logo Sign In

1997 Star Wars Special Edition 35mm Project (a WIP) — Page 20

Author
Time

RU.08 said:
Right, the dirt is something I noted to poita, in fact the very first thing I noticed and I gave him quite a few examples. It’s not on the print it’s in the print. Not just black dirt, but white dirt also (i.e. dirt on the positives). And yes it would still be in both because it’s a composite shot and the dirt is on the film used to make the composite.

Black dirt is dirt on the positive, white dirt is on the negative. So this dirt can’t have been on the original photography element else it would be white. It’s also not on the grindhouse version of 4K77.

Author
Time

I’m more inclined to believe that the saber differences are simply due to the different path the media takes for telecine, but that’s just my opinion.

However, I do have evidence to support the claim that “All of the wipes and fades were redone optically for the Special Edition,” rather than digitaly:

“He found the original negative at Fox and the separation masters in a salt mine somewhere in the Midwest, but he couldn’t find the original negatives for the wipes. They looked all over Los Angeles—no luck. Nobody asked me. They were in my ILM editorial warehouse… We had the negative and everything else he was looking for.”

Ironically, just as ILM was retiring optical printers and moving into the digital realm, the technology was resurrected again. Pacific Titles had eleven state of the art, modern optical printers with new lenses, which, when combined with Kodak’s finest film stock, gave “a boost in resolution and color saturation,” according to company vice-president Phillip Feiner. They re-composited all wipes and transitions (the “bread and butter” opticals, as Feiner calls them).

TheStarWarsTrilogy.com.
The007Dossier.com.
Donations always welcome: Paypal | Bitcoin: bc1qzr9ejyfpzm9ea2dglfegxzt59tys3uwmj26ytj

Author
Time

But the question is were any of the effect shots optically recomposited? Particularly this shot in ben’s hovel which would need to have a new rotoscoped saber element created.

Author
Time

Chewielewis said:

But the question is were any of the effect shots optically recomposited? Particularly this shot in ben’s hovel which would need to have a new rotoscoped saber element created.

I don’t think so. I don’t think all of the effects shots were redone, but from what I’ve read, those that were were done digitally. The negative (or in some cases the IP) were scanned in at 2k, digitally manipulated and then output back to film as a new piece of negative.

“Soon all was ready to cut into the original Star Wars negative. Before doing that, however, Tom wanted to check something with George. He explained to George that everything was ready to cut in, but he wanted to know whether they should order a duplicate negative of the original version. It could easily be done, he said, and then there would always be an original version of this historic movie. George said, “This is the original version,” meaning that what everyone else had described as a restoration, he considered the final version of the movie.”

TheStarWarsTrilogy.com.
The007Dossier.com.
Donations always welcome: Paypal | Bitcoin: bc1qzr9ejyfpzm9ea2dglfegxzt59tys3uwmj26ytj

Author
Time
 (Edited)

Chewielewis said:

RU.08 said:
Right, the dirt is something I noted to poita, in fact the very first thing I noticed and I gave him quite a few examples. It’s not on the print it’s in the print. Not just black dirt, but white dirt also (i.e. dirt on the positives). And yes it would still be in both because it’s a composite shot and the dirt is on the film used to make the composite.

Black dirt is dirt on the positive, white dirt is on the negative. So this dirt can’t have been on the original photography element else it would be white. It’s also not on the grindhouse version of 4K77.

That dirt splodge is not ‘on’ the print. I can tell that by looking at the IR dirtmap.

The wipes were done optically, not digitally and were done by Pacific Title.

Yes, not all shots were re-done digitally, but those that were… it was in 2K and put back out to film and cut into the negative. The removed pieces of the negative were kept, and Disney now has them.

Having had to do optical composites, I find it really unlikely that this shot was recomped optically just for the telecine. Firstly, there would be no need, the sabre comps really transfer easilyto video, and it is nice and wide, there would be no reason to think it wouldn’t transfer that well to video.
Secondly, I can see any way that the alignment would end up so precisely perfect to the release print if it is an optical re-comp. There would be more discrepancy, it is really, really, hard to get that level of precision.
It looks like an adjustment made on the telecine, threepio has also been affected by it. Nothing else in the scene is bright enough to be affected, so it makes sense to me.

I’m not saying it is impossible, but there is really no reason to redo this scene for telecine, no reason to believe that it would have been problematic as is, and it would have been a pain in the arse to seek out the elements and re-comp unless there was a really compelling reason to do so.

So, it could have been done, but I’m not convinced so far. Look at the ‘split sabre’ syndrome on the Definitive Edition laserdisc and you can see how much the transfer can mess with the look of the sabres.

