Inside The Production Process Of Sinatra At The Palladium In London, Part II
Back to the craziness. Last month, we discussed the early planning stages of getting the London production of Sinatra Live At The Palladium on the stage, up to choosing the right gear. Now, we take you to the studio.
While gear was being assembled, we leapt into production at the studio. Our director, David Leveaux, had been busy putting Rosie O'Donnell in his production of Fiddler on Broadway, but by early November, the creative team convened again, this time at MODE Studios in Seattle for a thorough breaking down and definition of the design. We would all be looking at available footage. We worked for two weeks on defining screen movements and arrangements, matching footage to moments in the show, and discussing the look and feel of the more narrative and scenic content. We had decided that text would be a notable component of the content. We had also decided to use a technique called “2 and 1/2 D” to construct intricate moving composites of photographic elements. This technique was pioneered early on by Disney and is also called multi-plane animation. In an extension of the rotoscopic nature of Frank, we were going to cut hundreds of people, elements, scenic locations, and architecture out of the mass of Sinatra's life and assemble them into virtual pastiches that Frank's footage could move among and through, separate and related. In this way, we hoped to navigate some of the more challenging moments of communication, like the story of his difficult birth, or the telling of his tragic/romantic relationship with Ava Gardner in the number, “I Got It Bad, And That Ain't Good.” We came out of the two-week meeting with a more challenging design mandate than ever and with little time to actualize it. Leveaux and scenic designer Tom Pye flew to London to begin preparations there, while MODE studios went into overdrive creating two hours of beyond HD content for nine surfaces.
Divide And Conquer
Divide and conquer became the mandate. Our lead compositor, Larry Butcher, began the construction of an intricate multi-plane composition featuring a travel through Hoboken tenements with an aerial pass by the isle of Manhattan circa 1915. Lead editor Brandon Oosterhof began some of the more delicate rotoscopy work, as well as syncing audio and video. Colleen and I began construction of various numbers, making the broad stroke “sketches” of numbers that could be refined at rehearsal. Late in December, we ramped up again, adding three interns to the workload from our class at CalArts: Roz Fulton, Rebecca Makkus, and Jason Baeri.
After five weeks, it was time to decamp to rehearsals. We had been continuing the process of collaboration and idea approval via a client website and by iChat and Skype video conferencing. This had appeal to the inner technological fetishist, but ultimately, the distance proved difficult. We discovered that the team worked best in the same room. What team wouldn't? But now, we would at last be convening in London for rehearsals.
Rehearsals were held at the feature film soundstages of 3 Mills Studios outside London. Leveaux wanted to work with the company and a reasonable facsimile of the visual production to build and define the show. We weren't going to have the full control system of eight HD Hippos for this period. In fact, we would only be able to get one initially. Green Hippo (personified by Nigel Sadler) rode to the rescue, reconfiguring the software of one HD Hippo to simulate being eight different Hippos for programming purposes. Sean Cagney, our super Jedi programmer from Scharff Weisberg, would also be simulating automation in his programming.
Since our content would be moving automatically with screens, we weren't going to have to work with positioning values in the actual production. But for the purposes of rehearsal, we used four Barco R12+ units on a big muslin sheet. Our super Hippo would simulate nine surfaces on this one plane. We would, therefore, program automation moves on the MA Lighting grandMA in order to visually simulate that happening.
Cagney and I could see immediately that we were going to be working some long nights. Clearly accomplishing this volume of programming in real-time rehearsal was going to be untenable. This show construction phase would last a little more than six weeks before we loaded into the Palladium in London's West End.
One consequence of creating a show for the first time (or even the second) is that there is no road map. The rehearsals at 3 Mills Studios revealed a lot about what we had planned and a lot about what we had not planned. The show grew, we explored, and things changed fast. I had arrived in London with Butcher. Shortly after, Colleen arrived with Jo Johnson (our first assistant), Oosterhof, and Laura Kozuh, who had joined the MODE crew as a compositor/animator. But the workload had quickly swamped even this group. Colleen used contacts to get in touch with VMI, a London effects and film personnel resource. VMI was able to locate two more fantastic compositors, Andrew White and Shane Costar, as well as provide additional computing resources to our own.
Our little mobile studio now had nine Macintosh computers, all accessing in excess of 12 terabytes of RAID storage, chock full of Sinatra media. The area we occupied might have sounded like a wind farm for all our fans. We were not lacking for heat, either.
The pace was unremitting. There were no days off, and most of the crew fueled themselves with caffeine or can after can of Red Bull. Content would be directly built in Adobe® After Effects®, rendered to the RAID, dropped into watch folders, encoded by ProCoder on Boxx Render Nodes, and then moved directly into the Hippotizer file systems, allowing Cagney to program and implement updates quickly.
In the theatre, the pace was also unrelenting. Scharff Weisberg project manager John Ackerman was supervising the XL Video crew in the install, working out exact projector positions and making sure signal paths were distributed and secure. He was working closely with Richard Bullimore, who was the production supervisor for the show. I was visiting the theatre as much as possible to try to keep an eye out for the unforeseen and to make sure the design would work out physically.
With one week left of rehearsals outside the theatre, the entire projection team moved to the Palladium to get ready for the arrival of the company seven days later. The process of refinement, construction, and selection continued, and the spreadsheet containing our notes yet to be done was pages and pages. I had been involved with huge productions before, and I could feel that familiar weight of, “Oh God, can we really do this?” settling in.
