How film restorers brought The Apu Trilogy back to life

How film restorers brought The Apu Trilogy back to life

A restorer works on a section of celluloid (Photo: Criterion)
A restorer works on a section of celluloid (Photo: Criterion)

In entertainment, an awful lot of stuff happens behind closed doors, from canceling TV shows to organizing music festival lineups. While the public sees the end product on TVs, movie screens, paper, or radio dials, they don’t see what it took to get there. In Expert Witness, The A.V. Club talks to industry insiders about the actual business of entertainment in hopes of shedding some light on how the pop-culture sausage gets made.

The films of Satyajit Ray’s The Apu Trilogy are not only the best and most influential in Indian cinema, they stand among the greatest made anywhere, tracking a life from childhood to parenthood, through despair to triumph and maturity. Yet despite their acclaim, it’s likely cinephiles have never fully experienced them due to the depreciating audio and visual quality of these films since their release about 60 years ago. Beyond the normal wear and tear that comes with time and making prints, the films’ original negatives were housed in London’s Henderson Film Laboratories when that warehouse caught fire in 1993. Many reels were burned beyond repair, and the rest suffered heavy damage.

Enter Criterion, the beloved restorer and distributor of classic films. Years ago, Criterion decided to try to resurrect Ray’s masterpieces, a process only now at completion. After a massive worldwide effort, the films have been restored to shockingly good visual quality; they probably haven’t looked this clear since their premiere, if ever. The re-release of The Apu Trilogy will begin at New York City’s Film Forum on May 8, followed by a broader release throughout the summer. The A.V. Club spoke with Criterion technical director Lee Kline about the process of returning Apu to cinema-level quality, and the importance of film restoration.

The A.V. Club: How did it come about that you would be restoring The Apu Trilogy?

Lee Kline: We had wanted to put these out for a while, but they were kind of eluding us. Sony had put the trilogy out about 15 years ago, but its rights had lapsed and reverted to different owners. Because the new owners were in India, they were difficult to make a deal with. But eventually, after a lot of negotiating back and forth, we signed a contract and put them on our schedule.

AVC: How does that compare with other projects?

LK: We pick the films in a variety of ways. Sometimes, if we’ve heard a lot about a title, we’ll seek it out and try to get the rights. Other times, filmmakers will reach out to us to include something in the collection, or people will write to ask us to work on something. It’s always different.

AVC: Once you have the rights, how does the remastering process work?

LK: You’re only as good as the quality of your film. It’s obviously best if you can start with the negative, the thing that actually ran through the camera, which is much better than a copy or print. Something that goes through those analog processes may have additional problems like flickers or a lack of grayscale, or dirt or scratches, anything like that. Obviously a negative can have those same issues, but it’s far more likely to be worse in subsequent prints.

The Apu Trilogy was known to be problematic because there weren’t good film elements to begin with. We had heard that the negatives had burned in a fire, so we didn’t even consider them at first. Instead, we started looking at anything else that had existed, various copies and prints, and it was disappointing. Everything was dirty and scratched, and while some parts were better than others, it just wasn’t very good.

Then I went to L.A. to work on some stuff at the Academy Of Motion Picture Arts And Sciences, and the people I was working with mentioned that they had a bunch of boxes with stuff from The Apu Trilogy in the Academy Film Archives. It turned out to be the original negatives, the material that was damaged in a fire. I have to thank the Academy on this, because someone there had the films brought to the U.S. 20 years ago, after the fire happened. Someone knew about the fire but still asked for everything to be sent over rather than having it be thrown out. Even though there wasn’t a way to restore it at the time, someone was smart enough to say, “don’t throw it out, keep it.” That saved the movies.

The people at the Academy asked if we wanted to see the material, and they sent it over. I had never seen film that was burned in a fire, and I really had no idea what to expect. It was very interesting. There were some sections that weren’t as exposed to the fire, but the outsides of the reel were really bad. The film was flaking and brittle, just falling apart. It was extremely dry and fragile, with a lot of cracks. This is just looking at the reels, mind you, not holding it up or spooling it through a projector. We didn’t even know if we would be able to do use a projector because it was all so fragile.

It’s sad holding film in that condition, and it’s doubly sad to imagine the film that didn’t make it. You’re always surprised when you open up a film can from 30 or 50 years ago; I compare it to a bottle of wine. A vintage bottle of wine sounds very exciting, and it looks great on the outside, but then you open it up and it smells like vinegar. Film does the same thing, and actually film also smells like vinegar. It’s a terrible thing to open up a film can and see something rotting away, which is why preservation is so important. If stored right, film can last a really long time. We’ve seen film from 100 years ago that still holds up. We’ve touched the original negative of Charlie Chaplin’s The Kid, and the film is in great shape because it was stored well. How many films from that era are around in that condition today? Very few. Before someone realized that prints needed to be made from copies, a lot of prints were made from negatives, which meant the negatives just fell apart.

