Trevor Hogg chats with senior visual effects supervisor Dan Glass; visual effects supervisors Stéphane Ceretti, Matt Dessero, Geoffrey Hancock, Alessandro Cioffi, Florian Gellinger, Angela Barson, Clark Parkhurst, Russell Earl and Falk Gärtner; and executive visual effects producer Ismat Zaidi about their work on Cloud Atlas. Be warned there are spoilers...
“In a strange way our work relationship isn’t like it ended and began again; we continuously collaborate and have done for many years now,” says Dan Glass of his long-time creative partnership with filmmakers Andy and Lana Wachowski which dates back to the sequels
The Matrix Reloaded (2003) and
The Matrix Revolutions (2003). “The only break in it was
Batman Begins [2005]. While I was doing
The Tree of Life [2011] I also did
Speed Racer [2008] with them and
Ninja Assassin [2009].
Cloud Atlas [2012] came up after
Ninja which was directed by James McTeigue. Andy and Lana were given the novel by Natalie Portman [
Black Swan] back on
V for Vendetta [2005] and had loved it and always wanted to turn it into a film so they put it together. It took them quite a long time because of the complexity of it but they wrote one of the most brilliant scripts I’ve read.” Penned by author David Mitchell, the 2004 novel explores how the lives of others are interconnected through six stories spanning different time periods.
The Pacific Journal of Adam Ewing chronicles the 1855 passage of an American notary to San Francisco,
Letters from Zedelghem recounts the 1930s tale of a poor amanuensis who has a fatal relationship with an English composer,
Half-Lives: The First Luisa Rey Mystery features a 1970s journalist discovering that a newly constructed nuclear plant is unsafe,
The Ghastly Ordeal of Timothy Cavendish stars a present day book publisher who attempts to escape a nursing home, the futuristic
An Orison of Sonmi –451 revolves around a clone who sparks a revolution, and the final story is the post-apocalyptic
Sloosha's Crossin' an' Ev'rythin' After where a primitive man encounters a woman from technologically advanced society.”
Alterations had to be made when adapting the source material for the big screen. “The novel sticks to a chapter per story and goes through the time periods and comes back out through them for a second chapter on each so it’s far less intercut,” states Dan Glass. “That works for the medium but obviously it ends up posing different kind of challenges and the way the themes come across is slightly different. The feature of intercutting I think makes total sense not just for the medium but the way you can contrast some of the roles and the characters and the situations they get in is fantastic. But of course that means the way you can break things up is even more single scenes in different locations than you might otherwise have had. They’re in one place then you cut away to another of the stories and you come back and they’re somewhere else. The continuity is less direct. There is obviously keeping a mood or a sense of the different worlds when you arrive back at one from another. The novel, even though the chapters are distinct, they’re written in distinct different styles so the shipping tale is written in an Old English, the 1970s story is written an Chandleresque film noir, the future story is written in an odd future English/Korean, and the post-apocalyptic one is written in an almost Ridley Walker style of some made up English. Those are distinct and it’s quite hard reading as a result. With the movie they consciously wanted to make it clear it was one film. Although you’re jumping in different time periods they didn’t want it to be jarring so the colour pallets were for example were designed not to get too distinct. They wanted to help the viewer to understand when you’re going from place to another.”
Having three filmmakers helming the cinematic adaptation, Tom Tykwer (
Perfume) as well as the Wachowskis, did not over complicate matters. “They definitely worked differently but they were collaborative and agreeable,” says Dan Glass. “Tom would be ready to say, ‘Look this is more your area of expertise. You make the call here.’ And vice versa. The bigger challenge was making sure that we got appropriate feedback quickly from all parties without it affecting the process. We had to have a quick system to getting data delivered all over the world simultaneously and we had a great team, Marc Kolbe helped lead in order to get all of that in place. It went smoothly considering.” Production stalled as various studios declined to finance the project. “It was too big a risk and interestingly the studios thought they couldn’t do it for that money. They thought it would be much more expensive. The unfortunate thing even with $100 million a lot of that money ends up going, when you’re independent, into insurance and other costs that don’t really show up on the screen so the effective budget was actually quite a bit less than that.”
“The financing which did turn out to be very complicated but even before that was how to schedule and manage what was like six different movies, would be it smaller ones within a larger one,” states Dan Glass who served as the Senior Visual Effects Supervisor on
Cloud Atlas. “How you plan and prepare for all of those different time periods and variations? They had the idea early on they wanted the same actors to play the different roles and they would have two main units running in tandem because Tom Tykwer would take one and the Wachowskis would take another. The scheduling complexity alone, the most of which fell to these directors to figure out was enormously intricate and very little if any room for error. After they figured the initial schedule Lana, Andy and Tom reckon they could slip one day in total on one of the units or the schedule would break.” Even though Tykwer and the Wachowskis divided the shooting of the different stories between each other, the three filmmakers were creatively entwined. “It’s a fairly blurred line. They wrote and planned it together. The first meeting we all got called to was about two and a half years ago and showed up in Berlin for two weeks, myself, the production designers, the directors, the first ADs, and the DOPs. We all looked at the script, broke it down, worked out how we could parcel things up, strategize, plan and budget; that was a kick-off but the directors were very much involved, my meetings were with all three of them. It took about a year and a bit before they could raise all of the financing; once that was raised all production meetings were held with all three directors. The previs was done with all of their input. It was only when we got to the physical shooting that by necessity they had to separate in order to run everything in parallel and they divided it three stories per team.”
“I don’t know how much they had to deliberate about which stories that they wanted to do because they loved all of them,” notes Dan Glass. “One of the shorter answers for it is that the heavier visual effects material fell to the Wachowskis which made sense because they obviously have a ton of experience in that area and Tom did stories that were more actor-driven.” The native of Britain explains, “Tom took Luisa Rey which is the 1970s, Timothy Cavendish which is the present day, and the Frobisher story which is the 1930s. The Wachowskis took the 1850s shipping story, and the two futuristic ones being Sonmi and Sloosha’s Crossin’.” The cast which plays multiple roles in the $100 million production includes Tom Hanks (
Forrest Gump), Halle Berry (
Monster’s Ball), Jim Broadbent (
Iris), Susan Saradon (
Dead Man Walking) Hugo Weaving (
Captain America: The First Avenger), Jim Sturgess (
Upside Down), Doona Rae (
The Host), Ben Whishaw (
Skyfall), Keith David (
Platoon), James D’Arcy (
Master and Commander: The Far Side of the World), Xun Zhou (
Suzhou River), David Gyasi (
The Dark Knight Rises), and Hugh Grant (
About a Boy). “With Sturgess,” recalls Glass, “[Hae-Joo] Chang the character he plays was one of the toughest we had to do because it involved at some points full CG eyelids and eyes.”
“Most of the departments on the movie were split in two,” states Dan Glass. “The ADs were split, there were two DPs, there were two Art Departments but three departments sound, editorial and visual effects were all one single department. Alex Vanner was the editor; he put it together. Sound similarly they needed to run it singularly. Visual effects essentially ran as a single department and I oversaw. I brought in Stef Ceretti who I have known for many years and he works out of our Method London office; I brought him in as a colleague because firstly we knew we were covering two units shooting. One of them needed coverage so Stef came on and looked after Tom’s stories principally, that work and we continued onto post more or less. It ended up also helping to have Stef to look after material that was based in Europe because he is Europe based. I looked after stuff that was for the most West Coast between Method Vancouver and Los Angeles and ILM; although I also saw the Scanline work through one of us in London. The brunt of the work I took was based on the fact that we would split stories the way the directors had.” Glass explains the work process for the visual effects shots, “When they got to a certain point Stef would show them to me and we would talk; he would take it back with more feedback or we would show the directors. Pretty much vice versa. I would always like to get his opinion we did work as a team but it did come under one roof in a sense.”
