• Blogs
• Forums
• Tradeshows
• Partners
• White Papers
• Events

Digital Vision

Aug 1, 2006 12:00 PM, By Michael Goldman

The Workflow and Philosophy Behind Miami Vice Imagery

Web-expanded version
For more from Carroll and Beebe on the Miami Vice workflow, click here.
For audio from Goldman’s conversation with Carroll, click here.

Filming Miami Vice for director Michael Mann would have been difficult even under the best of circum-stances. After all, the crew was charged with capturing imagery that matched a specific vision designed by Mann, using delicate digital camera systems in remote locations across the globe. Furthermore, much of the work involved capturing complicated stunts with cameras and recording decks strapped into speeding cars and boats, as well as extensive visual effects plates. That made Miami Vice “a really hard production,” in the words of DP Dion Beebe, ASC.

As the shoot went on, many wondered how Mann's decision to use Thomson Viper FilmStream cameras to record the entire movie to HDCAM-SR tape through Sony SRW1 VTRs in Rec. 709 HD color space would turn out. Beebe and co-producer Bryan Carroll recently spoke with Millimeter to give their thoughts on that issue, and their workflow.

Both men say the production is a continuation of what Mann has been doing for years — most recently on Collateral in 2004 (click here). The notion of giving the movie a video look, they add, was therefore an integral step in a long-standing plan.

“It's the same basic concept we have been using for roughly six years, but the workflow is constantly evolving,” Carroll says. “The general idea is to start at the end, determine how the material will be displayed — in this case, primarily on film in theaters — and then work backwards by testing to make sure everything will achieve what Michael has envisioned.

“But creatively, when we shot Collateral, we determined early on that we fully wanted to embrace the capabilities of the video camera, which means you commit to video [Rec. 709 HD] color space. Therefore, what Dion and Michael saw on set, on a 24in. HD monitor, was meant to stay the same throughout production, with very little color timing involved. All we really do after that is duplicate tapes, so that when it gets back to us to view in dailies on a 2K projector, it should look exactly the way we saw it on set, only bigger. For Miami Vice, we spent about seven months testing in preproduction, and an important part of that was the development of customized LUTs, which were designed to let us know what we could expect in different situations.”

Among the big departures from Collateral on this project was the amount of contrast in the movie's look. “For this movie, we wanted it more contrasty,” Beebe explains. “This time, we were taking the digital medium into daytime. We did a lot of experimenting, keeping in the back of our mind that we did not want to try and mimic a film look.”

Filmmakers quickly determined that the production would need to be multi-format in nature, with Vipers serving as the dominant acquisition tool.

“We used Thomson Vipers, Sony HDW-F950, and Sony HDW-F900 cameras, as well as 35mm film cameras for certain elements,” says Carroll. “During our extensive testing period, we were not just defining what cameras we would use, but also what software will be in those cameras, what modifications we needed to make to that software, as well as what kind of LUTs we would use in order to visualize on set what the end product would look like. Of the digital material that we shot, about 75 percent was shot with Viper cameras. Maybe 4 percent of the movie was shot on film — mostly for slow-motion, ramping, speed elements, and for some wider visual effects plates. We tested recording to hard drives, but felt that would not be right for what we were doing because the main thing we were testing was to figure out how the set would function within the context of what Michael Mann wanted. And what he wants is for the set to function the same way as when we shoot film. That means grabbing a camera and shooting instantly wherever he wants us to shoot. So, we had the Sony F900 cameras for those particular scenarios, and we recorded all material shot with the Viper and the F950 cameras onto Sony HDCAM-SR tape using SRW1 decks.”

Still, to capture daytime exteriors with these tools, filmmakers developed an extremely stylized approach.

“I'm not talking just daylight — I'm talking about white clouds, blue skies, white airplanes going through clouds, oceans of blue with whitecaps with light boats on the water, and more,” says Carroll. “If you laid these out to most people in the video world, they'd say you are going to clip the highlights. But, in fact, that was a way for achieving Michael's vision. What he and Dion were able to do was deliberately use images, where they do start to clip, in an artful way.”

