The Original Trilogy restored from 35mm prints (a WIP) — Page 18

Just checked it. I’m seeing movement on the gout and the 35mm.

EDIT: 15496 on reel 3 is gout 75072 if we assume that the first frame of the gout is frame 0. I didn’t see any duplicate frames anywhere in the shot with the hologram commanders.

I think the problem is that whatever you are using to view the file probably parses the file on-the-fly and comes up with different results depending on where you start viewing the file from. Properly indexing the file is the best way to be almost sure you are looking at the right frame without any errors.

althor1138 said:

It occurs to me that I don’t know what an “irreparably damaged” frame is. I did not annotate all of the red splotches or emulsion scratches but these are the ones that stuck out to me. I guess it really boils down to how much motion is in the frame and how much time you want to spend trying to fix it.

Those “red splotches” are the mould. By the sound of it they will either all need to be replaced with another source, or you’d need to stack the print scans and erase the damage. That may be possible, they are digitally pin-registered/aligned to the sprocket holes, but I’d be guessing they won’t align themselves perfectly without further work as they’re in two different languages, hence made from different inter-negs.

MikeV stacked his scans to erase damage without needing to resort to interpolating the damaged areas, I think poita’s got the same strategy in mind. It must take incredible computing power to do that! Then again, I guess 4K is only a rather recent thing anyway - what was the first film scanned at 4K does anyone know?

Here’s an interesting topic on the issue of scanning. The preference with restoration is to scan at 6K or 8K even if outputting to 4K as it resolves the information better. Although poita hasn’t said, his 4K scans may be captured at 6K or 8K and then outputted at 4K, according to the thread that’s actually standard practice. It probably explains why everyone seems to think Wizard of Oz was scanned at 8K when it was done on a Spirit 4K film scanner that has a maximum output of 4K according to these specs!

I just popped in my 2009 Wizard of Oz bluray and had a look - the most disappointing thing is they used the inferior VC-1 codec rather than AVC (Sirius Pixels) and the movie is only 21GB. For comparison, Deep Red 4K is 37GB and AVC (x264 encoder) its film grain looks better than Wizard of Oz. But WOO still looks amazing, and it was scanned on a Spirit 4K (according to the restoration video) which is now an older obsolete/inferior scanner compared to what ESB is being scanned on. The technology going into this project is incredible, on par with top commercial film restorations, and even better than what was done for Oz!

We should also talk about the mess that Lucasfilm made with all of this as well. What we’ve ended up with can really only be described as a mess. They haven’t done a restoration on these films since the 1997 Special Edition when the o-neg was upgraded with newly struck film to replace the aging stock (aside from the unused 4K version done by Reliance Media in 2012, but until it get used it doesn’t count any more than Legacy does). We were then told this made restoring the theatrical versions impossible, which isn’t true because even if the o-neg is now different there would still be internegatives or interpositives that can be worked from, or the camera negatives that are still stored, or of course prints, or the parts of the o-neg that were removed as well.

The 2004 DVD was just a high definition video transfer for home release. The surprising thing is that’s the version that was screened in a 2014 cinema marathon. The 1080p transfer done in 2002 we would not even call a scan by today’s terminology, we’d call it a telecine. Don’t get me wrong, it’s certainly good by telecine standards, and Lowy did a decent job at digital clean-up given the short turnaround time required, but the limitations of the technology are quite obvious. For example, there’s a scene where Luke is in Ben’s hut - 36:11 on the SSE and 36:32 on the blu-ray - watch that scene and look at the wall above and to the right of 3PO. It shakes! Yet there’s no shaking at all on the SSE, it’s an imperfection in the 2002 telecine. Or the film is warped (which modern scanners are designed to cope with, and 2002 technology wasn’t). That unsteady image can be seen all over the 2002 transfer actually, it’s just a particularly good example of it.

One of the major differences with what poita is doing is that he has an IR damage matte - that didn’t exist for the 2002 home-video transfer, it was a 10-bit RGB telecine. In their press releases they did call it a restoration, but restorations should produce a theatrical quality result, and we know Fox/LFL’s policy until very recently was that screenings of these films had to be the 1997 SE prints, not the 2002 home-video transfer as confirmed on Film-Tech and also Home Theatre Forum. So I’m not sure what happened in that instance, but my guess is they got permission from Fox/Disney to run the BDs through DCP-o-matic, or they were directed to do that because it was cheap.

According to that Lowy press release they removed up to a million of pieces of dirt from some single scenes in the movie, such as the desert scene. “We’ve cleaned up more dirt on these three movies than we have on any movie we’ve ever worked on, including Citizen Kane – and that was almost impossible” (Lowy). LFL color timed the movie prior to sending it to Lowy, whereas MikeV said that color timing is the last thing you do in a film restoration.

That whole process would be much simpler today, just look how clean and gorgeous the reels poita had scanned came out after ultrasonic cleaning, I believe telecines picked up more dirt than modern scanners as well. Plus just the fact of having a 6K or 8K sensor - even if downsizing to 2K - means that the image and grain structure is of a much better quality as talked about in that cinematography forum thread where DavidC says 35mm needs 8K even though the detail resolved is around 3K to retain the fine film grain.

