poita

The deck needs to be cleaned so that we get the best possible transfer, and I had to soure a 7pin dub cable so that we can convert the Y/C 688 to NTSC Y/C levels, as this gives much better quality than coming out via composite.
There is a lot happening, it all costs money and time, the time being multiple people’s time, all of whom have to fit it in around families, life etc. and I am managing it from across oceans to get it done.
So it will take as long as it takes, but assuming the tapes are okay, and the cable works, and the JVC TBC/Transcoder works then it will get done fairly soon. I’ll update the thread with any news, at the moment cables and other items are yet to be tested or are in transit.

Rather than have the discussion all jammed into one thread, I’ve decided to steal Harmy’s idea and start a feedback thread for the THX 35mm restoration and grindhouse version here.

Please post all feedback in this thread, try to keep this thread to actual feedback on the scans.

Quite a few people downloaded the version of the straight scan, I’d like any feedback here as to changes between it and the Laserdisc versions and the 16mm version, as well as restoration and scan issues that might otherwise be missed.

So far only one person as given me any feedback re the scan, so please feel free to post feedback notes here.

Thanks!

Very very much appreciated!

SilverWook said:

Is it the one that someone around here had previously? The old THX thread was shared with AG discussion early on.

It was bought ages ago, I need to replace the CCD in my scanner, so it has been sitting there awaiting funds unfortunately.

I am also in serious need of more 4TB HDDs and another 1TB SSD too if anyone has any spare. At the moment everything keeps grinding to a halt when I have to unload 4TB of data and wait 8 hours at a time before continuing work.

I have an original 16mm American Grafitti here waiting to be scanned actually.

For anyone wondering what U-Matic is, let these lovely ladies from Sony fill you in.

https://www.youtube.com/watch?v=yGNXwZooiPc

Yup, I have some high band tapes from 1982, and the tapes may be high band, as they played back in a BVU but not in a standard U-matic deck, so we will know soon.

Yeah, I know all about cleaning U-matic decks, I have four here, and have passed on instructions to be used if required.

Thanks Puggo, the VO-9850 isn’t high band is it?

If the tapes need baking, are you able to offer that service?

How to donate is right there in my signature 😃

The tapes should be arriving early next week, they ended up costing me $95 with shipping, the U-Matic deck was pretty costly though.

Hopefully the tapes won’t need baking, but we will know soon!

LDs and home re1eases have radically different exposure to the theatrical versions, you can see that in the explosions as well.

Not many are pure white, they vary quite a bit.

It is cinemascope, so it is twice the height, you can reduce by 50% to get it to the correct size. I didn’t want to do that and potentially lose stars.

Right click to view full size.

You are in luck, I hadn’t unloaded Reel1 yet to get back to THX.

Here is a quick extraction, the planet is probably a bit messed up by the stacking algorithm, but the stars are what you after, so I hope this is of use.
I won’t get back to this for months, so if it isn’t, then it will be a while till I can get to it again

Full size 16bpp TIFF 2422×4068 image is here:
http://tinyurl.com/zqy2hdm

I concentrated on the bottom half, so the top may have less stars than the previous image, but you could always comp the two together.

I’ll see what I can do. My HDD shortage means to work on this I have to unload the THX files from the RAID to my one free drive, then load the SW Reel 1 sources (four of them) onto the RAID from their drives, it takes around 4-8 hours every time I interrupt one project for another.
The alternative is to work without a backup, but I lost so much last time that happened, that I am not willing to do that again, but it currently means backing up the files every time I need to work on them, and erasing the backup of the files I was working on previously. It is rather painful.

Just as an aside, I have viewed many Ep IV crawls, and the prints I have seen a have almost no stars in the Starfield, I think it was a compositing issue with the redo of the crawl. Audiences would have seen a fairly Star-free Star Wars at the time.

To recover the stars from these prints I am actually using Astronomy software (I was also an astronomer during my production days, both were work that rewards an inability to sleep) and it does a great job of digging the stars out of the noise from multiple prints and frames. It is time consuming, but effective

It is helpful to calculate the shadow fade, as well as absolute black to use the space between frames, but the shadows midtones and highlights all fade at a different rate, so it is useful, but mostly for the blacks.