I will be able to answer definitively whether it is a digital comp or not when I receive the full files, there is a difference in the framing surround in the shots that have digital content, but unfortunately it is cropped on the ProRes Proxy, so I can’t quite see it, so frustrating, but we will know for sure.

That is something else we will be able to log, which shots have digital enhancement, and which don’t, for the entire movie.

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!

Author
Time

Two frames missing at the head and 5 at the tail, in case you were wondering.

Is there a GOUT style sync standard for the SE?

Author
Time

Director 10K, triple flash.

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!

Author
Time

I was looking at the home video releases of the 1997 SE recently, and I was surprised to see how inconsistent ANH is in terms of colors, compared to the other two films, with the color balalance being all over the place, and larhe parts of the film having a blue cast/pink cast. TESB and ROTJ look really great, and the colors of TESB are really close to Puggo Strikes Back. Do you know if the print versions of TESB and ROTJ are as inconsistent as ANH in terms of color, poita?

Author
Time

Both the original 1977 film and the 1997 Star Wars SE are all over the place as far as the colour grade goes.

The original ESB is better than Star Wars, but the grade still varies a lot.
So many comp shots and different base films and a lack of digital colour grading at the time all lead to an inconsistent grade. I haven’t watched Jedi enough to be able to confidently say how it looked in 1983 as far as colour goes, other than that the Jabba palace scenes were extremely dark.

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!

Author
Time

The Jabba’s palace scenes gave Puggo a few headaches on the 16mm transfer.

Forum Moderator

Where were you in '77?

Author
Time
 (Edited)

I found an interesting scientific paper on the resolution of 35mm film going from the negative to watching a release print on a movie screen. If I understand correctly, the average discernible resolution by experts of a 35mm film viewed in the theatre is about 720p, but I could be mistaken, so I think poita can probably shed more light on the process used by these people:

https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7262540

Author
Time

I think I’ve talked about this paper before.

The main issues with their testing procedure are:

The low contrast test patterns
The projection lenses at the cinemas chosen
The screen textures at the cinemas chosen
The sine wave test patterns used.
The gates and gate pressure of the projectors used.

While the test was worthwhile and produced some useful data, the resolution findings are pretty arbitrary, apart from comparing resolution loss at a relative level between the negative, inter positive, answer print and release prints.
The ‘absolute’ resolution figures are not really useful, all they really tell us is the quality of the projector and screen being used at those particular cinemas.

Also, it is at 1.85:1 so the numbers are off compared to a 2.35:1 or other ratio image.

Had they tested the projection lenses, and the gate pressures and done some analysis of the screens in use, the numbers would have been far more useful.

Cinema presentation varies radically, and gate wear and pressure has a huge effect, as does the light scattering properties of the screen in use, and of course the projection lens, and the lens mount.

As for discerning the resolution of a release print, you can find that out directly and much more accurately by shooting a wider variety of test patterns and then examining the release print with a microscope or a scan of the print.

The resolution discernible in the cinema itself when projected, will be lower than that number, but will vary massively from cinema to cinema, even with the same projector and lens, the wear and aetup and screen properties could halve the discernible resolution compared to different setup.

The resolution present on a release print can easily exceed 1080P. I can show that directly on the scan we just did of the SE, and it isn’t a very sharp print. When projected, well, that number could literally drop to anything, depending on the cinema.

However, even in a terrible cinema, it will look better than a 720P digital projection, the fixed 720P grid will look clearly pixelated at normal projection sizes, whereas even with a poor setup, the release print will look a lot smoother and more detailed, and the effective resolution will be higher due to image information not falling between pixels or being averaged across them.

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!

Author
Time
 (Edited)

poita said:

I think I’ve talked about this paper before.

The main issues with their testing procedure are:

The low contrast test patterns
The projection lenses at the cinemas chosen
The screen textures at the cinemas chosen
The sine wave test patterns used.
The gates and gate pressure of the projectors used.

While the test was worthwhile and produced some useful data, the resolution findings are pretty arbitrary, apart from comparing resolution loss at a relative level between the negative, inter positive, answer print and release prints.
The ‘absolute’ resolution figures are not really useful, all they really tell us is the quality of the projector and screen being used at those particular cinemas.

Also, it is at 1.85:1 so the numbers are off compared to a 2.35:1 or other ratio image.

Had they tested the projection lenses, and the gate pressures and done some analysis of the screens in use, the numbers would have been far more useful.

Cinema presentation varies radically, and gate wear and pressure has a huge effect, as does the light scattering properties of the screen in use, and of course the projection lens, and the lens mount.

