Tweet Criterion: How We Restored The Man Who Knew Too Much

[silentfilm]

http://www.criterion.com/current/posts/2629-how-we-restored-the-man-who-knew-too-much

Very interesting and well done. One question, though -- they said that the wet gate printer used water. I thought that these wet gate printers actually used acetone or some similar chemical, which has the same refractive index as nitrate film, and thus helps to cover scratches and other problems. I also seem to recall that the problem with the wet gate process was that the chemical (similar to what was used in dry cleaning establishments) was banned by the DEP as carcinogenic, and could no longer be used.

So have they now adapted or improved the process to use plain water? Just curious. SETH

Very interesting and well done. One question, though -- they said that the wet gate printer used water.

I thought that these wet gate printers actually used acetone or some similar chemical, which has the same refractive index as nitrate film, and thus helps to cover scratches and other problems during the printing process. I also seem to recall that the problem with wet gate printers was that the chemical (similar to what was used in dry cleaning establishments) was banned by the DEP as carcinogenic, and could no longer be used.

So have they now adapted or improved the process to use plain water? Just curious. SETH

Thanks for sharing! I found it very interesting.

I do not believe a modern film duplicate printer on modern duplicate film stock, would ever render a so visible resolution loss as pointed in this Criterion video in 3:40.
Besides, there is also a density loss in thes comparison of the video above, that is not true in modern film duplication.
Modern contact wetgate printer produces very sharp images, rivaling the original negative, and no white halo around objects.


Even a fine grain from 1950's (if very well made) would not render a resolution loss that we would noticed in a 480p or even 720p video !!!!!!!!

Count generations it's not a reliable way to determine a film resolution, or resolution loss compared to camera negative. Film duplicating . The film emulsions used for duplicating make the difference, as modern fine grain film emulsions are great, and also the printer used a bit of and lab work.
Many early films have bad dupes cause they did not have or did not use good film emulsion. Fists fine grain stocks only appeared in the late 30's.

Even so... Take a look at All QUiet in The Western Front : It looks very good, and it's a lavander, but not a fine grain. It's a ortochromatic emulsion used for film duplicating in 1931 or so, but the silver grains was ver similar to a common print of that time. The lavander of that era was more to help keep the contrast low.



This Critirion video It is smelling like a agreement to promote the new film scanner !!!!!!!!



Anyway, when industry will create a good film scanner, able to handle skrunked old nitrate films and also have wet gate option ???

I do not believe a modern film duplicate printer on modern duplicate film stock, would ever render a so visible resolution loss as pointed in this Criterion video in 3:40.
Besides, there is also a density loss in thes comparison of the video above, that is not true in modern film duplication.
Modern contact wetgate printer produces very sharp images, rivaling the original negative, and no white halo around objects.


Even a fine grain from 1950's (if very well made) would not render a resolution loss that we would noticed in a 480p or even 720p video !!!!!!!!

Count generations it's not a reliable way to determine a film resolution, or resolution loss compared to camera negative. Film duplicating . The film emulsions used for duplicating make the difference, as modern fine grain film emulsions are great, and also the printer used a bit of and lab work.
Many early films have bad dupes cause they did not have or did not use good film emulsion. Fists fine grain stocks only appeared in the late 30's.

Even so... Take a look at All QUiet in The Western Front : It looks very good, and it's a lavander, but not a fine grain. It's a ortochromatic emulsion used for film duplicating in 1931 or so, but the silver grains was ver similar to a common print of that time. The lavander of that era was more to help keep the contrast low.



This Critirion video It is smelling like a agreement to promote the new film scanner !!!!!!!!



Anyway, when industry will create a good film scanner, able to handle skrunked old nitrate films and also have wet gate option ???

There are a number of issues with how this promo video describes the problem. First, sprocketless scanners have existed for quite some time now. There's NOTHING novel about that. The issue has always been how to handle the focus problem caused by the film not sitting flat in the gate. Lots of sprocketless scanners couldn't handle that. Later pin-registered scanners solved that problem, but the warpage problem also creates horizontal and vertical shifts on splices, especially with nitrate, and that's something you can only solve with computer software.

There no longer is ANY need for wet gate to be built into a scanner, as computer software is capable of removing the scratches. To me, this preferable to the softness that occurs from using wet gate. It just a matter of being willing to pay for the digital restoration work.

