poita

Final scans have arrived and I am working on them now. Unfortunately for the past 48 hours I and most of the rest of the family have been laid up with a particularly violent vomit bug. The kind where you think you are going to die, and then start wishing you would.

The print is very beatup in places, I have been working on some algorithms to help automate the process. I’ll pop out a low rez grindhouse, and a first cleaning pass version soon, so you can all follow along with the process.
Basically I analyse the film, work out some semi-automated routines, run those passes and then look for what didn’t work, and then start finessing it from there.
I’d be appreciative if people could join in and help ID the sections where my routines fail so I can modify them and then finally go through and do a manual pass to clean up any remaining problem areas.
It is particularly tricky due to the CRT flickering screens and fine details in some scenes, but overall it is great to watch a ‘film version’ even if it is in a language I don’t speak. It is a mostly visual film anyway 😃

You can tell if an IB print has projection fade by comparing the masked area with the main picture area. Even the worst case fade is really minimal, not enough to visibly affect the sky for instance.

The ‘noise’ is actually in the print, the blue layer is terribly grainy, and that is true for film generally, the blue layer has the most noise.
In this case, I also haven’t set the black levels, so they may be too high in the blue layer, causing you issues, the black levels should be set so that the base noise level is obscured.

Just on film grain, many people incorrectly think it is the grains are the ‘pixels’ that form the colour information. In reality the ‘fundamental particles’ that hold the colour information are the silver particles and colour dye clouds. The grain is an order of magnitude larger than these particles and is effectively laid over the image, obscuring it with noise, it is not the source of the image itself.

Just out of interest, here are the images from my scan of the LPP, with the -1 scans on top, mine on bottom. No colour correction done.

-1 Scan:

Poita Scan:

-1 Scan:

Poita Scan:

-1 Scan

Poita Scan:

Poita Scan with basic white balance done:

You can right click on any of the images and ‘view image’ to look at them in 1080P, zoom in etc.

It will be interesting to see how the algorithm works with these.

Finnius said:

…but it’s the little flaws that fans have also endeared to their hearts. Now, if you grew up with the SE versions…THAT is your memory of Star Wars…but for old bastards like myself, the matte lines, missing frames in death star explosion, x-wing hiccups are part of the experience. It’s what I know, its what I love.

Another perfect example would be Blade Runner. I LOVE Blade Runner. I love the stunt double man in Pris’ death scene…the stunt double man during Zhora’s plunge thru the glass, etc etc…these things don’t detract from the power, the presentation, the greatness of that film. Oh yes, I own the ‘briefcase’ collectors edition…HAD to get it. And the remastering is wonderful…but I still pull out that original theatrical just because that was MY Blade Runner…the one I grew up with. It’s the same with Star Wars. I need MY Star Wars back.

The odd thing is though, that many of the flaws the fans ave burned into their consciousness are from watching the home versions on LD, Laserdisc, TV etc. and in some cases are quite different to the projected versions. I’ve seen it projected more often than from home releases, so ‘my’ Star Wars is different to many. Everyone has their own ‘Star Wars’ 😃

There are two things at play here.
One is channel misalignment on the IB Tech prints, this is common on IB prints, and you can check against the non-IB prints to see what is channel alignment and what is not.

Two is chromatic abberation, this is visible on all prints in certain areas, epecially starfields, particuarly at the edges of frame, i.e. there was more CA as you got away from the centre of the lens.

Getting back to the negative is an interesting goal for preservation, for restoration, not necessarily. There is plenty of visible stuff on the neg that never would have made it to a release print, and some detail on the prints that certainly would have been invisible when projected.

Restoration is tricky, remastering is easier as far as decision making goes.

There definitely would have been presentations in 1977 where the garbage mattes would have been invisible or barely visible - there would have been others where they stuck out like a sore thumb.

Mike’s project is clearly a remastering, and would result in a viewing experience of Star Wars which is probably how most people would ‘remember’ seeing it, and would be easier to immerse yourself in the film, without the glitchier bits pulling you out of it. That is something I would love to see.

I’d also love to see a full ‘restoration’. However, a restoration, well, that is another thing altogether, and you will probably never get two people anywhere to fully agree, not just about what is kept and what is not, but even to what degree… Just how visible do you make those garbage mattes…

titanic said:

poita said:

titanic said:

Sorry if I sound anxious (can’t help it), but is there any update on this?

(and Alice in Wonderland that was mentioned)

The scans are all done of Fantasia and Alice, just looking for anyone that wants to handle the restoration, I really do not have enough time even if I cloned myself. The files are currently in the US.

Thanks for your reply.

Are the raw scans that bad that need restoration?

I mean, the raw scan of Jurassic park was by all means watchable.

And by restoration you mean artifact/dust removal and color/gamma/brightness correction?

A print well over 60 years old is not in as good a shape as a 90s polyester print unfortunately, and that Jurassic Park print was in incredibly good shape, it was by far the cleanest release print I had ever seen. Fantasia is, as you would expect, in far, far worse shape.

There are some good priced ones over here in Oz at the moment, An SP of The Lion King and a Cinderella tech have turned up recently, they were under a grand each.

Mavimao said:

poita said:

I have scanned that print, it was unfortunately infested with mold.

Gross. Is it unrecoverable?

No, it ate its way inbetween the film layers. Not at all unusual for IB prints in England, sadly.

junh1024 said:

Imo, with 35mm you will NOT get much detail beyond 1080p, and with 70mm, you will NOT get much detail beyond 4K. This is AT BEST.

Not true, but the real answer is complex.

On some of the Star Wars 35mm prints I have scanned, there is plenty of detail at 4K that just isn’t there, and in some cases, cannot be there at 2K/1080P.