“I had done
The Matrix Reloaded [2003] and
Revolutions [2003] with Dan Glass, and Lana and Andy Wachowski,” states Stéphane Ceretti. “I have been working with Dan at Method Studios and it seemed natural for me to join them to help on the film as a VFX supervisor.” There was not much time to become familiar the source material. “I heard Lana, Andy and Tom wanted to adapt it for the screen but also heard it would be quite a challenge. I got the book and started reading it but got the script very little time after so I jumped on the script.” Though Ceretti mainly focused on assisting Tom Tykwer throughout the production and post-production, he also did some work on the stories being handled by the Wachowskis. “Dan was focused on the futuristic parts of the story, but he retained oversight on everything in the movie as he was the senior supervisor. Tom has a very photorealistic approach to what he wants to achieve; he wants to shoot as much as possible and has a great understanding of what looks good visually.
Cloud Atlas was a very big project in terms of visual effects for Tom, but he handled it very well and had a very sharp eye.”
Vendors based in the United States, Canada, UK, and Germany were recruited to produce the required 1150 visual effects shots; they included Method Studio facilities in Los Angeles, Vancouver and London Trixter, RISE Visual Effects Studios, Scanline VFX, BlueBolt, Lola VFX, Industrial Light & Magic, and Exozet Effects. It was a combination of strengths and individuals I have worked with before who I have trust in,” states Dan Glass as to how he chose the visual effects companies. “It was the sort of situation where time was going to be tight in terms of how quickly they wanted to put a temp version of the movie together. We had to spread ourselves across a number of places for that reason. But it was also the nature of what we were trying to do again which was so different and we wanted or were looking for a lot of places that would be part of and collaborate with the design itself. For that reason I reached out to a lot of companies famous for doing that or had the direct experience with that case. Scanline is an example obviously. We went to them originally for water but we sent them some other things that were quite complex and design oriented like the slaughter house; they did a fantastic job at. I was excited to the level of collaboration they brought to it.” There were shared shots between the VFX vendors. “It certainly was a challenge because in post-production there was one main edit. The Wachowskis went back to Chicago and Tom stayed in Berlin. We had to deal with different continents showing material and getting creative comments back on things. The way we split things was not always by story which was a logical one but by type of content. Particular environments or views would go to one vendor so they would understand all the continuity issues related. The characters because there was quite a lot of work as you might imagine sometimes fairly simple to help prosthetics to more involved to help take the prosthetics somewhere it couldn’t achieve practically.”
“Dan Glass is our Executive Vice President of Method and Senior Visual Effects Supervisor,” states Method Studios VFX Supervisor Matt Dessero who is based in Los Angeles. “He was on the project with the Wachowskis and Tom Tykwer for the last couple of years and we’ve been talking about it and looking at different pieces of what we could do here at Method L.A. and the other Method facilities.” The contributions of Method Studios were divided between three facilities. “Method Los Angeles got the biggest bulk as we did about 200 shots on the film. The Vancouver office did about 50 to 60 and London did about the same 50 to 60. We shared assets between all of the facilities; we both modelled the vehicles and a lot of buildings here in Los Angeles. The Vancouver office also did a lot of buildings and we shared our assets with ILM and they shared back with us.” Method LA handled the clone revolt tale. “We mainly dealt with Sonmi, the fifth story which takes place in 2144. Dan Glass was the supervisor for the sequence.” Environments needed to be created and the actors digitally enhanced. “We were building big establishing shots of that we did about 90 VFX shots. We also did Chang, the hero there.” Though the prosthetic make-up got better as the production progressed, some CG augmentation was required. “You’re trying to take Jim Sturgess who is a Caucasian and turn him Asian; there are some challenges there. What they had to do was put practical make-up on him; it wouldn’t be consistent so our job was to go in there. We started off with 40 shots and ended up doing 100 shots of face clean-up on him.”
The ethnic transformation of Jim Sturgess could not be solved by the push of a button. “How do you make a guy with make-up on him look Asian but still retain his soul, character and look?” asks Matt Dessero. “Dan and I went through tons of Asian men reference, and found different eyes and bits that we liked. We did some concept frames. It was easy to make him look Asian but it didn’t have the essence of Jim. As we got the running footage we started doing testsfor the clean-up and had a couple of guys working on it in 2D. What we noticed early on was there was no way that this was going to be consistent. Each artist has his own interpretation of what this guy should look like.” A digital copy of Jim Sturgess was needed. “We opted to take the Cyberscan of Jim Sturgess with the Chang make-up on and we built a model from that which became our tracking model. We cleaned up all of the make-up inaccuracies, and made sure the eye sockets locked into Jim Sturgess’ face and pressed up against a couple of line-up frames. From there, we would bring it to Nuke. Take the plate and project it onto that model and unwrap it in UV space. In UV space everything is flat and you can pin on the image for the most part without it swimming all over. You remove the camera move from that. We did a lot of our clean-up there and were able to setup some nice templates on what this guy should look like. While the Nuke guys are working on that, those initial tests we would wrap them backup and go, ‘It will be nice to have some shadows or some inclusion or some extra skin speck.’ We realized that the character was getting soft. While our compositing leads, Troy Moore and Jeff Allen [Compositor Supervisor], continued to improve the process in 2D, we created a hybrid which took the original model that we had of Chang and deepened the sockets a little bit to maintain the Jim Sturgess look and adjusted the eyebrows ever so slightly. What we then do is take that into Nuke and project our unwrapped head back onto the hybrid model. That hybrid model is where we would also render skin textures, and made CG eyeballs and spec maps. Blake Sweeney was one of the CG Supervisors who put together our CG Chang and that became the base of what the compositors would use to add specular reflections back on to the heads. It gave us the floor detail and the wrinkles back into the face. It’s a complex process.”
“We a few environments to do,” remarks Matt Dessero. “We had the prison break and then from the prison break we went to the prison exterior. As Chang is stopped by the enforcers we cut to another story in the movie. When we come back the enforcers and gunships are leading Chang down into the deeper part of the city and they land on a platform on top of the substratum building. The substratum was the name of a set location. We did a LiDAR [Light Detection And Ranging] of the set which was a box where the fight takes place. We built the building and all of the surroundings around it. There were multiple views we had to explore. We had our in-house concepts and matte painter Chris Sanchez worked us some simple sketches. We started with simple sketches of what we wanted it to look like. A lot of our reference was cities in Asia, Kowloon was one of them; it’s a crowded city. Buildings are stacked on top of one another and that was our base image reference for a lot of these portions of the city. From there rough texture would be created. We would give it to modelling and modelling would workout a simple model to understand the form and scale. I would take that and go back to our concept painter and he would paint on top of those images so he was working in proper perspective. We would add refined detail where needed. At the same time the modellers were adding little pieces of pipes and tubes that we were going to need. We worked tightly with the modellers with their modelling on this sequence with Lim Bunupuradah. There was a lot of back and forth between concept modelling, and we improved the image rather quickly. This is the pro of having a concept painter do concept for a couple of weeks and giving that to our modeller. We wanted to create this world where both guys were sharing, and going back and forth. It worked out well for us.”
“The environments are pretty much all CG with the exception of that substratum fight where we had one tenth of the world there and we created 90 per cent of it all around,” states Matt Dessero. "There are a couple of shots where there is a gunship and the pilot has his graphics in front of him and is looking down at Chang. It is right before the gunship gets blown up. We shot a set piece with all of the actors and the actions. The camera was pulled back but not far enough and they wanted to go further. We adjusted the camera so we could come slightly off centre. It looks great. We shot texture and the actors. We cut the actors out and put them on a card. We ended up replacing the majority of that set with CG including the prison truck and all of the surrounds and the ground. It makes it easier because we can reflect everything back into the pools of water on the ground.” Dessero explains, “They weren’t happy with the buildings so we ended up spending a lot of time reconcepting that and finally building in CG. The prison truck we did shoot some on-set. When we replaced those it was easier to replace with our CG rather than try to force it. They weren’t a hundred per cent happy with a couple of the camera moves. A good example of that is when the prison truck leaves the gate and is coming out onto a transway. In that shot we had a plate for it but the camera couldn’t get low enough to go under the truck The Wachowskis wanted the truck come right over top so we ended up replacing the prison truck with a CG truck.”