So I was in L.A. working on something else, and at one point I was working with some guys scanning film into a computer. I mentioned the burned film to them, and we put a sample on a light table to examine it. We couldn’t really tell what it was like, not even with a magnifying glass. All we could see was a warped, and very brittle, image. And this is from the least-damaged reel, where only a little was flaking off. We decided after that to scan a section to see what the image quality was like, and when we did that, lo and behold, there was actually an image in there, one that was pretty good. We all got very excited, because it seemed like negatives that had been burned in a fire could be brought back to life and used for something.

AVC: How did that reel compare with the others?

LK: Because it was the best reel, the results of the scanning test were a little misleading. There were some sections that weren’t usable, and not even all the reels were there. I think we had nine reels of Pather Panchali, 11 reels of Aparajito, and just two reels of Apu Sansar (The World Of Apu). That’s out of maybe 13 for Aparajito, since you assume 10 minutes a reel. So they were not complete. We got excited too soon. But we thought that if we could get it all to work, that would be pretty amazing.

Now, I had never worked with film that had been damaged in a fire and I didn’t know anyone who had. We called Kodak for help; we called Fuji; we called every lab and film person we knew or could think of. It turns out, not many people know how to deal with it, since a lot of the people who would have experience with this are either retired or dead, and most fire-damaged film isn’t salvaged in the first place. No one had a very good answer; they would say things like, “You could try this, but that might not work, so you could also try this.” That wasn’t helpful because we figured we only had one shot at this given the delicacy of the film. If we messed up, that could destroy the film. So we needed something that not just worked, but a process that could let us remaster it in super-high resolution.

We had been doing a lot of work with this lab in Bologna, L’Immagine Ritrovata, which is probably the premiere lab facility in the world. Those technicians are the ones who know how to do basically anything with film, because they’re among the last people to really deal with it. They’ve transitioned very well from film to digital, but they still have film technicians, something like 120 people. They get film from all over the world, and not only do they know how to deal with pretty much anything, they’re not scared to take it on.

One technician looked at the Apu reels and said we would need to rehydrate it in a solution of glycerol, acetone, and water.

AVC: Do you have a chemistry background?

LK: I don’t, no. I have a video background. But I know film, I know how film works, I’ve been in labs where film is being developed. I’m not a chemist, but I can look at a piece of film and know what it is, by reading edge-code and things like that. Just by handling it for 20 years, you know how it works.

After the solution, the technician said we would need to repair all the perforations, cutting out the old ones, putting on new ones, and taking apart splices and re-splicing at every cut. The films had been covered with a layer of wax back in the ’50s. Ironically, it had been put there for protection, but it melted into the film in the fire, so this would also have to be taken off. Basically, everything that could need to be done for a project would have to be done for this one.

It was the kind of undertaking that would cause some people to run away, but we all knew that the films of The Apu Trilogy are major, major movies. This was a chance to do something that could bring them back to life.

So, we had L’Immagine Ritrovata do a test on a reel. They ripped apart the negative and put it back together, scanning it in a few different ways in super-high 4K resolution. They did what’s called wet gate, or wet-transfer film gate, which is where you run the film through a solution to help with things like scratches and the wax buildup. As the test went on, we became increasingly sure that we would be able to do this, but we were concerned because steps like the wet gate kept adding to the cost.

AVC: What is the cost of a restoration project like this?

LK: It’s difficult to put a price on it. L’Immagine Ritrovata did the analog remastering and we did the digital remaster. So from our end, since we have an in-house lab, we can dedicate all our resources to the project. We don’t have to watch a clock or have a lab say that we’ve used up the 36 hours that we’ve paid for, meaning pay more or quit. It can be a blessing and a curse to have an in-house lab, because we would want to spend forever improving the print, but eventually you need to actually put the film out. I would guess that the cost would run in the hundreds of thousands of dollars, which may not be enough to justify the release, though the results makes a good case for film restoration.

A before and after comparison of a frame in the film (Photo: Courtesy Janus Films)

The Academy ended up being a great partner for The Apu Trilogy, and by partnering with both them and the lab, we were able to finance the project in a way where we felt we could really do justice to the films. From there, we began the process of cleaning them up and getting them back out into the world.

Like I mentioned before, not every shot or frame from the negative was usable, which meant the second step of the process was searching for suitable replacements for the moments we couldn’t get from the negative. The BFI [British Film Institute] in London had some; there was some footage in the Harvard Film Archive; the Academy had some. We scanned multiple pieces of the film and found which source had the highest quality for every moment.

AVC: If you didn’t have all the reels, were you comparing the footage you had to a script?