No one is using Rec 2020 yet, you can’t even monitor it fully, even with $40,000 monitor solutions - and for Star Wars there certainly isn’t any need or it.
Basically if you master in ACES then you are future proofed for any conceivable delivery colour space.

If you have a passion for it, get yourself to some kind of film school, you need to be working with people on projects, there is really no other way ‘in’ that will get you the experience required. Knowing the software packages, though eventually necessary, is really the least important part.

I’d say 70% or more of what I use is custom code, or scripts or custom plugins, so being able to write in C++ certainly doesn’t hurt, and the skills you develop as a coder are useful accross the board. The people skills are crucial though, as are the contacts and the production experience.

Just in case it was a serious question, the first step is to go to university and study film production, you will need a solid backing in the technical aspects of film making, and reading about it on the net is very different to being involved in a production environment. It also gives you the discipline and teamwork skills, as well as a deeper understanding of the medium itself. Just as importantly, you make a ton of contacts that will be crucial to finding work.
Learning to work with directors, DOPs, colourists and other people is just as important as having the technical skills.
Study colour theory, it is a huge topic and requires a lot of understanding to be able to restore film at a high level. Learn to grade, the old way with just primaries, and learn to do it fast, relying purely on lift/gamma/gain to get the results you want. We used to have to do it realtime whilst a transfer was happening, so by fast, I mean, do it until it is second nature.

Once you have an understanding of grading both on an intellectual and muscle memory/physical level, have learned how to work in a production team, how to talk to sometimes problematic directors and other artists, have some real production experience you can put in your portfolio, even if only on student films, then you are employable and can work your way to restoration. If it sounds like a long time, a lot of work, too hard, then the industry isn’t for you. That is the fastest/shortcut way in really, the long way will be the subject of another post.
Your location is a huge advantage, you already have it easier than most anyone else in the rest of the world just because of where you live and the access to people, schools, festivals and tools. It is still hard, and you will be broke and living on noodles and working ridiculous hours, but it is great fun. So if you want to get into it, nothing is stopping anyone other than the motivation to do so.

It used to be so much harder, almost impossible to get into, but now it is there for anyone willing to do the work and develop a decent eye and to ask a million questions and accept that all of us pretty much know nothing and are constantly learning. Never pretend to know something that you don’t, you won’t last a week in production. The smartest people are the ones that ask the most questions and never let anyone down through making assumptions or by covering up for their ignorance.

I’d take the $2500 fee quite happily at the moment 😄

They unfortunately again are not much use. THX was specifically timed after it was shot, to convey emotions suitable to the scenes. For instance, there is an intentional green tint in some scenes, meant to emphasize the un-natural conditions and make the skintones look a little off. The sex scene was warmed so that the skintones would ‘pop’ against the almost monochrome backgrounds etc.

The print we have is an IB Tech, it is battered but has zero signs of fading. It is how the print would have looked when projected in 1971. If you watch the version I worked on, the colour on it is accurate to the print.

You or anyone else can make a version with a different colour grade, one that they like or feel improves on the theatrical release, but the colour in the sample is 95% matched to the print.

We aren’t used to the variation within a print these days, but when everything was done with chemicals and lighting, that sort of inconsistency was normal. Telecine transfers for home video tended to be warmed up and exposed to stay within the NTSC or PAL legal levels and often looked totally different to the Theatrical release, so this will look very different to the LD and VHS etc. versions as far as the grade goes.
I’ll be matching to the print, but the pre-grade files will be stored in ACES at 16bpp for anyone that wants to make their own grade to suit their own tastes as an exercise.

You might be waiting a while, I am awaiting the arrival of the tapes, and have purchased a better U-Matic deck for the transfer, I have had the deck sent to a friend in the US to do the transfer to save having to get the tapes and gear to Oz.
Donations as always are more than welcome, this was an unexpected expense, but seemed too important to pass up, plus having a good quality U-Matic deck in our hands makes future projects easier if they turn up.