Why the Criterion team waited over a year for an Arri scanner capable of doing what they want is a mystery to me when this scanner (introduced in 2006!) was already available and being used by companies like Warner Brothers:

http://lasergraphics.com/pr060612.html

It's sprocketless and they say it can handle badly warped film and has a 4 side pressure plate to maintain focus. It uses LED illumination to eliminate heat.

Derek

Modern contact wet gate printers can handle a fine film duplication.

Frame by frame it's possible to press a bit the film source to the virgen film, solving the problem about warped film.


Scratchs are still a serious problem, even with modern film restoration. it's always better to get a clean image, removing scratchs by filling it (when scratch did not rerached emulsion) than using a digital tool.

There are some abrasion (many scratches) that are so intense that no digital tool can handle. Metropolis 16mm print with the missing scene are a example. No digital toold was able to remove the pattern of abrasions. The abrasions was printed in the 16mm, since it was originally from a overprojected 35mm print from where the 16mm was shot.

Audio Audio Audio. I guess the audio is as good as it can get. I still can't understand the words on the new blu-ray. I never have been able to enjoy the movie because of the audio track/accents.

Since that is true, I was going to watch it tomorrow with the audio turned up much louder than normal (while the roommate's out of the house). But...

I've just now downloaded a file (srt) with English sub-titles for the film and will be splitting my attention between screens, TV and laptop. http://www.subtitlecube.com/download-su ... he:1484926" target="_blank" target="_blank" target="_blank" target="_blank

edit: I should explain that the .srt file can be opened in a text or word processor to see the words; you don't have to be running a movie file on your computer to see the words. The .srt file simply has the words and the times in the movie when they are spoken...

(The Criterion Blu-ray has no sub-titles, as far as I can see, for the deaf, English as a second language, etc.)

Rick

Scratchs are still a serious problem, even with modern film restoration. it's always better to get a clean image, removing scratchs by filling it (when scratch did not rerached emulsion) than using a digital tool.

Wrong. Wet gate always leaves the image softer than non-wet gate. IIRC, the recent Cinerama restorations were deliberately NOT wet-gated for just that reason. They used Image Trends Digital-ICE processing to remove the scratches. Wet-gate is cheaper, but not preferred.

There are some abrasion (many scratches) that are so intense that no digital tool can handle. Metropolis 16mm print with the missing scene are a example. No digital toold was able to remove the pattern of abrasions. The abrasions was printed in the 16mm, since it was originally from a overprojected 35mm print from where the 16mm was shot.

I've read through every available written piece on the restoration of the discovered Metropolis 16mm footage. None of them have suggested that digital couldn't do it. It seems to have been strictly a money decision, as that much scratching could not have been removed using an automated method. It would have required human intervention at every shot, which costs big money to do. Money that likely wouldn't have been recovered by sales of the restoration...

Derek

Derek, all film copy make image a bit softer, but usually not much, and certainly not the like the example the Criterion clip showed, especially about the weird halo.

Optical printer make image softer and contraster, cause projects the frame image to the virgen film using lenses.
Contact printer creates sharper images, cause the film source is in contact with the virgen film, and the light source tends to have light near parallel.

In a wet gate step contact printer, the liguid from wet gate can create a layer between the film source and the virgen film, but it a very thin layer. If make image softer is not as showed in the example, unless the printer it's not good to join the source and virgen films flat.
I presume the warped film source it's a problem if the virgen film and source film are not compressed one to each other. In cases like that it's a slow copy, frame by frame, expansive and very carrefull.



Metropolis digital restoration for the new 26 minutes footage, was very expanasive.

The abrasions on metropolis are too much intense and no digital tool can handle, cause it creates a texture an leaves no clean base of comparisom to allow remove the scratches.
It will be solved only in the future, when better tools appear.



Anyway I imagine that a scan without wetgate, and a scan with wetgate, could be combined. A software would analize the image with scratches, compare with "sister image" with no scratches, and the difference would be easilly found, as have different light values. So only the area of the scratch, in the no wet gate scanning, could be replaced with the image from wet gat scratch free.
But for that would require two scans, and precise alligment of both scans.


Cintel had a film scanner, or telecine, that use a alternative to wetgate. It was called Oliver, and uses refraction properties of light to certains optical devices to avoid get the light refracted by the scratch. A scratch as we see it's the refraction of light caused by the fact a portion of the film base (transparent plastic) was missing.

Derek, all film copy make image a bit softer, but usually not much, and certainly not the like the example the Criterion clip showed, especially about the weird halo.

Untrue, I've been successful in making copy negatives where the print from them was indistinguishable from prints made from the original negative.

Optical printer make image softer and contraster, cause projects the frame image to the virgen film using lenses.