“We did the orisons, the spinning disk and what that does is generate other little viewing disks for them to web search,” states Matt Dessero. “Chang takes Sonmi [Doona Rae] to the safe house which transforms. The idea was that the LCD panel can show whatever we wanted it to show. You walk into a cement room and all the walls start moving and sliding; they create this cherry blossom scene he has programmed in, and with whip of his hand and the orison, the petals start to float.” A orison causes a wall in the tiny room to become a movie screen. “The tough challenge there was all of the footage that had to go into the disk. One of our CG supervisors Doug Bloom created a nice model with a translucent back with a reflective glass back. When you’re looking from the reverse side you can’t see what the person is looking at. When you’re looking from the front you get these sharp crisp images. We modelled up some simple disks, creating a texture map, refracted the back with a glass shader, and went back and forth with production and Method Design; they created a lot of the graphics for us.” The in-house design group assisted with the signage that is everywhere in Neo Seoul. “We had a library of signs. Production provided a chunk of signs to us and we created some of ourselves with Method Design and our visual effects team. Vancouver created a couple of signs also. We would share assets and put them in. The challenge was not having them repeat. There were a fair amount of signs but you would see the same ones pop up so we said, ‘We’re going to use this group and you’re going to use that group.’”
“Everyone was a unique challenge,” states Matt Dessero. “There is one shot in particular; it is an over the shoulder of a character and he is looking out at an enforcer on top of one of the motorcycle skiffs. So many people ask, ‘What’s the CG? Just the background?’ Its like, ‘Yes the background but it also includes the guy on the skiff.’ They shot that on a gimbal out there. Dan and I opted not to shoot our sequence on the gimbal because we were going to replace it with CG. We didn’t want to be a slave to it and knew we could pull it off. The character is in beautiful costuming, and has face shield and we’re reflecting everything into it and it integrates so well.” Another cinematic moment stands out to Dessero. “The Substratum Sequence is beautiful to me. There is so much detail and richness. There are little floating banners and moving lanterns. The Wachowskis and Dan Glass pointed that out to us and said, ‘Nice detail in there.’ The effort paid off. “I remember our own team was like, ‘Why do you want to do that?’ I want to do it because it is not going to take us that much time to do and it is going to add a little bit of richness and when the directors noted it was like, ‘That’s why we do it.’”
“I couldn’t have done it without the crew on any of the shows,” says Matt Dessero who was also serving as the visual effects supervisor for
Argo (2012). “They were heavy design, compositing, and lighting. The time frame we have these days are minimal. The
Cloud Atlas work was under seven months to do 200 shots, a 100 face replacements and a bunch of environment shots. It’s a testament to the crew’s tenacity.” Dessero remarks, “The biggest challenge was the amount of design we had to do on production in a film. We were supplied enough concept to understand the world the Wachowskis and Tom were trying to build but not enough to resolve our environments and sequences that we had to work on. We spent a lot of time designing. You’re designing futuristic bits of the world, robots, a scan box that floats around, and the prison yard. There is so much design that went into this show which is awesome to have an opportunity to work on such a great film and have the leeway to design. It’s rare. Usually, it’s all supplied. It was great. We would sit here with our concept painter Chris Sanchez; he was helping me concept things out and we would send them off to Dan Glass and the Wachowskis and they’re like, ‘That looks cool. Do this and this.’ It was a great experience.”
“We had early talks with Dan Glass to discuss what sequences would best suit our previous experiences virtual locations and digital extensions,” states Method Studios Visual Effects Supervisor Geoffrey Hancock who operates from the company facility situated in Vancouver. “Dan was excited to find a synergistic way in which the various Method Studios locations could cooperate to take best advantage of each of our specialties.” The project required a trip to Europe. “It was excellent to have some real face time on-set while we took part in the shooting of our sequence in Berlin. We knew that this production would be a real test of the merits and challenges of a global production. Stéphane and Dan had work out a smooth set of technical specs and submissions guidelines early on which allowed us to fit in with their workflow of reviews. Occasionally during production we would have a visit from Dan and spend as much time as we could in our theatre reviewing material and turning around revisions as fast as possible within the day. For the majority of the time we had regular CineSync sessions and would receive annotated still frames with notes. Dan and I also swapped photo references early on to help zero in on the specific not-too-pretty dawn look that the Wachowskis were after.”
“We worked on the Sonmi sequences inside and outside an apartment that is being used as a safe house,” states Geoffrey Hancock. “Once their pursuers get close Sonmi and Chang escape out a window onto a narrow extendable scaffold between buildings. The majority of our work was creating the virtual environment around the safe house. It was an older, run down part of Neo Seoul that bordered on the old partly submerged Old Seoul. This environment was seen both at night time, lit up by all the pervasive advertising; as well as at dawn with the sun rising over the ocean beyond the city.” It was important to maintain an Asian sensibility to the imagery. “We did a ton of research on all the Korean traditional architectural forms as a way of imparting an Asian feeling to the futuristic neighbourhood. We took special notice of the layering and improvisational nature of structures in big Asian cities. We wanted to show and area of town that had been grand at one time, but was old, with many repairs, additions, and retro fits.”
“One of the themes of the film is the layering of time, and we were trying to bring that into the city,” explains Hancock. “Given the immense height of the structures, we had a lot of room to play up this layering. The deeper you look in the city the older the structures become, one being built on top of the other. There was a network of walk ways and levitating road ways that snaked between all these buildings at various levels that added complexity and life to the city. These thoroughfares where populated with digital extras and traffic. The major difficulty of this sequence became the fact that no one shot could be finaled before the rest in a sense. Just when we thought we had got a particular angle looking good, further refinement, elaboration and creative grading in another would push it out of continuity. The surrounds of the safe house were conceived as 6 interconnected matte paintings with mid ground CG buildings and two ‘hero’ CG buildings on either side of the gap. The continual mantra around the production floor in the final days became ‘oh wait, that little piece over there, that’s out of date.’ This spoke to the nonlinear methods that where required on this fast paced production.”
Initially, we saw some areas that would overlap between studios in the awarded sequences,” says Geoffrey Hancock. “We knew that the Gunship that attacks in the end of our sequence appear in other sequences. Method Studios LA had many featured uses of that asset and we agreed that they would share the model, rig and look development for it. This meant that Vancouver would need to adopt Vray as a rendering platform for that asset. In the past we have most often used Mental Ray for our environment work, and Mantra for the digital extras. Due to the complexity of the city environment, it’s inter-reflections and dynamic lighting requirements. The renders that were going to be the starting material for the matte paintings began to be unusually cumbersome, and since we had Vray integrated into our pipeline for the Gunship it wasn’t long before the lighters where championing to use it for the city. Eventually the majority of the renders for the environment where completed in Vray. The matte painters still used Mental Ray for specialty renders and the crowds and vehicles stayed in Mantra. We also knew Method Studios London had a lot of matte painting of Neo Seoul in their sequences. As soon as we had rough CG buildings for use in the early blocking stages we shared those with the other studios working on shots that could benefit from them. We also got a good batch of assets from LA for numerous rooftop details, vents, fire escapes,and things that we used to clutter up the city and help it feel lived in. Near the end of the production Dan and London saw greater overlap between shots surrounding the gate to the Old City, so we were able to flip a camera angle around, and do a render of our environment from the vantage point that London had in there shot. We also sent over the entire city scene which I believe they cherry picked assets out of. In turn we used a portion of a background that London created for LA as our distant horizon and reflection environment. It was a good chance to see how each studio was working and take advantage of similar types of work being done.