If you resize the 4K scan to 1080P that detail is lost, the classic example is the door seal on the Tantive at the opening of Star Wars (Ep IV). On the 4K scans you can see the ridged detail, on the Blu-ray and any 1080P source, it is a non-descipt grey.

However, the Blu-ray, even in that same scene shows some detail not in the 35mm prints and yet is missing detail from the 35mm prints at the same time

So what is going on?

It is down to transfer. When you scan from the neg, you pick up all sorts of detail that will never make it to a print, and some of this survives the down-rez to 1080P, so the Blu-ray is more detailed in some areas than the prints. Yet there is some detail that is simply finer than 1080 lines, so can never be seen on any 1080P source, uncompressed or otherwise, yet will show up nicely on a 35mm print.

So the answer is… it depends.

There is definitely detail on the Tech IBs of Star Wars, that are well above 2K, but if your final delivery is 1080P, then it probably doesn’t matter. If you want to make a UHD/4K version for viewing on your snazzy UHD/4K TV however, then that detail can be retained if you scan at 4K or higher.

I tend to scan at a vertical resolution of 3400 pixels or so, and that seems to retain all the detail from the 35mm prints, dropping to 2160 pixels vertically definitely loses some visual detail, so I won’t scan archivally below 3000 pixels, but if your delivery is 1080P, then you could probably scan at UHD resolution, and end up not seeing any difference.

That is all talking about 35mm prints. A 35mm neg however easily can have detail that exceeds 4K, but again, unless your final delivery is better than UHD resolution, then you may not need to scan any higher than 4K full aperture.

FrankT said:

It’s… not mine. I have to borrow someone else’s machine to use it. I’d… rather not talk about it, actually.

I’d strongly recommend not talking about that package unless you own a license 😃

I am happy to help anyone that will genuinely undertake a restoration, but they have to supply the drives and freight etc.
What I am not doing is sending out files to people that just want a copy or have a look, genuine restoration or research projects only, so anyone that asks, please forward details of your restoration project via PM - I am not ‘distributing’ files as such.

I have scanned that print, it was unfortunately infested with mold.

FrankT said:

4 TB? That’s… I’m not exactly made of money anymore…

PFClean works fine with ProRes files, I’m surprised you would be using it, it isn’t cheap.
Make sure your license is up to date, PixelFarm are extremely litigious and go out of their way to protect their software.

If you can’t afford a 4TB HDD and postage, then I would suggest that you won’t be able to undertake the restoration.

You need at least twice the space of the scan to be able to do a restoration, just for working files, and you need to be able to output the final result.
That is part of the reason I am so broke, the need for working space, and preferably fast working space is a huge requirement, but one you really can’t do without.

Send me a PM, it would be in your interest I think to get a 4TB HDD, you will need some working space!

I’m old.

So it will just cost you your youth.

I can think of 8 specific viewings over the years, but it would likely be 12 or so times.

AllAboutThatSpace said:

Don’t have much to contribute except support.
I’ll try to donate in a couple of months when I have some more moola.
I saw this in 70mm at the Prince Charles cinema in London. The remarkable thing about film that I hadn’t noticed before is that it highlights detail you want and disguises detail you don’t. On the blu-ray, in the Dawn of Man prologue, you can clearly see paint roller strokes or cleaning marks on the back projection screen behind the savanna sets. It’s distracting and annoying and you think why would such a perfectionist and pedant as Stanley Kubrick put up with that when it’s so obvious? I was looking out for those marks on the 70mm print (because I’m a nerd) and they just plain don’t show up. The film print just ignores them. The print’s sharp as a knife alright, but only in areas of intense bunched up detail or bright highlights. Same kinda thing with Star Wars, the green screen and effects mattes just blend better in the release prints when compared to the blu rays.
In many ways scanning in high def from the camera negative is not a restoration of an old film in any way. It’s a preservation for sure, but it’s getting back to a raw material that wouldn’t ever have been seen. Kubrick didn’t pick up on those marks, because even on 70mm they weren’t a problem. That’s why these scans of release prints look so good, the generational grain binds the images together to give you an intense feeling of ‘story’ rather than ‘photography’, that’s why we like em so much 😉

I first noticed the patchwork background of the front projection screen when viewing the film in 70mm, I haven’t seen the blu-ray, but I notice it every time in the cinema.

About 2TB for ProRes

The scan was done, but no restoration as such has taken place, if anyone wants to do the restoration, just drop me for a PM.

This is true for a negative, but for prints, especially release prints, there is rarely much information above 4K, and generally many films struggle to represent 2K resolution.

The blu-ray often has more detail than a release print, but not always and not consistently.

It is a bit hard to explain, but being able to scan from the negative lets you capture detail that never made it to the print, but sometimes there is fine detail in the print that is lost when you downscale to 1080P.

Grain is definitely a ‘fault’, it is the very definition of noise that obstructs the signal, but in many cases it was ‘used’ to help hide matte lines, blend separate elements or breathe a bit of life into a static shot. In other words, it was a known limitation, and was factored in to the final look of the film, and that makes it worth preserving in many cases.

Yes, matching the grain and sharpness and black and white points basically, which would mean I would treat those scenes separately, as I wouldn’t scrub the entire film quite as heavily.

Yeah, once the other drives arrive, I will put together a semi-cleaned up version quickly that should be quite watchable, I ned it anyway to plan the proper cleanup workflow.

If someone could put together a scene list from the bluray, of what has been changed, then I can do the full restoration of those scenes first, before tackling the restoration of the entire film. I’d also restore those to match the blu-ray, which would be quite a different process to what I have planned for the film itself.

Actually Harmy, I’ve been meaning to talk to you re despecialising THX, it would make sense for me to do the needed scenes first, if you have a list.