Yes, it's softer/contrastier due to the lenses.

Contact printer creates sharper images, cause the film source is in contact with the virgen film, and the light source tends to have light near parallel.

Yes, contact is better than optical in that respect.

In a wet gate step contact printer, the liguid from wet gate can create a layer between the film source and the virgen film, but it a very thin layer. If make image softer is not as showed in the example, unless the printer it's not good to join the source and virgen films flat.
I presume the warped film source it's a problem if the virgen film and source film are not compressed one to each other. In cases like that it's a slow copy, frame by frame, expansive and very carrefull.

Right, you can't do a contact print of a warped source negative/print. Scanning technology eliminates all this. In most
cases, you'll get a normal frame. For the ones that aren't 100%, there's now software to re-render them flat. Scratches
can be removed digitally and without the softness added by the wet gate fluid.

Metropolis digital restoration for the new 26 minutes footage, was very expanasive.

And expensive, too!

The abrasions on metropolis are too much intense and no digital tool can handle, cause it creates a texture an leaves no clean base of comparisom to allow remove the scratches. It will be solved only in the future, when better tools appear.

Before posting this, I went back to examine the restored Argentina frames more carefully. Some scenes show signs of lightening or softening of some of the scratches, while the darkest scratches in the same frame remain sharp. There are certainly scenes so badly scratched that each frame would have to be retouched by hand. Most of them however seem like you could use the software to identify the scratches, and then a human would have to pick which ones to process and by what how much. I disagree that digital cannot handle this - it certainly could be improved a great deal over what was done for the 2010 restoration if funding were available. As it was, I think they spent something like 600,000 Euros.

Anyway I imagine that a scan without wetgate, and a scan with wetgate, could be combined. A software would analize the image with scratches, compare with "sister image" with no scratches, and the difference would be easilly found, as have different light values. So only the area of the scratch, in the no wet gate scanning, could be replaced with the image from wet gat scratch free. But for that would require two scans, and precise alligment of both scans.

What you suggest sounds remarkably like WB's Ultra Restoration process for three-strip Technicolor negatives. Each negative's
scan is merged/aligned, and the UR software starts picking out differences for the human restorer to take action on.

Cintel had a film scanner, or telecine, that use a alternative to wetgate. It was called Oliver, and uses refraction properties of light to certains optical devices to avoid get the light refracted by the scratch. A scratch as we see it's the refraction of light caused by the fact a portion of the film base (transparent plastic) was missing.

Oliver was introduced in 2002, and still remains part of the Cintel products. It sounds similar to the Digital ICE process, developed by Kodak and licensed by Image Trends (and refined further by the team that developed it at Kodak who are now at IT). Digital ICE uses infrared scanning, and Oliver uses a different process (measured reflectivity) to identify defects in the films. Without the original 35mm scratched nitrate to work from, I don't believe either process would be of any help to the Metropolis project.

Derek

Derek, I know the best B&W fine grains duplicate stock of today can make internegatives whose prints would look nearlly identical to prints from camera negatives, if very well made (shot &developed). That's what Koday says in their site.
Film duplicate can very vastly in quality, and that's why I don't like reference like count generations to judge image quality.

Take some frames of a very warped film with scratches , clean it, and in a dark room, place it over a glass with liguid to fill scratches, place a virgen duplicate film over it, and another layer of glass. Make some pressure to turn the warped film flat while over the virgen film, and expose it with a parallel light source.
I supose it would work to create a copy nearly as sharp as the original. I never saw, but I presume exist some printer able to do that, but in a slow process, frame by frame adjusting each one.

There are some chemicals that can make the film base smoother, more flexible, but it's used in last cases, when films are in danger of be lost and needs to be copied, cause after the chemical treatment the film tend do desintegrate in few months.
IThe restoration of A Trip to The Moon was one case, espite of be copied with quality digital cameras. But wasn't due scratchs, but to be able to handle the film with less risk.

Some scanner, if do not eliminates the scratchs, at least could try dettect all preciselly, and sending a map of each scratch in each frame to make the digital restoration easir.

I once imagined a way to not just send the position of the scratch in each frame, but send information about the depth of the scratch, buy using a liquid with a dye. The deeper the scratch the more saturated if would get, due the dye get deeper in the depression caused by the scratch.
For emulsion scratches it would be usefull.


I know Metropolis 16mm prints have scratches printed in from the 35mm original print from it was shot, and not due the 16mm film base scratches itself. Indeed I said that some posts above.