Besides the interconnectedness of our shots within the virtual location, the other big challenge was finding the right emotional mood and grading for the scene,” recalls Geoffrey Hancock. “Given the huge breadth of material in the film, the numerous time periods and locations there wasn’t a single simple look and feel that we could say our scene needed to match. The night time shots were in some ways homage to other classic science fiction visions, but with a real personal twist from the Directors. There was a clear desire to avoid pinks and purples in the multi-coloured hues of the signs, advertisements, holograms, neon and architectural lighting that gave form to the darkness. This resulted in an eerie cool not altogether inviting feeling. The remainder of our work took place at dawn. There was a real challenge to find the specific emotion and style for what at first seems like a classic lighting situation. We began with photographic studies taken at dawn, and extensive research for reference of large polluted cities near water at dawn. In the end Olivier Dumont was instrumental in his capacity as an art director for our sequences; his explorations eventually lead to a decided not warm and fuzzy morning. We really pushed the plate photography of the actors on the green screen stage and grading our matte paintings in compositing to find gritty combinations of warm brown tones in the bright areas and near green in the shadows. This really brought forth the feeling of the pollution in the air and the age of the surfaces.”
“Other than the two main actors on a short section of practical scaffold, and a bathroom window led, we replaced and created from scratch the entire environment outside the safe house,” says Geoffrey Hancock. “There were numerous city layout scripts both in Houdini and Maya. Our Houdini crowd system was extended to include traffic ways. Vray was integrated through our pipeline. And we continued with the integration of Shotgun and Tank into the studio.” Hancock remarks, “I was really proud of the quality of the work in the dawn sequence. At each turn when a creative decision was being made we asked ourselves, ‘How this could support and strengthen the emotions that our characters are feeling here?’ In subtle supportive ways, we increased the sense of danger, closeness, separation, and fear that intermingles in the scene.” A particular software program turned out to be an indispensable tool. “Vray proved itself to be a superb renderer. Dan Glass and his global team run a tight ship, allowing for great efficiency. The processes and workflows set in place to work with so many other vendors kept up the pace better than many productions. They were a pleasure to work with. It’s always a pleasure to work on films with great stories and
Cloud Atlas was an inspiring story to be part of.”
“
Cloud Atlas is a project we strongly wanted to be a part of, since we heard the first rumors about its production,” states Trixter Visual Effects Supervisor Alessandro Cioffi. “We sensed that it could become a sort of milestone in recent cinema history. I've personally read the book and loved it. Simone Kraus [CEO and Co-founder] started gathering information about it. A few months later I went to Berlin to meet Dan Glass and Marc Kolbe on-set and offered them our collaboration, and finally, we got the great opportunity to contribute to its completion.” German based VFX facility dealt with Visual Effects Supervisor Stéphane Ceretti. “Right after the first preliminary talks and briefings with Dan, we mainly had contacts with Stef, with whom we had almost daily meetings via CineSync, Skype or telephone. We have worked with them in the past. By now, we have quite a good understanding of their visual wishes and method of work, which helped us to promptly interpret the directions they wanted to go.”
“We worked on a few sequences,” states Alessandro Cioffi. “Worth mentioning are those where Meronym [Halle Berry] and Zachry [Tom Hanks] climb up a mountain in order to scout out an abandoned observatory, or where Cavendish [Jim Broadbent] meets Hoggins [Hanks] at the author's award party and Hoggins throws the critic Felix Finch [Alistair Petrie] off the building. Plus we worked on other shorter sequences and sparse shots and effects here and there.” The senior visual effects supervisor and the visual effects supervisor for
Cloud Atlas were great collaborators. “I must say the biggest credit for the visual research on those sequences naturally goes to Dan and Stef who went on location and shot tons of visual material both for reference and for actual use in the VFX-making process. Our part was to consolidate that considerable volume of footage and create consistent, believable images. For the Meronym-Zachry Sequence, the actors were shot on big green screens in Berlin. Whereas the location of choice for the background was in Palma de Mallorca, a popular climbing rocky formation called Es Queixal where Dan went with a helicopter unit to shoot plenty of coverage, from different angles, in order to recreate the same geography in post. The place is particularly interesting for appearing very remote and lost in time. Once delivered, we stitched the images together to create a first layout covering all the angles in the sequences. Then, based on that first pass we started working creatively with Stef to create more interesting compositions, on a shot-by-shot basis, by twisting the geography through reprojections of parts of it. After a first layout approval, we dressed up the 'new' geography, so to speak, enhancing it with drifting clouds [of course!], moving vegetation, and various details, sometimes reprojecting on actual 3D surfaces to get a better sense of depth and parallax. A very close eye has been kept on picture consistency and on the continuity.”
“For the so-called Skybar Sequence, where Hoggins throws the critic Finch off the building, the approach was similar,” explains Alessandro Cioffi. “The scene takes place in London this time, and the live-action was shot in studio. Stef Ceretti provided us with a complete set of images taken on location. First we combined the various elements to form up the expected background, a fairly big matte painting to cover all the angles. Then we established the mood, very specific for the sequence, and brought this big image 'to life' by adding animated details such as domestic light and life in the surrounding buildings' windows, cars and people in the streets, blinking airplane lights in the sky or even a slightly spinning London Eye in the distance. Once complete and approved, that was reprojected in compositing on geometries and cards to accommodate every camera and to generate the establishing shot, a full CG one, featuring a pushing-in camera, gently hovering over the roofs of London.”
“Of particular enjoyment was the pivot moment of the sequence, the actual throwing of the critic out from the terrace,” reveals Cioffi. “Here we animated his digital double, shaded and rigged out of a 3D scan of the actor, following a deadly parabola before smashing itself on the street below. I know it may sound mean, but these type of actions are very fun to animate, with attention on the details, such as the scarf, which shortly follows the heavy body hitting the ground. The contrast between the lightness with which the scarf gently lands and the shocking moment of Finch's violent impact, I think creates a moment of pure black humor. ” When asked what was essential in making the cinematic moment believable, Stéphane Ceretti replies, “Animation, it had to look real. But we wanted to retain an element of comedy in it; it did not have to be just gross. The arms moving as if he is trying to stop his fall and then protect himself from the ground sold it for us. The splash on the ground was a bit over the top but again, that was there to help sell it as a ‘funny’ moment.”
“More than a challenge was the work we did on the sequence where Cavendish [Broadbent] and his friends do a stopover in a pub while escaping from their persecutors,” states Alessandro Cioffi. “It has been both interesting and fun. In this sequence, as they arrive at the pub, a fight starts and one guy gets hit in the face and his tooth flies through the air, ending up in Cavendish's beer. Besides being a fun moment in the movie, we had some good laughs while animating the digital tooth flying off, in a spatter of blood, from the guy's mouth, describing a parabolic trajectory right in front of Broadbent's eyes. The situation is just hilarious in itself and every review we did in the house was a good occasion for new funny comments and more laughter! Technically the scene was shot at 72fps with a hand held camera, with the idea to speed ramp it dynamically in post, while following the tooth trajectory. We first tried to preserve the camera move, but, apart from being naturally too jittery at normal speed, Stef and Dan wanted to reserve the freedom to redesign the tooth trajectory and consequently the camera path. The cinematographic concept here was to let the camera lag behind the flying object. For this reason we tracked the camera and reprojected it on a plane in order to gain some room for adjustments, working both on its path and speed, while designing the desired tooth flight trajectory.”
