
We emailed Eric Grayson to find out more about how the restoration of the color sequence in Seven Chances was done. Here are his responses:
NITRATEVILLE: Your blog is called Dr. Film, you've been known to sneer the word "digital," at Cinesation you told how you soaked a print of Song of the Open Road for a year or something to get it to go through a projector flat (it looked brand new, by the way). In short, you're well known as an advocate of film preservation— on actual film. Yet you did this digitally. Why?ERIC: I'm not sure it could be done with photochemical restoration. It could barely be done with current digital tools. The elements were in very bad shape.
Kino did a standard issue of this in the late 90s that was mastered from a different print at standard definition. I remember examining that transfer and showing that the color might be restored using computer technology.
For the 2011 transfer, they used a nitrate print of the film from LoC. In theory, that should have been better, but it wasn't. The problem with two-color Technicolor is that it's highly unstable. It's not like the Technicolor we remember later on, the dye transfer method. The old Technicolor prints faded very badly and many turned to goo in the cans. In those days, the used two pieces of film that were welded together with a glue that tended to swell, and it caused the color to fade, especially the greens.
But we've seen lovely restorations of The Black Pirate and The Toll of the Sea with excellent color. Yes, we were lucky with those. In those cases, a Technicolor negative survived, on black and white film. It only needs to be reprinted on modern stock. With
Seven Chances we simply have the prints that have survived and been preserved over the years.
So you used the earlier master as well?Yes, the master from the earlier Kino version. It was made from a film print had been copied many years earlier, probably by Raymond Rohauer, and it was done on Eastman color stock in the 1970s. Eastman fades, too, so I had that to contend with as well. Technicolor fading and Eastman color fading— and nitrate decomposition, which causes red flares on the left side of the picture.
The color in the nitrate print had faded so much that it was basically not there anymore. I could get a little density out of it, but not enough to make it worthwhile. The old master, despite being a copy on faded Eastman, hadn't faded as much, and I was able to get the color out of that. So I used the new HD transfer by Kino for the background and aligned the color from the older version with the new transfer. This way you have the best of both worlds: the color that filters well, and the sharpness of an HD transfer.
The BBC restored some Dr. Who episodes this way, didn't they?Yes, their process is fairly similar. The BBC had some bad color recordings of several episodes along with sharp black and white master material. The concept is based on an old color model that is different from the Red, Green, Blue idea. You can also map color as hues overlaid on a black-and-white picture. It's an old engineering conversion.
But the color image must not be as sharp.That's true, but it doesn't matter too much. The human eye is easily fooled by a softer color image so long as the black-and-white beneath it is strong and sharp.
I assume you used some sort of movie program like Final Cut Pro on this.No. Final Cut Pro doesn't have color correction to the level I needed. I had to load it frame by frame into Photoshop. Each shot of the film had its own set of color curves based on the fading for that particular shot.
Then what made it more complicated is that the old transfer was done at 30 frames per second, which was the NTSC standard. The HD transfer was at 24 frames, which is the film standard. I had to convert back to 24 to make it work. That can cause artifacting and other problems.
The commentary track says it took 80 hours to get this fixed. Is that accurate?Actually, it's a little low. Bruce Lawton, who was consulting on this project and did the commentary track, recommended that Kino contact me based on what he'd seen of my attempts of this process earlier. They'd had problems doing any sort of restoration, and they were working on a tight deadline. I got the material on Friday and it had to be back to them by Tuesday AM. I sent it back over the internet to save time.
Did you sleep?A little, but I didn't leave the house or shower. It took me a day and a half to figure out how to automate this process, but even then I had to compose a Photoshop script that forced me to click by hand on the movie one frame at a time. Then I had to figure out how they've changed the aspect ratio of the pixels in HD vs. old NTSC, which caused a lot of headaches. It took about two hours to run a single iteration of the restoration; I couldn't look at it until it was finished, and it meant clicking the mouse 2000+ times on the Photoshop script. I think I ran five iterations before I got something I could use.
I was a little surprised that it worked as well as it did. Bret Wood at Kino was very nice. He'd been working with the HD transfer and wasn't holding out much hope. I don't blame him: I couldn't get it to work either. I think he was pleasantly surprised that it worked and then he asked me to record a commentary for it.
Since you're an advocate of film restoration, would you like to see this sequence output to film?I'd like to see all the parties and archives involved cooperate and produce a new 4000-line full-frame color restoration. I suspect that very high resolution scanning with high color bit rates would enable a much better restoration than I was able to do. You'd have to go back to the film elements to do it. I hope it happens before all the materials get any worse.
But then your restoration would be obsolete...No, it would be a stepping-stone toward getting it done the right way. If it leads to a better, higher resolution restoration, then nothing would make me happier.