DI As Workflow
PDF of Digital Intermediate 2004
While the digital intermediate process has hardly been ubiquitous, it has become a finishing option that is now routinely and seriously considered for almost all major feature film projects. In recent years Millimeter has regularly detailed the DI options, workflows, pipelines, technology, artists, issues, and trends impacting studios, postproduction facilities, and filmmakers.
This issue, Millimeter takes a look at what has recently changed in the DI world–where the technology and techniques are likely to head next as the process continues to mature. To do that, six veterans of the DI wars–all of whom, coming from different areas of responsibility, have spoken with Millimeter about major DI projects in the last two years–offer their recent experiences and opinions about their industry. The group includes two prominent colorists on major features (Stephen Nakamura of Technicolor Digital Intermediates, Burbank, and Steve Scott of Efilm, Hollywood), two DPs well experienced with DIs (Newton Thomas Sigel and James Muro), a DI facility executive (Bob Eicholz, VP of corporate development at Efilm), and a major studio postproduction supervisor (Fox's Patrick Esposito).
Millimeter: What are some of the recent advances or improvements that have given you more control in the DI suite?
Steve Scott: As a colorist, things have changed for the better in recent years, but a lot of it is proprietary advances, since each facility takes its own approach. At Efilm, we have incredible color science, very advanced, that now allows us to make digital files deadly accurate when compared to film. We are now using Barco 2k projectors (in the color-timing suites), so between the projectors and our improved color science, there is a marked improvement in terms of getting closer to approximating the true depth of blacks and detail on film.
Since these improvements are now more visible on the screen, we can spend less time helping the director or DP translate in their mind from what they are seeing on screen, in video space, to what they will see on film. What they see is what they get. There has been a quantum leap made in the proprietary software in terms of making the whole process faster.
Stephen Nakamura: Projectors are a typical example. We have gone through three generations of projectors in just 21/2 years (at Technicolor). That illustrates how fast the technology is improving. Couple that with improvements in technology for conforming, within our color-correction tools, and so on, and it's a rapid advancement. For example, when I started on features in 2002, you could not do defocus at 2k resolution. That might seem like a simple thing, but it was a problem back then. Now, it's very common.
Newton Thomas Sigel: Just look at the degree of resolution–how quickly that is improving. Now, (facilities) are routinely scanning at 2k. Some are on the verge of scanning at 4k, and they are able to color-correct at 4k. Every year, the degree of resolution that we can do for DIs is increasing, as is the number of tools that are available within the color-corrector itself: things like tracking mattes, the number of Power Windows, noise reduction. All those tools are growing in power daily. The initial speed is also greatly improved in terms of scanning speed from when I did my first theatrical DI (for Confessions of a Dangerous Mind in 2002).
Millimeter: Speaking of resolution, what's the feeling about migrating to a 4k data path for DIs, as Efilm did for much of Spider-Man 2?
Bob Eicholz: The 4k data path is something many people are now asking us about, and we expect to increase the number of 4k DIs we do quite a bit in 2005. Spider-Man 2 was 4k for everything but the effects, but we have just started (at press time) on a true 4k data path for an entire film–Ocean's 12.
There are two reasons why clients are interested in this. First, it looks better, especially when you are making show prints off an Estar negative. 4k plus an Estar negative really shows the resolution difference, even for a non-sophisticated viewer. The other reason 4k is important is for archiving. You can save your movie for all eternity so that all the pixels are at the best level possible. I don't see any reason to worry about archiving beyond 4k, but there are people who would debate that.
James Muro: For most release prints, it isn't worth it to spend the extra money, unless it is a really big-budget picture. Most people will never see the difference. But 4k is mainly helpful now for the theatrical release negative–2k is more than adequate for HD and DVD or home video releases. Only the bigger films will be doing 4k any time real soon.
It will be a good development for the smaller films, though, because the price of 2k will come down and become available to people currently limited to HD resolution. Maybe the smaller films will start getting some breaks in price.
Sigel: Once we get to 4k, everyone will want 6k. That's the next big hurdle, and it's just a question of having enough memory at the facilities to do it. The more you can get into doing things like custom curves and secondary color correction without creating any video noise or artifacts, that would be a great development. Once digital projection gets into its big phase and more DIs start happening, I'm sure they will get there.
Nakamura: A typical movie is about two hours long. In 4k, between your original scan, the color-corrected movie, your dirt pass, and so on, you might need 75TB of storage for that one movie. That is humungous. And then, you need to move all that information around properly and quickly. If it's a big effects movie, changes and approvals will come in last minute. You need to scan and integrate cut changes coming in constantly as the production gets to the wire. If you can't integrate all that information quickly, it puts your facility in a bind.