A lot of background replacements and CG augmentation had to be done by Trixter. “Besides the above mentioned full CG backgrounds we worked on several other replacements and, more generally speaking, picture enhancements,” says Alessandro Cioffi. “For example, the work we were asked to do in order to complete the transformation of Doona Bae into a Caucasian character, Lady Tilda Ewing. Completing, meant in a way that we had to intervene digitally on her facial features, in addition to the heavy prosthetic makeup – warping, removing, and adding details in order to adjust the look of the character accordingly.” No tools needed to be invented. “We made a broad use of all sorts of techniques, in both orthodox and unorthodox ways, but I wouldn't talk of new technology development.” Cioffi remarks, “For the entire production time there has been a very special atmosphere here in the studio. Each and every member of the team was aware they were working on a memorable project and delivered some further contribution, in terms of enthusiasm and positive spirit. We're all are proud of having associated our name to such an epic show.”
“We worked previously with Dan Glass on
Ninja Assassin in 2008,” states RISE Visual Effects Studios VFX Supervisor Florian Gellinger. “Over the last couple of years we met every now and then and I showed him our work on
Harry Potter,
Captain America [2011] and a German movie called
This Is Love [2009]. One of the signature shots that we did for
This is Love is a violent full frontal car crash, very different in style yet technically similar to one of the shots that Dan and the directors wanted to do in
Cloud Atlas. When he set up his production office in Berlin in 2011 he gave us a call and we went for a beer.” The German VFX company took an extra step to make sure that the requirements of filmmakers were achieved. “We moved workstations and personnel to their office for pre-visualizing the VFX heavy sequences, especially the ones we were supposed to do in post production. Dan, Stéphane and the directors worked closely with our artists until they had something they liked. Our artists created technical breakdowns for the shoot that provided a perfect reference for all departments. I joined Stéphane on-set whenever one of the sequences was being shot that we were to work on. Also Stef was located near our office during post-production. I would hop in my car almost every day and we would look at the latest iterations of our shots on his screen.”
“We served as the main vendor for the entire Luisa Rey story set in San Francisco in 1973 revolving around the Swannekke Lake nuclear power plant,” says Florian Gellinger. “The power plant is supposed to be located on an island somewhere in the San Francisco Bay Area and had to be created digitally for day, sunset and night shots including the bridge leading to it. The day shots of the VW Beetle driving towards the power plant were shot on a bridge spanning a river in Scotland. We extended the bridge with a 3D model that we built from a LiDAR scanned section of the real bridge, added street lights, the island and the power plant. For night views of the bridge we shot the Volkswagen Beetle on an airfield in Berlin to simplify practical lighting. We then added the digital bridge model around the runway including street lights, CG water and the power plant at night. All shots inside the car with Halle Berry were shot in front of blue screen at Studio Babelsberg with the exterior of the car being almost always entirely CG.”
“The villain Bill Smoke [Hugo Weaving] rams Luisa Rey [Halle Berry] in her VW Beetle off the bridge causing her to drop into the water underneath,” states Florian Gellinger. “There are two shots of this in the movie: One shows the VW Beetle directly from above, hitting and destroying the railing, rotating upside down and dropping. In the other shot the camera remains inside the car framed at the beginning from the assistant driver's seat on Halle's face and then moving around her to the back seat looking through the windshield as the car drops. While it drops everything inside the car is starting to float spinning through the air to give the shot the sensation of zero gravity. As the car hits the water surface below all the props are catapulted forward. We achieved this with a mixture of rigid body simulation and keyframe animation of CG props. In the following shots the VW Beetle sinks in the nightly Bay. We shot the sequence as dry-for-wet in front of black. We then added a simulated full CG bubble and plankton environment with the 3D model of the Beetle being pushed through a fluid container. The fluid container reacting to the car model provided us with the perfect swirls and turbulence for the murky water. The cracks that build in the windows of the car were procedurally generated in Houdini as geometry that would then refract everything around it correctly.”
Corporate sabotage leads to an aviation disaster. “There is also an exploding airplane in the Luisa Rey story that we simulated and rendered entirely in Houdini using rigid body simulations for the bits and pieces breaking off and pyro for fire and smoke,” states Gellinger. “We modelled the plane in both flying and destroyed states to have control over even the smallest bits and pieces, something a fracturing tool could never provide. The huge advantage of rendering all of this through Mantra was consistent lighting on all CG objects, scattering of light through smoke volumes and direct reflections and lighting from the fire fluids.”
“Some daytime shots with Halle Berry in San Francisco were shot in Glasgow and we used our LiDAR scan of the location to virtually dress the street with CG telegraph poles and wires, to replace houses and to add the Oakland Bay Bridge and the Transamerica Pyramid,” remarks Florian Gellinger. “One night shot of San Francisco was shot in Düsseldorf and extended in the background using stills provided by production and CG cars, another was a virtual camera pan based on stitched stills with animated details like flickering neon lights, driving cars and boats - and if you look closely: there is a party in an apartment with our office crew dancing as silhouettes.”
The visual effects contributions of RISE Visual Effects Studios were not entirely situated on land. “For the episode set in the South Pacific in 1849 we added thunderstorm skies to plates of the ship Prophetess,” remarks Florian Gellinger. “We colour corrected the shots as day for night and used the suppressed direct sunlight as light source when lightning was supposed to strike. We did some sky comps, period clean-ups, added cars and extras to shots of the Edinburgh monument for the storyline set in Cambridge 1936. The doctor who frees Sonmi of her collar in Neo Seoul 2144 needed a laser scalpel and a bionic eye patch to help with surgery. We provided five concept art paintings and three variations of the eye patch, modelled and textured the preferred design and tracked it to the surgeons head.”
A pivotal CG creation is featured during the post-apocalyptic story
Sloosha's Crossin' an' Ev'rythin' After. “For the Satellite Communication Center on the mountain top on Hawaii in 2321, production provided us with magnificent concept art,” recalls Gellinger. “The main entrance of the SatComCenter is based on a LiDAR scan of the power plant's entrance in the 1973 Luisa Rey
story. We tracked the helicopter plates and generated a dense point cloud for the mountain topography. Using our LiDAR meshing workflow we were then able to generate a detailed 3D model of the mountain as the foundation for the SatComCenter base. The SatComCenter became one of the most detailed models that we have built to date. We added destruction and decay to the facade, textured rust onto steel scaffolding and cracks onto the concrete floor. We then positioned CG grass growing from the cracks and dust and debris on the Satellite dish to crumble off once it starts to move, causing wires attached to the structure to start swinging. Clouds slowly moving around the building scattering the sunlight were generated using Houdini and rendered using Mantra while the building was done in Maya, Mari and MentalRay.”
“The biggest challenge in creating the power plant was that its design was based on the San Onofre nuclear power plant south of Los Angeles,” states Florian Gellinger. “There was no way of getting a shooting permit even close to it. So we built the power plant from images Stéphane took at other industrial compounds and nuclear power plants in Germany. For the Satellite Communication Center I researched stills of cable cars and restaurants in the Alps providing us with perfect references of steel, concrete and stone high up above the clouds in the mountains. And with San Francisco, Stéphane provided us with plenty of high resolution raw stills that we could stitch and turn into HDRs. That's pretty much as good as it gets.” No shots were shared with other VFX facilities. “We did provide ILM with the VW Beetle asset; they added it hovering somewhere to the background of the Skiff Chase in Neo Seoul. The SatComCenter was also rendered as a still from great distance for Method to comp into a mountain vista. And because we did the LiDAR 3D scanning of locations, sets and vehicles for the entire show we provided meshes and point clouds of almost everything to production.”
“Our biggest challenge was dealing with Houdini as the newest addition to our tool set while relying so heavily on it for a huge variety of shots,” notes Florian Gellinger. “There was little room for error but the payoff was gigantic. Also the diversity of shots and with it the number of models, textures, plates, elements and tasks was overwhelming.” No particular sequence stands out. “We love them all. The San Francisco shots are absolutely seamless, perfect set extensions. The SatComCenter is just beautiful in shading and design. All the details on its facade and the mechanism that moves and opens the satellite dish were a lot of work and I think it shows. The car crash on the bridge at night was conceptually so well laid out and planned by Dan and Stef. In the end there is plenty of eye candy that works very well on so many levels.” Gellinger remarks, “I would like to thank Dan, Stef and the directors Lana and Andy Wachowski and Tom Tykwer for letting us create what we had always dreamed of creating; we are eternally grateful. And thank you to my team that was so incredibly good at individually creating all those bits and pieces of the puzzle that ended up fitting so perfectly together.”