“Balancing these [Viper] cameras to shoot day exteriors is tricky, though,” Beebe adds. “You have a very low threshold, and you have to ride highlights carefully. If you really blow something out, there is no going back. But that's why you have to define your look in advance, and find exposure latitude you like, and really work within those parameters.”

Filmmakers also substantially improved their method of viewing dailies, according to Carroll.

“A big improvement came with the availability of the NEC iS8-2K DLP projector,” he says. “Wherever we went, Michael had the iS8 and a 12ft. screen with stereo sound in his location screening room to watch HDCAM dailies. Now that the iS8 is available, we can see close to 100 percent of what we shot — so if there were any focus or software issues in camera, or digital hits, Dion could spot them instantly. The editors [William Goldenberg, ACE, and Paul Rubell, ACE] usually turn the show around every 24 hours, and Michael views their latest cut with his 2K projector almost every morning. We brought an Avid Nitris system into the editing room specifically to online the show every night, so that he can watch a 2K HD version of the movie every morning. In that respect, we use Nitris the way we used to use film benches. The editor finishes cutting, the EDL goes up to Nitris, the new version is conformed, and everybody gets to watch it.”

These methods meant that there was no need for major color or contrast changes during the digital intermediate process at Company 3, Santa Monica, Calif., with Mann's colorist, Stefan Sonnenfeld. It was, instead, more of a tweaking process, Beebe explains.

“That's part of the whole debate over DI, now that there are so many digital tools to acquire and manipulate images,” he says. “I think that makes it even more imperative to be clear about your direction, to define your look early in the process. Yes, you can go into the DI suite and take things in other directions, and that's a wonderful tool if you use it right, but the integrity of the original intent for the movie should be honored, and that has to happen at the start of the process, which is why we spent so much time and money testing workflow options. By the time we got to the DI, we knew exactly where we were going with everything.”

---

Carroll and Beebe Talk Back

As they were guiding the movie into its digital intermediate phase at Company 3, Santa Monica, Calif., in collaboration with colorist Stefan Sonnenfeld, Carroll and Beebe took time to sit down with Millimeter to chat about the project. They explained how they used Viper cameras to craft Miami Vice ’s palette and satisfy Mann’s requirements, how they designed and implemented their workflow, and what their philosophies are about the current state, and future, of digital filmmaking.

Millimeter: Please explain your choice of camera systems for the project, and why you made those choices.
Carroll: We used Thomson Vipers, Sony HDW-F950, and Sony HDW-F900 cameras, as well as 35mm film cameras for certain elements. During our extensive testing period, we were not just defining what cameras we would use, but also what software will be in those cameras, what modifications we needed to make to that software, as well as what kind of LUTs we would use in order to visualize on set what the end product would look like. Of the digital material that we shot, about 75 percent was shot with Viper cameras. Maybe four percent of the movie was shot on film, mostly for slow-motion, ramping, speed elements, and for some wider visual effects plates.

We tested recording to hard drives, but felt that would not be right for what we were doing because the main thing we were testing was to figure out how the set would function within the context of what Michael Mann wanted. And what he wants is for the set to function the same way as when we shoot film. That means grabbing a camera and shooting instantly wherever he wants us to shoot. So, we had the Sony F900 cameras for those particular scenarios, and we recorded all material shot with the Viper and the F950 cameras onto Sony HDCAM-SR tape using SRW1 decks.

Millimeter: What was your strategy for capturing daytime exteriors during the production?
Carroll: This was about exploiting what was unique about these cameras, and what they are capable of doing. One thing we learned during this process was to take advantage of enormous depth of field in combination with day exteriors. A lot of cinematographers try to work against the incredible depth of field that these cameras have since it does not resemble what you would expect from film. In our case, though, we emphasized that look and got this fantastic, deep focus effect, which Michael really loved.

And I’m not talking just daylight—I’m talking about white clouds, blue skies, white airplanes going through clouds, oceans of blue with whitecaps with light boats on the water, and more. If you laid these out to most people in the video world, they’d say you are going to clip the highlights. But, in fact, that was a way for achieving Michael’s vision. What he and Dion were able to do was deliberately use images, where they do start to clip, in an artful way. It gives it sort of a three-dimensional look, almost a silver lining to the clouds that was incredible. That’s about knowing your medium, just like a film negative—knowing when it is going to clip and using that. Our daytime aerial photography and the shots on the ocean with boats—that is some of the most stunning photography I have ever seen. Usually, when you see daytime on video, filmmakers and cinematographers have been very cautious. They will come down to a -3db on the camera to make sure the image doesn’t clip. But we were out there just pushing it to the limits in order to get that look.