LK: We had release prints that we knew were complete because they had screened at festivals. They weren’t in good condition, but at least everything was intact. We made an EDL [edit decision list] from those. It was basically like a puzzle, where we would figure out which source had the highest quality of a particular moment. Once it was all slugged in, the films were complete. The EDL became our roadmap to the restoration.

AVC: How long does it take to restore a film?

LK: A typical project could be anywhere from a few days to a week, maybe three weeks. It depends on the quality. We’re doing The French Lieutenant’s Woman right now, and since that’s a fairly contemporary film, it hasn’t had the time to get damaged, and it will only take a couple of days. Our technicians will go through and clean up dust and specks and little scratches, that kind of thing.

The Apu Trilogy was far more complex, and it took months. We started talking about it and looking at the film almost two years ago. The physical repair of the negative took over a thousand hours. That was literally someone cutting and snipping, putting the sprockets on, changing the splices. That was done by L’Immagine Ritrovata, which had a team that worked pretty much around the clock on it. It was tedious work; it was slow; and it must have felt pretty endless. Going frame-by-frame like that is a lot of work, a lot of busywork, and you have to be a certain type of person to do the job. Now, the people who have dedicated their lives to this tend to be those kinds of people, and they tend to be pretty good at it. Still, when I went to visit them one time, all the technicians looked at me with daggers in their eyes because of how tedious the work was. I joke that because they only had part of the negative, in some respects they got off easy.

The digital restoration was all done here in-house at Criterion, and we basically stopped working on everything else to finish it and make our deadlines.

AVC: Do the parts of the film from different sources end up looking different?

LK: We’ve tried to put the film together in a way that lets you forget that you’re watching some stuff from one piece of film and other stuff from another piece of film. That’s difficult. We tried to do that with the color grading, the difference between the grays and the blacks, as well as with tricks like grain management, which is where we’ll use a computer to remove film grain so that it looks more like the negative. There are a lot of tricks, and sometimes it doesn’t work and you’ll be jarred out of what you’re watching because the visual quality changes so dynamically. There’s a section of Pather Panchali that was very difficult because the only source film we could find was very mediocre. It had a lot of scratches, which are the hardest thing to remove because they stay in one place, meaning you can’t steal the information from other frames.

The process is, we’ll load a film onto a computer and our technicians will look at it frame by frame, trying to identify flaws within the image. If it’s a piece of dust, the software will look at the frames before and after the one with the dust and be able to determine what that pixel, essentially, should look like instead. So if it’s a speck on a wall, it will know what the color should be based on the color of the previous and subsequent frames. Sometimes the computer will make a mistake and want to eliminate things like stars or raindrops or a guy’s pinstripes. Little details like that, which are essentially moving specks, read as flaws even though they aren’t.

AVC: Is this easier because it is in black and white?

LK: Actually no. The computer does better when there’s more contrast and color to work against. It helps if there’s a good grayscale, but because that fades when films decay, black and white is typically more challenging.

AVC: Is there a limit to how well a film can be restored? Will it ever look as good as it did at its premiere, for example?

A before and after comparison of a frame in the film (Photo: Courtesy Janus Films)

LK: I think you can actually make it look better than the way it looked originally. Assuming a film was kept in good condition, if you make a 4K DCP [digital cinema package] from a negative, if the DCP has been restored and cleaned up, it will probably look better than a print first looked. Not to say that the first print of Pather Panchali didn’t look great, but prints can flicker or move around, they have reel-change marks punched into them. Those analog things can happen, and they can be removed in a digital print. Our Pather Panchali looks great, and I think that anyone who looks at our version compared with a print would say ours looks better, unless they’re just a celluloid purist. There is something great about watching a print in 35 mm, and of course that’s how everyone saw it back then, but it is just a fact that digital versions can eliminate some of those issues.

There is a limit to what you can do because you can’t change how a film was originally made; you can only help it. By using film from multiple sources, it will never look like the original camera negative. In Panchali there are scenes I wish we had been able to get the negative for, but those were burned in a fire. There wasn’t anything we could do about that.

Another issue was that Satyajit Ray used a lot of dissolves, which are optically printed sequences. Today, doing that electronically is easy, but in his day, the way you would do such a transition was to print two pieces of film and make the transition optically by using a darkroom enlarger. There were optical-printing houses that did this kind of work, and those sections of the film are always going to look inferior because they were done poorly or crudely or quickly. They have problems like light coming in from the side, or they click in rather than dissolving smoothly. Because Aparajito has so many dissolves, it really pulls you out of the film because at each one, the image will change to super-light or super-dark for a second before changing back to where it’s supposed to be. So not only were these crappy pieces of film, these parts specifically don’t hold up to the negative. It was a problem.

AVC: Can a computer be used to smooth over issues like that? Would that be ethical, as far as manipulating images?