“We had heard about this show and it was great news that a German production company was planning to realize such a challenging project,” states Scanline VFX Executive Visual Effects Producer Ismat Zaidi. “Of course we were very curious to find out more about the script, and be a part of this show. We got in touch with Dan Glasss and met him in Berlin to discuss which parts of the VFX work might be the best match for Scanline.” A multi-media approach was adopted by Dan Glass when articulating what was needed for the movie. “First of all we received breakdowns and concept art, and had a few CineSync sessions with Dan to clarify what the directors wanted and the look they wanted to achieve. Dan sent a lot of references regarding to the look and feel of certain elements, which was helpful. For example he shot references of smoke layers for the transway. Throughout the entire production time we were in close touch with Dan, had regular CineSync sessions to discuss the shots we did in Munich.”
Along with a couple of individual shots, Scanline's work included sequences, such as the Slaugtherhouse Ship, the Reservoir Tube Sequence and the Prescient Hydrofoil Sequence,” states Ismat Zaidi. “A lot of the shots required water interaction, shots perfectly designed for Scanline´s in-house software Flowline. Scanline worked on the reservoir sequence tube, where Sonmi and Chang are being chased by the enforcers. Chang and Sonmi finally escape after Chang destroys the tunnel. We created the entire environment for the reservoir tube, underwater explosion as well as the water simulation after the tunnel is destroyed. Furthermore we worked on the slaughterhouse sequence with various shots of the exterior and interior of the Papa Song Freighter. We did a couple of shots with the Hydrofoil and the Hovercraft and a few set enhancements as well.” Additional visual reference material was sought out to complete the Reservoir Tube Sequence. “It is very hard to find visual references for underwater explosions; most of the references are from other movies where they were either CG or miniature. Dan shot references for the force fields the vehicles travel on, but we still needed a lot of look development back and forth to achieve the look the directors liked. For the slaughterhouse interior where we did two full CG shots of the processing facilities and did a lot of research on human bodies.”
“Dan had carefully chosen the sequences he awarded to the various facilities in Germany,” observes Ismat Zaidi. “Other than a few assets such as enforcers digi doubles and a BG DMP for Neo Seoul we did not receive any assets from other vendors. We did not share any of our shots with other vendors, though Method did some clean-up work on Chang’s eye makeup for a couple of the shots in the reservoir tube.” The biggest challenge was in producing the grim fate suffered by the Sonmi clones. “I would personally say it was the two full CG shots where we had to create the environment and show the entire “processing” of the clones in the slaughter freighter. The processing plant shots were huge and we ended up integrating almost every feasible piece of CG effect. We created an entire processing chain with rigid body simulations, cloth simulations, hair, fire, water, fluid simulations, and blood running down the bodies. Fire, gas and fluid simulations were integrated in several areas of the shot to show off big bubbling fluid containers, blood droplets and fire gouts. Other than blood-colour liquids used as elements, the entire scene was CG, featuring complex digital machinery that process the dead clones. For each part of the process, we had separate 3D models that needed individual texturing and shading. All of the machines had to work and move, which required a lot of animation and simulation. On top of that there was a pretty lengthy design process as the whole disassembly line had to be invented from scratch. Virtually every visible piece of machinery was especially designed for these shots.”
“One of the shots with the prescient hover ship was initially a filmed camera plate with a pan to reveal the prescient hover ship approaching the island,” states Ismat Zaidi. “As the shot progressed the production was not satisfied with the camera pan that followed the initial prescient ship animation. The shot developed into a rebuild where the camera move was replaced, the ocean was replaced and the remaining plate footage was reprojected onto helper geometry. Essentially that gave the production the ability to change the camera move on the already filmed material. For the prescient hover ship we employed a large scale Flowline fluid simulation that covered the entire length that the ship traverses. That was one of the largest areas of fluid that we simulated so far. There were other shots with the prescient hover ship moored in the island bay. All of the prescient ship shots were difficult to integrate so that the ship would feel sufficiently massive enough. We had to build in little details and effects that would jump out and put the prescient ship size in perspective. We also worked on an extensive hydrofoil shot where the hydrofoil whips past the camera at incredible speed towards the distant island. The original plate has been extensively remodelled to allow for adjustments to the camera move. The complete ocean was done in CG, leaving only parts of the sky and the background island projected and extended from the original plate. Simulating the hydrofoil ocean interaction was done in various steps because the production had very specific ideas on how the visual quality was supposed to look like. Originally we attempted to accurately simulate the interaction but discovered that we needed a lot more control over the process and created the final look by using half a dozen different simulations that were partially simulated attached to the hydrofoil to deal with its extreme speed.”
“Since we were using Flowline as one of our main tools for the simulation work, we have the great advantage that our developers sit ‘next door’, says Ismat Zaidi. “They were great in their support and devised various tools to make the work easier for our artists.” Looking back on the sequences, Scanline’s Executive Visual Effects Producer states, “We are proud of all the work we did for
Cloud Atlas, each of the shots had it´s challenges and it was great working with Dan and the directors on this show. Our artists loved working on the Slaughterhouse Sequence, as well as the Hydrofoil and the Reservoir Sequences.” Zaidi remarks, “Working on this show was a great experience and we are happy to have had the opportunity to be a part of it. A project such as
Cloud Atlas as a US-German coproduction is a novelty and we do hope that it is the first of many more to come.”
“We were approached by Stéphane Ceretti, who we have known for years, after Dan Glass watched
The Iron Lady [2011] and wanted to know who had done the prosthetic VFX work,” states VFX Supervisor/Director of BlueBolt Angela Barson. “We were given a description of what they wanted, and based on that we would do a few different versions on a hero shot. They would choose their favourite version which we would then work up to final for the hero look. Stéphane often came into the facility for reviews so we could look at the work together. When he wasn’t in London we would do CineSync sessions so we could be viewing the same footage and draw on the QTs.” The alternations needed were more significant than was required to transform Meryl Streep (
The Devil Wears Prada) into British Prime Minister Margaret Thatcher. “The work on
The Iron Lady was very minimal and subtle; we weren’t changing features as we did on
Cloud Atlas. The differences in the before and after images for some of the
Cloud Atlas characters were quite substantial.” The number of digital cosmetic surgeries needed to be done to the different performers. “Having a large cast of actors doesn’t make the work any harder. Each one is treated as its own case. A look was developed and approved for each character on a hero shot before the bulk of the shots were done.”
“The most complex character would be the hotel manager, played by Tom Hanks,” states Angela Barson. “He required some extensive work to get him to look natural as an older man. Originally the plates that we were given had the character with a bald cap and some visible glue joins around the temples, nose, mouth and neck. The challenge was to remove these issues while introducing sun spots and freckles for skin texture. Due to the nature of the way his head moved we required his head to be roto animated. With this animated model we were able to project on to it painted textures of his new skin. This was then composited and finessed with the main plate.” Prosthetic effects had to be seamless blended with digital enhancements. “With this type of work it’s always about the attention to detail. You tend to be working on areas which are the focus of the shot – a characters eyes or mouth – so the work has to be perfect. A lot of attention is paid to the subtleties of the skin grade and texture. We also had to continuously check the shots in sequence to make sure the consistency never drifted.”