As for discerning the resolution of a release print, you can find that out directly and much more accurately by shooting a wider variety of test patterns and then examining the release print with a microscope or a scan of the print.

The resolution discernible in the cinema itself when projected, will be lower than that number, but will vary massively from cinema to cinema, even with the same projector and lens, the wear and aetup and screen properties could halve the discernible resolution compared to different setup.

The resolution present on a release print can easily exceed 1080P. I can show that directly on the scan we just did of the SE, and it isn’t a very sharp print. When projected, well, that number could literally drop to anything, depending on the cinema.

However, even in a terrible cinema, it will look better than a 720P digital projection, the fixed 720P grid will look clearly pixelated at normal projection sizes, whereas even with a poor setup, the release print will look a lot smoother and more detailed, and the effective resolution will be higher due to image information not falling between pixels or being averaged across them.

Great, thanks! I was hoping for an answer like this. 😃

Author
Time
 (Edited)

I actually have another question. Does the process of going from the negative to a release print distort the image in any way? In other words is it technically possible to retrieve a more or less exact copy of the original interpositive by averaging and then deblurring an infinite number of release prints? In general I’m wondering what the limitations are of doing a restoration from a set of release prints, rather than using an interpositive, or the original negative.

Author
Time

DrDre said:

I actually have another question. Does the process of going from the negative to a release print distort the image in any way? In other words is it technically possible to retrieve a more or less exact copy of the original interpositive by averaging and then deblurring an infinite number of release prints? In general I’m wondering what the limitations are of doing a restoration from a set of release prints, rather than using an interpositive, or the original negative.

In other words, is Mike Verta actually accomplishing what he says he is? I’ve also been curious about this. I know that the multiple duplications to get to the release print leads to the addition of generational grain. But does the image actually get softer outside of that fact?

Author
Time

DrDre said:

I actually have another question. Does the process of going from the negative to a release print distort the image in any way? In other words is it technically possible to retrieve a more or less exact copy of the original interpositive by averaging and then deblurring an infinite number of release prints? In general I’m wondering what the limitations are of doing a restoration from a set of release prints, rather than using an interpositive, or the original negative.

Interpositives and prints are made with contact printing. That is to say, they sandwich the two pieces of film together and shine a light on it. Because there’s no lens, there’s no risk of light refraction.

But I could be wrong…

What’s the internal temperature of a TaunTaun? Luke warm.

Author
Time
 (Edited)

Yes you can get distortions, especially at splices where the film may not sit perfectly flat. No mechanical process is perfect, so you do get some distortion sometimes that is highly visible, with warping, out of focus areas etc. and very slight distortion that would only be noticeable if you compared them directly, but you would’t notice otherwise. Contact prints can also get newtons-rings type problems, which again, might not be highly visible, but can distort the image slightly.

Averaging a bunch of release prints gets you a little closer to the original, you negate some of the effect of the individual print grain, and if you use the correct stacking methods, might pick up some extra detail/resolution that is in one print versus another, but it is tricky, you might just end up substituting noise from one print to another in some areas.

You are never going to even get close to the negative however. The release print is timed differently, often throwing away 4 stops or more of exposure latitude, often intentionally.
And there is loss going from an IP to a release print that is lost forever it won’t be there in any of the prints, (e.g. the dynamic range etc) so it can’t be recovered, but yes if you had an infinite number of release prints, you could get close to the IP, assuming you could scan at a crazy high resolution so that the warping of every frame that you would need to do to assure perfect alignment between each print before stacking didn’t just smear out the detail anyway.

So what you get is something that gets closer to the timed interpositive, rather than getting something that is close to the original negative.

In practice, unless you are averaging a lot of prints (like a dozen) the effect isn’t always better than using smart algorithms to reduce the noise in a single print. The prints need to be perfectly aligned for it to work properly, otherwise slight misalignment is reducing detail, so the net affect can be worse. It is however helpful for recovering data where you have a lot of damage on one frame, and no damage on the same frame on other prints.

Back to resolution, it is kind of difficult to put a set number on the resolution of film, measuring a pixel grid is different to the randomness of grain and dye blobs, and the resolution will vary with different exposures, colours and contrast. Even the grain representation looks radically different at different resolutions, we just finished testing grain reproduction at different resolution levels. I’ll post some of our results later if I can get clearance to do so.

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!

Author
Time

Thanks, poita! This is all fascinating stuff!

Author
Time

Poita, have you ever considered giving some sort of online workshop on film restoration?

Author
Time

I have done workshops in person, but never online.

What would you like covered?

Donations welcome: paypal.me/poit
bitcoin:13QDjXjt7w7BFiQc4Q7wpRGPtYKYchnm8x
Help get The Original Trilogy preserved!