The most critical technical issue on the agenda that week was to determine if the hard work put in by the Green Hippo team and the talented brains from Stage Technologies had resulted in good screen tracking. This had been a killer problem at Radio City, and it was something I seemed to discuss with the producers daily. Was it going to work? Cagney was confident. “I'll have the thing doing figure eights by nine o'clock,” he brazenly proclaimed at the beginning of the day.
We had done a proof of concept demo at the scenic automation shops while in rehearsals. The tracking did work, but the sampling rate that the media servers were getting from the automation computers was down around four times a second. This resulted in some herky jerky movement. Also, the acceleration and deceleration was hard to track. The automation team decided to rewrite code to provide location feedback at least 24 times per second, increasing the tracking resolution. In turn, the Hippo team decided to tweak code to handle the accel/decel problem, incorporating a look-ahead algorithm to help the Hippo anticipate and assimilate these moves in image tracking.
Now that we were in the Palladium, the results of these efforts would be known. The crew at Stage Technologies had actually achieved a feedback rate of 100 times per second, which was phenomenal. And the look-ahead algorithm worked perfectly. Once the Hippotizers had been dialed in to the physical size of the projection raster, they tracked the moving screens smoothly and flawlessly. I promptly fished my phone out and started dialing; a collective sigh was heard on both sides of the Atlantic.
One thing not going so smoothly was the rotoscoping. Imagineering reported that the schedule was falling behind. We were going to have to make a subjective decision about what level of quality was good enough in order to get the footage done in time. Leveaux was not happy. The rotoscoped footage was the center of the show. It was impossible to contemplate it not being done in time. Nevertheless, we faced hard reality. After considerable difficult discussion, we decided that the contractors working on the Mokey footage would complete it to an 80% level of quality. Once completed to this level, Oosterhof and compositor Shane Costar would refine the masking work in Motor and Shake at the Palladium. We hoped that, by using many people getting us to that level and then handling refinement ourselves, we could still lock the show prior to press night.
Weeks slipped by frighteningly fast. The company joined us at the Palladium, and the arduous process of tech began. The integration of scenery, image, lighting, and staging proved difficult. The movement and blocking within the media was completely tied to and interdependent with the physical blocking and staging. Changes on either side meant considerable adjustment by the other. This manifested itself twofold, with Cagney spending a lot of time pounding out programming changes and the content crew contributing newly finessed pieces.
We were also attempting, in some cases, to synchronize the sound of performances that were different from the video performances. This was proving to be mind bogglingly difficult and was adding to our frustration levels. The opening song, “Come Fly With Me,” was an example of one of these numbers, and we were all stymied. At the suggestion of Charles Pignone, the estate's expert on available Sinatra materials, it was decided to use a different recording of the opening. The video and the audio sources now were from the same performance. Unfortunately for Stephen Mears, the choreographer, this version's tempo was just about twice as fast. The dancers would be wasting no time in breaking a sweat.
The Final Curtain
Speaking of time, it was up. When the first preview was 48 hours away, Leveaux declared that we had to lock what we had for the next several days in order to be stable enough to run for an audience. We hadn't had an opportunity to run it for ourselves yet. The first run without pause would happen for an audience. Good God.
The projection team immediately shifted into overdrive. We had huge pieces of show constructed, but the all-important transitions from sequence to sequence were non-existent. With Colleen leading the content team in cranking out interstitial pieces, Cagney and I engaged in a 48-hour straight programming session. You haven't lived until you've chased mice around in the Palladium lobby at 4:30am!
Against all odds, at 7:30pm on February 17, Sinatra Live At The Palladium went up in front of its first preview audience. It played without stopping and featured some rudiments of transitions between all numbers. For the first time, we could all see that we had something — something that was going to work. After the show, Leveaux gathered everyone and thanked them sincerely for their efforts. Now we had to enter an even more difficult phase: perfecting the show while running it.
With nine Hippotizers, the show represented a staggering amount of DMX channels. The programming was intensely complex and multilevel. With every change and update, we had to roll through many preceding and following scenes to make sure we weren't introducing unforeseen tracking issues. Balancing how much we could change in a day of rehearsal and how much time we would require to restabilize the programming became a daily conundrum. Leveaux, Pye, and associate director Eli Gonda were anxious to cover as much ground as possible. We had to try to accommodate this while keeping a conservative eye on the programming. It proved to be extraordinarily challenging.
With the final days counting down, Oosterhof and Costar did a tremendous job of refining the rotoscopy. By three days before opening, the show was looking sharp and really beautiful. Andy Bridge's lighting was tasteful and elegant, dancing around images and performers. The 24-piece onstage big band is probably the hottest assembly of players in the Western World right now. And out in front of a hot, hard working dance core was Frank Sinatra, larger than life and feeling decidedly alive. Leveaux pronounced it locked for opening. It was a moment I had begun to feel would never happen.
We knew it was really over when the Starbucks across the street from the stage door shut down two days before opening. The British crew joked that only a bunch of freaks from Seattle could pull that off.
It would have been easy to write another 5,000 words about this show. Every day seemed to be epic in its challenges and its achievements. If only the high stakes could have been captured in a reality show, we'd all really be making some money! We fought, we worked, we loved, and we made some very pretty pictures. Now, and through October, you can enjoy those pictures at the Palladium in London, on any day but Sunday. Go figure.