LK: You’d have to ask Ray that. He was an advocate for the preservation of his own films, and I think most filmmakers would be keen to use tools to enhance their films. If you manipulate the image to the point that it begins to look digital, that’s when—assuming the filmmaker isn’t around anymore to advise you—you’ve gone too far. That’s how we see it.

Now, if the filmmaker is still around, you can ask them what they think. We’re working on Truffaut’s Day For Night, and [cinematographer] Pierre-William Glenn is still alive. He helped us with the restoration, and when we were doing it, we found a lot of crazy close-ups that looked super grainy and weird. I asked why they were like that, and he told me that when Truffaut was editing the film, there were moments when he wanted close-ups that he hadn’t filmed. So, he optically printed other shots—medium shots, say—and blew them up into close-ups. He flung those in rather than reshooting, and by blowing them up, they became ugly pieces of film that really looked inferior to everything surrounding them. Glenn said he wished they could be fixed, and we told him we could with various grain techniques. I asked him if he thought Truffaut would be okay with our doing that, and he said Truffaut wouldn’t have cared. As a director, he made the choice because he cared more about having a close-up there than he did the quality of the image. It seems safe to assume that he wouldn’t have wanted anyone to think about the filmmaking at that moment, so he wouldn’t have been averse to easing the quality of the image.

To me, that conversation really put things in perspective. If Glenn had said that he didn’t like the way Truffaut color-corrected something, that he thought a green should be a red, that would be inappropriate. That would be an artistic choice, not a technical one, and it seems like it would be crossing a line.

AVC: Sometimes I’ll watch silent films and I’ll note scratches and so forth during the inter-titles. I’ve always wondered whether a restorer could just create new title with the same font and design and sub that in.

LK: Well if you wanted to make an inter-title you can just freeze a frame, scrub the image, and use that one frame for the card. We have all the tools to stabilize the image and remove any flicker. Now, a lot of people don’t like that because you feel like you’re in a digital world for a second. Take away the flicker, even if you get rid of the scratches, and it feels like a lifeless image compared to what’s surrounding it.

AVC: How do you know if something is as restored as it can be? Are different cameras from different eras capable of capturing different image qualities?

LK: Cameras improved over the past century, but it is amazing how good celluloid technology was from the start. If you look at a piece of filmstock from the ’30s, that negative picked up a lot of information. Certainly lenses got better and maybe the amount of grain changed as technology improved. But celluloid was so ahead of its time for so long that we estimate a 35 mm frame is basically a 5K image, digitally speaking. Certainly celluloid was capable of amazing images when The Apu Trilogy was filmed, to the point that we would have been doing a disservice to the image if we had scanned it in anything other than 4K, which is the most manageable of the high-resolution formats right now. Projects like Lawrence Of Arabia may use 6K or 8K, but that’s a bit different because those were on 70 mm celluloid so there was more information to scan. If we hadn’t used 4K for Apu, it would still be a restoration, but by using 4K we could really keep all that information. The quality is so good that you could translate it back to celluloid if you wanted. It is mind-blowing to watch these films after they’ve been scanned, graded, restored, and shown theatrically on a 4K projector. You can’t believe it was made 60 years ago. It just doesn’t look it at all. It looks super-contemporary, other than the subject matter and the style of the filmmaking. Modern-style films from the ’60s or ’70s look like they were shot yesterday, assuming they’ve been stored properly.

AVC: Is it the same for audio?

LK: Audio is a little different. You can take audio off of prints, or soundtrack negatives or off of magnetic audio tracks, especially when it comes to movies from Apu’s era. We fared better with the audio for The Apu Trilogy, but we did the same thing that we did with the video, piecing it together from different sources. Some copies had more hiss in the audio; others had crackles or were missing sound effects, oddly enough. Those are things you can fix with a computer.

AVC: Are people now pretty conscious about what they need to do to preserve film?

LK: A lot of cultures are very good about preserving their film. France has the Cinematheque Francaise, Spain has its own archives. We have the Library Of Congress, and studios keep their films pretty well preserved. Everyone is doing their part. But India isn’t doing a very good job of preserving its film, which is quite sad considering how big its film industry is. A few years ago there were only two guys in India who gave a shit about preserving film, and one made a documentary about the other, called Celluloid Man. It’s three hours long and was basically a wake-up call about how few people were doing preservation. The director [Shivendra Singh Dungarpur] started going to schools, trying to get them to offer courses on film restoration so everyone would understand what it was about. Earlier this year, he invited experts from all over the world to come to India to talk about the issue. I went, along with my sound guy, to talk about working on The Apu Trilogy. We were just a small part of the project, but there were 50 students or so who were selected to attend, out of hundreds of applicants. Hopefully, they all went back home and started their own restoration projects.