentages. They aren't based on the actual amount of exposed image per se, they're based on the .839 x .700 aperture that was the standard for Panavision at the time. I am not showing 0%-5%-10% cropping of the total used picture area, I'm showing 0%-5%-10% cropping based on a projector calibrated to the 1977 standard Panavision aperture.

As I just demonstrated, some shots actually use the maximum sound aperture, which falls outside of the SMPTE protected area. A comparison of the GOUT to the entire exposed image will always show a higher percentage of cropping for those shots; for example, even if the telecine operator kept the framing in exactly the same place for this whole sequence, the closeup of Ben and the Mos Eisley matte painting would have a higher percentage of cropping compared to the full exposed picture, because the amount of exposed frame increases.

Your images do basically contain the entire picture. However, for my 0%-5%-10% comparisons, I need to be able to see the black space around the exposed image on all four sides. Looking at many of these full frames with soundtrack, the camera exposure seems to "trail off" on the left - sometimes you crop this off. I also need to be able to see how close the right edge is to the sprocket holes, because of the scenes that use more of the frame; for those shots, even 0% cropping at Panavision aperture would take off some picture on the right side.

I'm trying to demonstrate the maximum and minimum amount of image that would have been seen in a theater in '77, to provide a guideline for how you should do your cropping. Thus, I need fixed points of reference - that's why I would prefer versions with the complete frame, including the top and bottom frame lines, the soundtrack and the sprocket holes.

I can wait for more images. I'll still try to do GOUT/JSC/Technidisc comparisons, but until I get more full frames, my 0%-5%-10% will either be guesses, or will be 0%-5%-10% of the entire exposed image, or I just won't include them.

I've been revisiting the scene of Han running after the stormtroopers, and I don't think there's really a zoom in the GOUT. -1's previous comparisons weren't completely accurate. The test captures he did seem to be stretched out too much vertically, besides being crooked/distorted, while his GOUT screenshots seem horizontally squeezed a little too much. He didn't perfectly match the proportions of the two images, and I can see places where the two sources are out of alignment.

I tried stretching my own GOUT screenshots over the same frames from the '04/'11 transfer (taken from Harmy's DeEd 1.0). I was able to come up with a rough border of where the GOUT image falls relative to the '04/'11.

As Harmy said before, the '04/'11 transfer has less image on the right and more on the left than the GOUT; for these frames at least, it also has less on the bottom and more on the top than the GOUT.

Here's the previous images cropped to roughly match the GOUT framing, compared to the actual GOUT screenshots. I vertically squashed the GOUT frames a bit, to undo the slight horizontal compression so that they'd match the '04/'11 proportions:

As you can see, the GOUT doesn't seem to zoom in like -1 thought. Besides the two sets of images not being the same proportions, I think that part of what might have thrown him off was the fact that the wall paneling on the left is messed up by the DVNR. Look at the third and fourth frames - some of the detail from adjacent frames is cloned into the blank areas. That may have confused -1 as to where the GOUT boundaries were in the complete frame.

I think that we'll eventually find that the GOUT framing is actually pretty consistent.