Criterion Goes to the Source(s) For Three Different Cuts of Mr. Arkadin

Perfectionist auteurs have become well known for monkeying with their movies up until the last minute and beyond, but how do you release a DVD director’s cut if you don’t know how the director himself would have finished the film? And how daunting is that job when the director is as formidable a legend as Orson Welles himself?
That was the challenge faced by New York’s Criterion Collection when it decided to restore Welles’ 1955 film, Mr. Arkadin. “The film had been re-edited many times by many different people in various territories, as Welles had been taking too long to edit it himself,” recalls Criterion’s technical director, Lee Kline. "There was film everywhere, but each was a different cut."
Those cuts included two different Spanish versions, a European release version (called Confidential Report), and two different U.S. versions. "In fact, the film was never finished," notes project producer Issa Clubb. "[Welles] was ultimately kicked off the project and sued by the producer, and there were all these different versions floating around. To make matters even more confusing, some even had different actors. The European version, which we released on laserdisc in the mid '90s, was completed but mangled by the producer after he'd kicked Welles off the film."

"Because no definitive version of the film exists, we decided to make that the point of the release and give the viewers as many possible versions as we could," says Kline. "So we ended up giving a full restoration job to three different versions: an early fine-cut version known as the Corinth version, Confidential Report, and finally an entirely new cut that includes as much of the existing material from all the different versions as possible."
So, in consultation with Welles scholars from all over the world, and with filmmaker Peter Bogdanovich also contributing, Criterion set out to construct an new version of the film. It was primarily created by Stefan Droessler of the Munich Film Archive and Claude Bertemes of the Cinà©mathà¨que Municipale de Luxembourg. "Peter had talked to Welles at length about the film,” says Kline, “so we had a lot of expert input, and we tried to create the most comprehensive version, one that Welles would probably have made if the film hadn't been taken away from him, and one that includes the most existing material, including some archival material, arranged based on clues Welles left in interviews or on his style."
Kline says their quest involved going back to all the source material-two 35mm internegatives in Spain, an original camera negative in France (“cut up beyond belief”) and a fine-grain dupe positive that documented a different cut made from the negative. The team also looked at the 16mm Corinth version Bogdanovich discovered in the U.S. in the '60s, which was named after its owner, Corinth Films. "That turned out to be very important as it was clearly very similar to what Welles probably wanted," notes Kline. However, there were major problems with this version. "It was worn and used and scratched, and it also had a major shake problem," he adds.
But each version contained shots that were missing from the others. "So we had to go through each version very carefully, and then, if we found a duplicate of a particular shot, we had to determine which was the better quality [version]," says Kline.
Having tracked down all available versions, the team ran into another problem. "No one was shipping their footage, so we had to go to France and Spain and find a lab that could transfer everything in HD and in 23.98psf, and to the best possible HD tape format that most international labs would have," reports Kline. "We chose Panasonic's D-5 system and recorded in 23.98psf."
This became "quite tricky," Kline recalls, because the labs in Spain used Cintel's C-Reality telecine technology while the labs in France and the U.S. used Spirits. Kline traveled from country to country to supervise all the transfers. The C-Reality system was a great help since its wet-gate processing system filled in scratches and dirt. In France, Kline went to Scanlab, part of LTC, and used both the original negative and the fine-grain. All transfers were done in 1:33:1 aspect ratio and a 16×9 HD frame.
"The transfers were all done in a best-light form where we did a general grade, knowing that we'd later grade further," he explains. "That let us record it all to tape without doing all the shot-to-shot color correction until all the editing was completed."
The French transfer was particularly "stunning," says Kline, "because the original negative was gorgeous and in good shape." Unfortunately, the French negative was missing key scenes that had to be tracked down elsewhere. One U.S. version was transferred on a Spirit datacine at Post Logic, New York, while the 16mm version was transferred at Modern VideoFilm in LA, also on a Spirit.
Once all the source material was transferred, the team edited using Final Cut Pro HD. "We brought in Peter and all the film scholars, who made sure we were getting the right version, and Issa pulled all the best shots we needed from the various elements and kept them all in Final Cut so we could pick and choose,” Kline explains.
"We replaced probably 99 percent of the 16mm material with much better 35mm material," reports Clubb. Once the edit was complete, the team took the final output to D-5 and did one more tape-to-tape pass to even out the look. "That included density changes, contrast, blacks, whites and luminance-anything to make it all look the same," Clubb says, "and the results were pretty amazing, considering all the different sources and machines we'd used."
The team was also able to stabilize some shots by running them through Apple's Shake. "We also added sharpness to some soft shots during the tape-to-tape phase at Modern VideoFilm, using colorist Gregg Garvin," notes Kline. "He was very familiar with the film, having worked on some of the other versions, and he loaded everything into a Clipster [made by German company DVS]. This system allows you to digitize the material in HD in telecine. Then you can replay it and color correct, re-size, re-shape, and re-frame as needed, as some of the Spanish material was slightly larger. So we could shrink and resize images as needed." Digital Vision's new Advanced Scratch Concealer with Motion Estimation and MTI Digital Restoration Software were used to further clean up the material.
Restoring Mr. Arkadin was "a bit of a nightmare," admits Clubb. "We try to have a very purist relationship to the films, to make them look like they were shown in theaters on first release. But here that's hard to determine and not necessarily desirable. There was no premiere of Mr. Arkadin that you would want to reproduce. That's what made it so challenging-to wade through all the different possibilities."