So the problem with 4k is file size and the ability to move it around. We haven't done 4k yet (at Technicolor), but we could and we will. The problem is doing multiple movies at 4k would be difficult.
Millimeter: What is going to be required in terms of pipeline and workflow for facilities to transition into the 4k data world?
Eicholz: Surprisingly, it did not require many changes at Efilm. We always scanned at 4k, so the change was mainly about needing to increase disk space to save data at 4k. It's not technically difficult to do that–it just costs money. But we are using serial ATA disk drives–a new generation of disk drives that are reasonably affordable. We also had to spend money to upgrade some of our computer infrastructure and add storage. We have about 220TB of storage right now, and we'll go up to at least 300 in 2005. That's online fast storage. Then, of course, you have additional offline storage. But the overall workflow–the approach doesn't change much.
But we are a resolution independent facility. For some hardware-based facilities, it might be a more difficult transition, because all their hardware is built around 2k. Some of them would have to go through a significant upgrade to do it, or replace their core color-timing equipment altogether.
Millimeter: Even with these technology improvements, there remains no industry standard regarding viewing environments, calibration, and color. Are you interested in the development industry standards?
Nakamura: There is no universal look-up table (LUT) for doing DIs. It might be nice, but right now, there is no standard. Everyone has to calibrate their own monitoring devices. Some people still color-correct off a monitor, though we only use projectors at (Technicolor Digital Intermediates in Burbank). There are lots of different philosophies. But either way, a facility is really as good as its colorist, but also as good as the calibration of its monitoring device to ensure the look when going back to film.
Scott: Efilm understands the importance of standards. We are actively working with and supporting groups like SMPTE and the ASC in their efforts to develop a common language across multiple viewing devices. That said, at facilities in our industry, you have a real advantage if you have specially calibrated digital projection and look-up tables that accurately emulate film color space. That's a big imperative for all of us. So, if you have that, you aren't likely to share it lightly. Good color is hard to achieve–lots of science and engineering goes into it. Accomplishing that makes us a viable business.
Even as standards evolve, I doubt there will ever be an off-shelf standard solution for all color issues. It's no different than how chemical labs achieve things with very particular mixes of chemicals.
Patrick Esposito: From the studio perspective, it would be nice to have standards. Right now, if you show up with color-timed 2k files from one facility that uses proprietary software, and you go to another facility with their proprietary software, you would have to start from the beginning with the raw scan.
It would be nice for us if everyone was working the same way, but I doubt the facilities would do that, because it removes a selling point regarding their gear and their initial investment. Everyone wants to have their own way of doing things, much like the visual effects houses.
Millimeter: How has proliferation of the DI changed the way you work or collaborate?
Muro: The colorist has become more of a collaborator for me–more involved in developing the look. I'm the type of cameraman who looks for input from everybody anyway, and it's a collaborative artform. As a DP, you tried to do something specific when you lit the movie, and the colorist is the one who really tries to help you stick to those guidelines.
There are now more and more people involved in the process–studio people, especially. They all have their own agendas. But the colorist is the one trying to help the DP maintain his agenda.
Scott: For me, as the colorist, the relationship with the DP is really the most important priority. I'm there to help sell his or her vision. There are more people coming into the process now, and I'm always trying to lobby for what I understand the DP wanted.
That said, the director and others have opinions and must also be happy with the end result. Part of the art of being a good colorist is helping everyone achieve a pleasing result.
Sigel: I'm shooting differently now that I have so much experience doing DIs. For example–doing 'stop pulls.' That's when you adjust your exposure in a shot from one part of that shot to another part of the shot. In the past, we always did that in-camera. But now, we don't always have to do it anymore, when we are doing a DI.
If you do it well in-camera, that's great. But if you are off even a little bit, that's not so easy to adjust. But now, with the DI, if you understand the latitude of the film that you are working with, you can actually put your exposure in the middle and do that stop-pull
or exposure compensation later, during the DI, and you can do it with more precision. I just save that part of the treatment for the DI now.
Millimeter: But often, the director takes total charge of the DI, especially directors who have commercial experience and are well acquainted with the process, or when the DP has another obligation. How do DPs protect their role?
Sigel: I'm doing Superman right now, and the DI was almost a deal-breaker for me with Warner Brothers. Initially, they were not going to put into my contract that they would pay me for it, so I originally declined. They eventually changed their mind, and I took the job.
This is a real important issue for cinematographers. There is only so much you can do as a cinematographer to protect your role, and most of that falls back on how solid your relationship is with the director and producer. Still, many cinematographers can't pass up jobs if they aren't getting paid for the DI. But beyond that, paying the DP for it is a way for people to understand the DI is a continuation of the cinematography process. Precisely because the DI suite can give you such control is why the cinematographer needs to be there. His work does not stop when principal photography ends. DPs are now beginning to understand that what you do in post is part of their job.