Beebe: Balancing these [Viper] cameras to shoot day exteriors is tricky, though. You have a very low threshold, and you have to ride highlights carefully. If you really blow something out, there is no going back. But that’s why you have to define your look in advance, and find exposure latitude you like, and really work within those parameters. But it was so exciting to explore and test all of that—exciting and tough. We have a lot of shots looking outside from an interior location through glass windows and walls, and we really had to protect highlights. But you can make it work and come up with something pretty amazing if you really test and explore, and don’t try to treat it like a film negative. It’s all about making the right choices in advance, which I think we did.

Millimeter: During production, what kinds of improvements or changes did you make to your workflow from how you handled things when you made Collateral?
Carroll: One thing we modified on this show occurred as a result of the distant locations [Miami, the Dominican Republic, Uruguay, Paraguay, Argentina, and Brazil]. Once the HDCAM-SR or HDCAM tape came out of the camera, it went directly to editorial for cloning backup camera masters, and to start editing. But, at the same time, they were also performing a thorough quality control process to look for hits or flaws on the tape. That way, we could get information back to the set immediately. We quickly learned during our tests that humidity and temperature in the tropical locations where we were shooting could be a major challenge for these cameras, and that anomalies could crop up. Thomson was great about writing massive amounts of software for us to help compensate for those anomalies, but a stricter process of checking everything shortly after it was shot was critically important.

One weakness of the Viper is the casing—it’s a video camera, after all. We wanted to use it like a film camera and attach lots of boxes and wires and rods to it. Being tethered by large copper cables in unpredictable shooting conditions is cumbersome, so we had intended to use fiber cables as much as possible. However, the heat and humidity limited that in many cases, and we were forced to use copper cables much of the time. Fortunately, a lot of the time, we didn’t need it because we were fairly close to our video village, which we called “The Mother Ship,” and the copper connection worked fine. If we were further away, we would send out the [HDW-F900 camera] when we felt the copper was not sufficient.

Millimeter: What about the dailies process?
Carroll: A big improvement came with the availability of the NEC iS8-2K DLP projector. Wherever we went, Michael had the iS8 and a 12ft. screen with stereo sound in his location screening room to watch HDCAM dailies. Now that the iS8 is available, we can see close to 100 percent of what we shot—so if there were any focus or software issues in camera, or digital hits, Dion could spot them instantly. Whether we shot tape or film, all material was transferred at Laser Pacific to HDCAM-SR or HDCAM, depending on the format of the camera original. Since we shot the Viper in scope mode [2.37:1], it needed to be unsqueezed at Laser Pacific. They would then make a DVCAM downconverted version for editorial to start working on, and an HDCAM dailies tape for Michael and Dion to view.

The editors [William Goldenberg, ACE, and Paul Rubell, ACE] usually turn the show around every 24 hours, and Michael views their latest cut with his 2K projector almost every morning. We brought an Avid Nitris system into the editing room specifically to online the show every night, so that he can watch a 2K HD version of the movie every morning. In that respect, we use Nitris the way we used to use film benches. The editor finishes cutting, the EDL goes up to Nitris, the new version is conformed, and everybody gets to watch it. Once again, we are maintaining the established film grammar and the basic functions of the editing room, except that we took the online of dailies and previews out of other facilities and brought them to our own editing room. That was the only way to turn around a cut for Michael to view on a big screen every 24 hours.

Our Avid Meridian systems for editing the movie have about 4TB of Avid Unity storage, and then we have just under 4TB for the Nitris system, which is more than adequate because that [Nitris] system is basically an online system for updating the piece so that Michael can watch it at 4:2:2, 8-bit resolution each morning.