“This work was all done with the standard set of software tools,” remarks Angela Barson who singles out a certain sequence. “The Archivist [James D’Arcy] needed a lot of work on his protruding forehead and deep set eyes, as well as changing his eyes and reshaping his eyebrows to look more Asian. We needed to rebuild his eyes in order to bring them out from the shadows of the forehead.” Maintaining a unified look was not a problem. “We shared a number of shots with Method, but the work was on separate aspects of the shots, so no problems maintaining a unified look. It worked very smoothly.” Barson notes, “We completed over 70 shots, working on seven different characters: The Archivist, Mephi [Hugo Weaving], An-kor Apise [Keith David], Seer Rhee [Hugh Grant], Rufus Sixsmith [James D’Arcy], the Conceirge and the Hotel Manager. Each character presented its own set of problems, but the general techniques used were similar across all characters.”
“We were approached by Dan Glass and Stéphane Ceretti based upon our previous body of work specializing in cosmetic digital enhancement,” states Lola VFX Visual Effects Supervisor Clark Parkhurst. “After initially receiving the plates and familiarizing ourselves with the photography, we would have a series of conversations several times a week regarding the progress of the shots. Most of these conversations involved using CineSync, so if Dan or Stéphane had a particular note we could draw it on the image, which we would then export and reference as we worked.” It was important to ensure that the digital changes did not take away from the acting performance. “In the case of
Cloud Atlas, especially, with the character of Nurse Noakes, the idea was to retain the characteristics of the actor underneath the makeup, while making her more feminine at the same time. The story dictates that you recognize the actors playing these different roles so as you can follow them throughout the different timelines.” Parkhust adds, “We only worked on a handful of characters in the 2012 sequence. We did the most amount of work on the Nurse Noakes character, played by Hugo Weaving. As far as working on multiple characters, we strived very hard to maintain a consistent look from shot to shot. That's probably the most important thing you can do in any situation involving this kind of work or any visual effects work.” The makeup prosthetics needed to be digitally augmented. “The practical effects looked great, but there were some problems with skin texture, seams, colouring and things of that nature that Production wanted to tweak and fix a little bit; that, combined with slimming down and reshaping some facial and bodily features to accentuate the feminine qualities of the character.”
“We always do a lot of visual research on the actor that's playing the role,” states Clark Parkhurst. “And we consult with cosmetic surgeons constantly to keep up with techniques and to maintain our understanding of how the human body ages and is perceived by an audience.” The biggest challenge was Nurse Noakes. “With anything of this nature you want to approach the work practically and efficiently. Our main goals were to maintain the likeness of Hugo Weaving while we touched up the practical prosthetics and to add some feminine qualities to his face and body by altering shape and bone structure.” No technology needed to be invented. “We didn't develop any new software for this project, but we are always striving to find new and more efficient ways of achieving visual effects like what you might see in
Cloud Atlas,
Captain America or any other of the films we've had the privilege to work on.” Parkhurst lists the bar fight as his personal favourite. “We have a great team at Lola and I'd like to say that this wouldn't have been possible without Max Leonard, our Producer, and several talented compositors including Jeremiah Sweeney, Kazuyoshi Yamagiwa and Demitre Garza. I'd also like to thank Tom Tykwer, Lana Wachowski and Andy Wachowski, Dan Glass and Stéphane Ceretti for involving us in the project.”
“I came onto the show after it had been awarded,” states ILM Visual Effects Supervisor Russell Earl. “But I believe the scope of work grew and Dan had done some work with us on
Speed Racer, so he thought we’d fit the bill. We wanted to be working with Lana and Andy so it was a great opportunity.” Video conferencing was conducted with Dan Glass. “We had CineSyncs with Dan where we would discuss what the shots should be and where they were headed.” The legendary visual effects company founded by George Lucas (
Star Wars) and recently sold to Disney handled a small but dramatic portion of the movie. “ILM worked on the Skiff Chase Sequence, essentially, the bit where Chang and Sonmi are on speeding away on the skiff being chased by the enforcers.” A classic science fiction film directed by Ridley Scott (
Prometheus) was used as a visual reference. “The look of Neo Seoul was described as ‘High Tech Blade Runner’. Dan Glass provided us with some concept art to use as a starting point. We looked at tons of different unique buildings throughout the world and used them for inspiration as well. We were building a future tech city with an Asian aesthetic, so we tried to include some traditional Asian architectural details in the city. We also had to develop the look of the transways, the sort of mag lev energy highways that the people of the city travelled on.”
“We didn’t share shots but did share assets and looks with Method,” remarks Russell Earl. “We built the digital doubles for Chang, Sonmi, and the Enforcers. We also built a bunch of the buildings and vehicles. Method supplied ILM with the Skiff and Gunship models. The hand off seemed pretty seamless on our side. Once we had the models with textures, we did our own look development and rigging as well as animation.” Practical effects had to be combined with the digital ones. “The biggest challenge was to make the dynamic camera moves work with the animation and blue screen plates that were shot. The plates were chaotic and the moves were dynamic. The challenge was to make the scenes believable, and to get the cameras and animation just right, as well as make the lights and the world they were travelling through fit with the lighting in the plates.” He states, “The ILM shots all had fully CG backgrounds. We also rendered the skiff in most shots as we wanted to better match the skiff lighting to the CG environments lighting. We had some shots where everything was CG including our hero characters. We also had multiple shots where we were doing blends from live action actors to our digital doubles and vice versa.” Some technological modifications were made to the visual effects process. “We were working in a new pipeline for us, harnessing Katana and Arnold. While the tools were new, we relied on the strength of the great artists here to realize the work.” Earl is pleased with the end result. “It was great fun to be a part of a project that had such passionate directors.
Cloud Atlas is an epic movie that we were glad to have the opportunity to collaborate on. We had a great crew here at ILM and I am proud to have worked on this film.”
“Exozet Effects has offices in Berlin and at Studio Babelsberg directly opposite from the sound stage where the Wachowskis where shooting,” states Exozet Effects VFX Supervisor and Lead Compositor Falk Gärtner. “I don´t really know how our producer got them over exactly; there might have been some coffee bribes involved! One day they came up into the studio, we showed them around and they talked to some of our team. Then we received a test shot to work on that was a makeup touch up from a sequence. We were also asked to help with some of the post-visualization work after they were done shooting. The post-viz happened inside the production office and we were able to build up a very friendly relationship with them, helping out in every way we could. After completing the test shot and post-viz work, the visual effects production team was happy with the results of our work and started awarding us shot packages.” Gärtner remarks, “We mostly worked with Stéphane and then Dan would come in to sign off on the Final Approval of a shot with the directors. Stéphane was friendly the whole time and explained patiently in a lot of detail what they were looking for. Most communication was happening through CineSync sessions which was helpful for general feedback, but became more challenging when discussing single pixels and grain due to the lower resolution quality of quicktimes - in the end, it worked out well enough and all sides were happy with our final results. Stéphane and Dan were really specific about the level of quality they wanted to achieve as I’m sure they were under a lot of pressure to deliver the best quality possible to the directors. Overall, it was really good to be pushed that far – it really benefited the film and our team.”
“The post-viz work that we did was distributed to various vendors including ILM and Scanline was to be used as reference for complex sequences,” states Falk Gärtner. “For the live-action visual effects work we completed, there was no sharing with other facilities involved; the artists finished the shots they had been assigned and we delivered final versions directly to Dan and Stéphane. The production did some reassigning during the schedule to adjust for the strengths of different companies. The concept of sharing between facilities is one of the strong points of the UK market and less so in Germany simply due to the volume of visual effects work within each country. Personally, I think this is something Germany can still learn a lot from, especially, since companies here are highly specialized so sharing would actually benefit the local industry as a whole.” A two-time Oscar winner proved to be the biggest challenge. “Touching the face of an A-list actor is always a huge undertaking as so many people will look it over and will criticize every little problem. The human mind is absolutely trained to look at a face and will see the tiniest glitches. You get into the uncanny valley very fast. Touching up Tom Hanks’ makeup when he plays Dr. Goose on so many shots, sometimes replacing half the face was quite a challenge. When we got the first test shots we tried to devise a 3D solution for the touch-up, but with a huge number of shots, the tight deadline and the varying degrees, lighting changes and places the makeup needed fixing made that approach not feasible. In the end we mostly went with a clever way where we spline deformed a painted patch with a planar tracker, that got us a long way. The rest we painted by hand or tracked tiny patches on.”