If it’s scanned at lower resolutions the grain can block and smudge together and you get that blotchy look that the SSE suffers from. As noted before, everyone seems to think that Wizard of Oz was scanned at 8K when the restoration video makes it clear it’s being scanned on a Spirit 4K scanner, and I think people are confused about the exposure resolution vs the resolution on the hard disk. I’d be interested to know if these prints are being scanned directly at 4K or at a higher exposure, and if poita agrees with the benefits DavidC on cinematography forum argued for.

EJones216 said:

That would be the 1993 restoration of “Snow White and the Seven Dwarfs”, scanned and output at 4K (I believe 10-bit Cineon). No idea if Disney has since done a new digital master, and it wouldn’t surprise me if they have. First new-release 4K DI was 11 years later with “Spider-Man 2”, though VFX were done in 2K.

Interesting, what’s the bet that after their 4K restoration they did what everyone does and did a 2K output to film instead of 4K? Well, if we can find a print from the 70’s/80’s we can do our own 4K scan.

I agree with David if you are talking a really fine grained negative in great shape, that was exposed well. For anything else, 4K full aperture really doesn’t look any different to 8K in the tests we have done.

Scanning a fine grain negative at greater than 4K is sometimes worth it, but for a release print, 4K already exceeds the resolution of the print in nearly all cases.
Also, 4K scanning is not what people generally think of as 4K or UHD i.e. 4096 x 2160 or 3840 x 2160

For a scope print, a 4K full aperture scan is 4096 x 3112, and there is image in over 3000 lines of the 3112 vertical lines captured, unlike a 4K Blu-ray of a cinemascope film, which would have around 1600 lines of vertical resolution.

So there is nearly double the pixels in a 4K full aperture scan of a scope print vs the resolution of a UHD Blu-ray.

The blotchiness on the SSE isn’t because it was scanned in ‘only’ 4K (I thought it was done @2K, but I could be wrong) it is mainly due to a combination of most of it being sourced from a dupe print, not an original print, plus they used a scanner that suffered from digital noise and had a less than ideal dynamic range, which means the image had to be pushed more, the image processing done on it, and the final compression.
This is not to diss the efforts on the SSE, but they were up against it from the beginning, and did a great job with what they had to work with. A lot of what people think of as grain in that release is actually a combination of sensor noise, dupe grain and processing issues.

I disagree with Mike re timing, we always do a primary grade early in the restoration process and then a final grade at the end, so whilst I disagree with a lot of what Lowry does, the placement of the grade I think is about right.

For release prints, even 4K full aperture is likely overkill, but if anyone turns up a negative, I will happily scan that at the full 10K of the sensor 😃

As a few of you know, I have health issues, and I am away from Monday through Friday for treatment, so may or may not be able to answer PMs etc. until the weekend, depending how hard it all hits.

I’ll be back online Monday week at the latest, so don’t take radio silence in-between as me ignoring anyone 😃

poita said:

I agree with David if you are talking a really fine grained negative in great shape, that was exposed well.

Ah yes of course. I was enjoying reading all those threads too much I forgot about that. Does the old nitrate stock Wizard of Oz was struck on have a fine grain too?

The blotchiness on the SSE isn’t because it was scanned in ‘only’ 4K (I thought it was done @2K, but I could be wrong)

Yeah 2K, I didn’t mean to say it was scanned at 4K. Have you seen Deep Red 4K by chance? That restoration looks stunning, even better than Robocop 4K in my opinion! Robocop seems to have more of that blotchy grain in the darker areas.

had a less than ideal dynamic range,

Yeah I’ve noticed that issue. The 2002 Telecine also had many color issues, and has those warped film distortions which the SSE doesn’t. The distortions are relatively minor, but a new transfer on modern equipment would eliminate them entirely.

A lot of what people think of as grain in that release is actually a combination of sensor noise, dupe grain and processing issues.

What’s interesting is that the grain on release prints, at least from what I’ve seen, when projected in a cinema is really faint and when more visible usually indicates an optical effect.

I’m not even dissing the 2002 Telecine, I was reading this thread on Home Theatre Forum and everyone in it seemed to love the new transfer when it was first released on DVD. It just an old out of date transfer now, intended for home video and broadcast, not for DCP/film-out. I don’t understand how they were happy to slap that onto the 2011 Blu-ray while every other watershed film gets a new pristine 2K or 4K restoration, and even less important Giallo films of the same age now look better than Star Wars!

poita said:

As a few of you know, I have health issues, and I am away from Monday through Friday for treatment, so may or may not be able to answer PMs etc. until the weekend, depending how hard it all hits.

Ouch, I hope your treatment is going well and you stay safe with all the travelling that must be involved. 😃

Best wishes Poita!

Take care poita!

Feel better Poita, we’ll all still be here when you get back, take all the time you need dude and happy holidays 😊

Best of luck mate!

RU.08 said:

What’s interesting is that the grain on release prints, at least from what I’ve seen, when projected in a cinema is really faint and when more visible usually indicates an optical effect.

Mostly because projection is really soft, the lenses aren’t particularly sharp, the screens are ‘rough’ and the image is ‘flashed’ multiple times due to the shutter and the film moves slightly with each ‘flash’, our brain does a bit of temporal processing on that - the grain is much more prominent in the screening rooms at work with really sharp optics, a smaller smooth screen and a film projector without much slop in the mechanism.

To view on a TV is a much different thing, so I believe for a home theatre release, you do need to reduce the grain slightly, to get the same effect as viewing film in a commercial cinema.