When we get to the final stages, of course, we’ll conform on the [Quantel] iQ at Company 3. Right now [at press time], we are working in HDCAM mode at 4:2:2, 8-bit resolution, but when we do the final conform, we’ll go back to our 4:4:4, 10-bit HDCAM-SR tapes and conform the final version from them, do our film color grading process at Company 3, and do the filmout at Efilm.

Another thing we did differently was to install customized TrueLight Video Inline Boxes, which we called LUT boxes that contained customized LUTs designed and installed for us by Company 3 and Efilm. They were used for viewing three different modifications to our standard LUT, which gave Michael and Dion a more accurate picture of how a particular shot would look on film if we gave it slightly heavier contrast, or made the blacks blacker, etc. They could see on their monitor an immediate simulation of what would happen when we go to film for exhibition, which automatically adds a bit more contrast anyway. This tool was largely used during our early tests, as well as in the screening room where Michael viewed dailies.

Beebe: That’s part of the process of never forgetting that we were going to end up displaying this movie on film to audiences. We developed a look we were happy with, but we wanted to define and identify in our research what would happen to that look during the transition to the photo-chemical world for exhibition. This particular look-up table was designed to help us bridge that gap by having a reference on hand that emulated our final filmout. By the time we started shooting, we really didn’t need to use the LUT very often, as we had run so many tests through to a filmout that we knew what to expect when we got there.

Millimeter: At press time, entering the DI stage, what does all of this mean for the digital intermediate process, if you committed to the vast majority of your creative decisions long before that point? Is it even accurate to refer to it as a digital intermediate in a case like this?
Carroll: For our purposes, we continue to call it a digital intermediate, largely because, as I said, that is part of the decision we made years ago, when we started down the high-def path, to keep our film grammar intact in terms of how we describe things. From our point of view, a work print is a work print, and it doesn’t matter if it was shot on film or digital or whatever. So, yes, we’re calling the process a digital intermediate.

But now [at press time] that we are into color timing, it is proving to be a little different from what we did on Collateral. On this one, there are very few changes being made in terms of color. There are some small, overall adjustments, but not nearly the kind of color timing that we had on Collateral or on film shows entering the DI world in the past. By deciding to stay in the Rec. 709 color space, we were able to get things to our liking on the front end, on set, for the most part.

Beebe: That’s part of the whole debate over DI, now that there are so many digital tools to acquire and manipulate images. I think that makes it even more imperative to be clear about your direction, to define your look early in the process. Yes, you can go into the DI suite and take things in other directions, and that’s a wonderful tool if you use it right, but the integrity of the original intent for the movie should be honored, and that has to happen at the start of the process, which is why we spent so much time and money testing workflow options. By the time we got to the DI, we knew exactly where we were going with everything.

Millimeter: With your workflow differing substantially from methods used on other high-profile digitally acquired feature films recently, how important is the role of the manufacturers and the rental houses in terms of support during production?
Beebe: It’s crucial. What we’re doing, what David Fincher and others are doing is fairly experimental work right now. [To read about David Fincher’s tapeless workflow on Zodiac, see the August 2006 issue of Digital Content Producer.] There is no one template in place right now for this kind of HD work. There are a lot of new systems in development and new cameras on the market, so there is no chance right now for one completely unified workflow. Therefore, the issue of support is really important. We are always pushing boundaries of these cameras and their abilities, and we are constantly needing to troubleshoot things and have backup equipment in place, and so forth. These companies have been very supportive in that regard, largely because it gives them an opportunity to see their technology in real-world production environments, and help make it more efficient for the next round of filmmakers that will be using it.

Carroll: Laser Pacific, for instance, was like a shield for us, identifying problems as they developed, and helping us get that information out to the companies to get to work on fixes. Plus 8 Digital—same thing. If we needed new equipment, or something to be milled, or more Vipers just to sit on a truck for parts and backup, in case something happened, they were prepared for all that stuff. Thomson and Sony both wrote a multitude of software for us. All that is possible because of the relationships that Michael Mann and I have established over the years. That’s the only way you can get cameras to test for months and months, and then have people ready to fly in on a moment’s notice to write software or bring replacement equipment or whatever else we needed.

But it’s a great testing ground for them, and I would imagine the next generation of Viper cameras will probably reflect much of what they developed with us for Miami Vice .

© 2008 Penton Media, Inc.