But there is a trend the other way. Some directors have backgrounds on commercials and have worked with colorists in the telecine environment for years, and have a strong visual sense. They want to do it themselves, so that is a trend. But in my case, I won't take a job if my contract doesn't say I get paid for the DI.
Muro: Sometimes, there are too many voices in one room, too many chefs to spoil the broth–that is true. Some cameramen move on to other jobs, but others won't be happy with their images unless they are doing (the DI) themselves. You do have the expertise of the colorist, of course, but if the DP isn't there, it will be hard to keep the thing consistent, unless a savvy director is there. The DP is integral to testing and preparing all this, so it doesn't make sense to have him miss the color-timing.
I'm now on my third DI, and I'm learning a lot about how this process can improve the images, and also, be affordable. I'm always selling the idea that we can do the DI affordably, even on smaller pictures like (Paul Haggis' indie film) Crash. But I just finished a job (Roll Bounce) where the post coordinator didn't think a DI was necessary for that type of a movie. We color-timed Crash in just five days at IO Films (North Hollywood), and that was a small film. I always try to demonstrate that it can be done affordably, if you give us a chance.
Millimeter: So, what's the financial truth then? We've heard that the digital intermediate process is more expensive than a standard photo-chemical approach, and yet, we also hear that it can lead to savings in the long run by allowing more efficient creation of the necessary deliverables for the movie's future incarnations.
Esposito: It's not very expensive to take a timed IP and do transfers and go from there for deliverables, so I disagree that a DI can be cheaper because of the deliverables issue. It does cost more money, especially for a major, high-end feature.
The reason to do a DI is not to try and save money. It should be about does it make the movie better? Obviously, with some filmmakers, like a Tony Scott or a Michael Mann, there will be a DI if that is what he wants, regardless. But usually, when I'm talking to filmmakers, and they bring up the DI, since it does cost more, I have to ask them why they want to do it. What do they want to achieve? What look can't you achieve photochemically, or in production, that we have to create digitally in post? That dictates it, as far as I'm concerned. It should be a negotiable process among the parties involved.
Sigel: I have a feeling cost will become less of an issue, except on very small, indie films. By and large, any major studio film that comes along, if they say they can't afford a DI, that is disingenuous. It depends on how you do it, of course, but it can be close to a wash, or only more expensive by an insignificant amount that won't impact most modern studio movie budgets.
But I do agree that the reason to do it isn't about the cost–it's about what you can do with the process if you have the right collaborative team in place, the right relationships between director, DP, colorist, facility, and studio.
Muro: Things are changing. Boutiques are now emerging that can perform a DI at 2k on a smaller film, and with experience and the right people, you can do that work in just a few days on those projects. That's what we did on Crash at IO.
It's a competitive business, and a few of these companies are just starting out, so filmmakers have to test these boutiques. But there are ways to get into this process with smaller-budget films. You might not be able to get all the bells and whistles of a major budget feature, but you can still get a quality DI in an affordable way.
Millimeter: Beyond cost, how does a production go about choosing a facility to perform a DI? What are the important factors?
Esposito: For me, it's more about relationships and past experience. That is why, right now, I'm skeptical about taking a Fox film any place but Modern VideoFilm (Glendale), mainly because of the pleasurable experience I had on the I, Robot DI, and that was their first big feature DI. I like other facilities, and there are lots of talented people and great gear out there, but we developed a great relationship with Modern.
We did a DI earlier this year at a facility I did not have the same kind of relationship with–I'd rather not say which one–and we ended up scrapping it, going back to the original camera negative, and color timing the movie traditionally, which is what I wanted to do originally. There were a lot of things that went on, but the look of the movie wasn't really helped by it, and we didn't know the colorist very well, and we went there anyway because the director and the DP pushed for it. It turned out to be a mistake. For me, representing the studio and the production, in addition to money, it's about relationships.
Millimeter: With studios seeking lower prices, from the facility perspective, where do you see the DI business heading in terms of revenue streams? Are there new business models you see developing out of your core DI business?
Eicholz: Facilities like ours, and some others, will be trending toward offering different ranges of services for higher and lower budgets. There will likely be three products: low-end, middle of the road like what most movies are doing now, and high-end that might include more sophisticated tools, a more experienced colorist, more color-timing hours and multiple negatives at 4k.
Another trend will be in the area of color management. This is all about making sure that what the audience ends up seeing is what the DP actually shot. The integration of color management all the way from HD dailies through the DI process is something companies like ours will offer.
The other thing is archiving. What do you do with digital files that make up your master? Do you just save the film element, or do you save all the digital files? I think most studios will want to save the digital file, as well. But where? How? This is already part of our business plan, and I presume others will be moving into this area, as well.