After the tapes have arrived, and have been checked to see if they need baking, we will get some samples up, but don’t hold your breath, we are pretty snowed and broke at the moment.

well, that is a 1080P version if you want to play with it, right click on it and save it out at full resolution.

I am honestly snowed at the moment, and won’t get back to the crawl until I get back to working on Star Wars itself. I am currently doing THX, organising The Shining, and am perilously short of disc space, so it might be a while before I get back to the Ep IV crawl.

I have been down the super-resolution path, with separating elements, and the results are uninspiring from the sources available.

We have scans at higher than 4K of the original prints, they contain much more genuine resolution than any of the home versions, so if you were going to work with anything, it would make a lot of sense to work with those.

Resolution isn’t really the issue, it is dealing with noise, damage, grain, recovering colour, etc. We actually have more resolution than is easily handled.

Feel free to pursue your theories, you may work out a really great workflow or a solution to an existing problem, but also be ready for robust discussion and feedback. I do film restoration for a living, have a horrible amount of study and materials on everything from esoteric colour theory, to film emulsion composition, to brain bending (for me) algorithms and math based theories on everything from noise analysis, to colour matrix transforms, and (for example) over 2000 pages of research on the way silver halide particles interact with dye bases, their clumping behaviours, diffraction processes and limitations and more stuff that I really never wanted to know about. I know about 1/4 of what I wish I knew about this stuff!

This is to say, it is a deep field, way too deep for any one person to know everything, but there is a reasonable chance others have already trod the paths you may choose to explore, so be aware that people may ask pointed questions or suggest that things have already been tried or may be a dead end. You may re-explore those paths and find a branch others missed, so go for it!

The main thing on these forums is results - there are a lot of good, bad and ugly theories out there - so your idea of trying a few seconds first and seeing how it goes is a good plan. Post your results and I’m sure you will get a lot of interest.

I only just saw this now. If anyone wants an isolated starfield from an actual 1981 print, I can probably extract a clean one.
I haven’t got time to spend on it now, so here is a quick extraction, but it may not be complete/may have some film damage, I only spent 15 mins on it.

Was planning on having another sample up today, but spent the day attempting (and sadly failing) to repair the washing machine instead. Might have something up later in the week.

The reason we tend to ‘throw’ out those sources, is that lighting, lenses and stock are everything, and the production shots, box shots, other photos etc. are under different lighting, different lenses and processed according to the whims or needs of the photographer or the publication of the day, then years later scanned by ‘who knows’ with settings chosen for reasons we don’t know, probably on uncalibrated equipment, then converted for the web and displayed differently yet again.

They are interesting, and perhaps a little useful in some situations to know for example that the red spot on the trooper’s belt is not a glitch for instance, but they are of little use when we have the actual film elements, most not-faded, except for Empire, and we do have unfaded transfers of ESB to Super8, and an unfaded LPP that we can view (but unfortunatley not scan). Also the film would then have been timed, so changed again from what was shot.

Even the props are of limited uses, and I have been lucky enough to study most of those that still survive. For example it doesn’t matter if the prop was ‘pure white’ if it was shot at the end of the day, through a diffusion filter onto a stock that renders colours in a certain way, under those situations the white prop should have golden/orange hues. It doesn’t matter that the prop was white if shot in the blue-orange lighting of the ‘Han lowered into the carbonite chamber’ scene, it will be affected by the light and grade of the scene. The reference is the prints themselves, the fade is a known quantity where there is fade, and printing errors are also a relatively known quantity, and for all three films, we have a pretty reliable reference in the prints themselves. They are much more accurate to the way the films were presented in 77, 80 and 83 than any photos will be. If as an exercise, one wants to know what colour the props and sets were when viewed under 5600K lighting, or another arbitrary light source, then that is an interesting thing in itself, and useful algorithmically to help with balancing, but doesn’t necessarily have any relevance to how those props and scenes look in the finished movie. It is useful if you want to recreate costumes or props accurately, but no much use when restoring the films.

That is not to say the algorithm is not useful, it absolutely is, but production shots, box shots etc. are of little if any use due to the reasons above.

(I apologise for the typos, my keyboard is dying)