“We did not specifically develop anything new,” says Falk Gärtner. The spline deform/planar track we actually adapted from some forum posts we found. We had to write two or three little scripts to make it work with our toolchain. We did use a complex liquid simulation tool for the blood sequences. We also had to ramp up and build an extremely fast pipeline with a lot of automatization inside our facility to handle the volume of work for the project.” The digital enhancements were achieved seamlessly. “The prosthetic fixes on Dr. Goose [Tom Hanks] were definitely challenging and worked out quite well. I personally couldn´t spot anything when I saw the movie the first time on the big screen and I knew where to look. The battle sequence in the far future was more fun to do. Doing realistic blood and having an extreme close-up CG arrow shot through the head of one of the Kona was quite satisfying. Also we developed a look for how these futuristic weapons would shoot – you know, that is stuff you always wanted to do when you first got into VFX, but rarely get the chance to actually do it for the German TV market.” Gärtner observes, “As one of the smaller vendors on the project there had to be a lot of trust from the production that we could actually pull this off in the quality they wanted. A lot of our shots went through in the first feedback round which was totally unexpected. We later heard that the VFX department, for the fun of it, had internal bets going on who will finish all their shots first. We ended up in the top 3 out of 15. Overall, the project was managed very well and we were always kept in the loop as to what was going on. The cooperation with Dan Glass, Stéphane Ceretti, Marc Kolbe and the whole production crew was on a very professional level with a lot of creative discussions and fun. We were fortunate to work on such an impressive project and we are ready to work on more of these exciting projects in the future.”
“We were splitting off a character and you could ensure continuity through that process,” states
Cloud Atlas Senior Visual Effects Supervisor Dan Glass. “It meant in some cases we had shots that were shared but at least we were supplying a character that went into a composite that someone else was doing. Where we had a view of the city for Neo Seoul; all of those particular places in the city would be done by the same vendor. We had occasions where ILM was doing the chase which followed immediately from Method Los Angeles segments and then leads into a Scanline segment. The junctures at which they changed were quite carefully designed whereas we would send reference work back and forth, models were shared. Method Los Angeles built the gunship and the skiff; they were supplied out to ILM and Scanline. Buildings were designed and built at different facilities and then we would share them so ILM would build some and that would get shared with Method Vancouver and Los Angeles and vice versa. We used it to our benefit so that we could get more variation out of things.” There were no pipeline issues between the different visual effects vendors. “For sharing the assets it worked well with ILM. There were conversations we setup beforehand to make sure that we were all talking the same language and flying the same way. Inevitably, there is some adaptation that takes place to bring something in but we weren’t sharing complex character rigs, for example, which can get propriety; these were hard surface models that didn’t have any mechanisms they weren’t over complicated. It wasn’t complicated to do that.”
“The biggest issue was the amount of sequences taking place in so many various time periods and places,” states
Cloud Atlas Visual Effects Supervisor Stéphane Ceretti when discussing the research conducted for the movie. “We had a huge amount of visual references for the futuristic cities but we also needed some good references for the past and present.” Key to introducing the different stories was the establishing shots. “We did research on the look of the San Francisco Bay Area in 1850 and 1973 [what buildings were built then, which ones were not, the type of advertising posters, cars, and the look of the roads]. Also we looked at images from Edinburgh in 1930 to find out how the surroundings of the Scotts monument looked like at the time, as well as images of Cambridge at the time.” Ceretti explains the opening scene for
Letters from Zedelghem. “I found a few images of Cambridge that I showed to the directors and they picked a view from the rooftops. I went to Cambridge and scouted the area a little bit and found the exact same place where these photos were taken and did a photo shoot from the same location. On this day the clouds in the sky were amazing and we only had to animate these a little bit to make the shot more alive, move the shadows on the buildings and add cars and pedestrians, smoke on the rooftops. We had to clean-up all of the modern elements to bring us back to the 1930s.”
As for the beginning of
Half-Lives: The First Luisa Rey Mystery, Stéphane Ceretti remarks, “We spotted quite an iconic view of San Francisco that everybody liked as well as another building that we wanted to use as the one where Luisa is before she gets stuck in the elevator. We took stills of both views and dropped the building into our hero view of the Transamerica building and the bay bridge. Again, it was all done from stills so we added moving cars in the streets and on the bridge, the people dancing inside the building, planes and boats, and ripples on the water in the bay.” Glasgow, Scotland had to be transformed into the coastal California city. “Having a good amount of references for what we wanted to achieve [was essential]. We pushed to get a photo shoot to happen to make sure we matched to reality as much as possible. A lot of attention to details and what was the essence of the streets of San Francisco in 1973. Rise who did the sequence did an amazing job, I think even people from San Francisco would have a hard time not thinking it was shot there [and some actually told me so!].” When it comes to the present day tale
The Ghastly Ordeal of Timothy Cavendish, the native of France reveals, “Here again it is a composite view of a hero building photographed in London, then another view of London showing the London eye, the parliament tower and Big Ben as well as other London landmark buildings. We blended all of them together and added the people inside the Starlight Bar at the top of the building.”
Maintaining a unified look throughout the movie as a whole was not a major issue. “We did not really have to because all of the stories had their own feel to them,” says Stéphane Ceretti. “There is a unified feel for the way the stories are told and how the characters react to the challenges they are facing, but all of the stories have their own distinct feel. What we did though was using the same sets and re-dress them to help the sense of recurrence in time. It is not always directly obvious but your brain does pick it up. The Startlight Bar in London is the same set as Papa’s song in Neo Seoul. The Swannekke power plant entrance hall is the same as the communication center in Zachary and the temple at the end of the Sonmi story where Chang dies. Ayrs music room is the same set as the Aurora House dining room and lobby. The most obvious one is The Ayrs Chateau being the same as the Aurora house itself.” The technique assisted in the editing of scenes such as when Frobisher climbs across his bed and transitions to the hotel room of the older Sixsmith reading the same letters. “It wasn’t even green screen,” remarks Dan Glass. “We set the cameras to align and shot it; that was an example of a redress set so we could line things up. It was a series of roto mattes to dissolve into different areas and transition them through. We did that with our in-house team. That was achieved relatively straightforward.” Glass points out, “You might make the choice because you’re on a budget and trying to save on allocating resources, but it was a conscious decision in order to help the stories tie together. There are a lot of things like that where it’s not just the characters reprising roles but sets and props come up again and again.”
Balancing the need to make the actors appear familiar yet different in portraying the various period roles was no small task. “In that sense Neo Seoul was the big challenge in a design sense,” notes Dan Glass. “Chang was probably the biggest challenge as a character because he has a fairly significant role but we had to adjust him convincingly. We didn’t want to Jim Sturgess to go full Korean. We felt it would be better to design his look as a Korean related individual. But the goal in all of our early concepts was that we wanted the actors to be recognizable. If we really wanted to go to make Jim Sturgess fully Korean we probably could have done it but then what would be the point? People are slightly thrown on the first viewing but we wanted it to be quite visible that once you knew that you could go, ‘Ah, that’s him. That’s his performance. Those are his eyes. That’s his mouth. It’s the same with Hugo Weaving as Nurse Noakes and the many roles that Tom Hanks plays. It’s almost like a theatrical device where you want to initially help the audience feel transported into the different stories but at the same time recognize what is underlining.”
A different approach was adopted for the holograms. “There are a couple of things in talking to the directors; they were keen to avoid the typical trail of video graphics,” states Dan Glass. “They had commented which is true that almost all the time in science fiction movies video graphics are represented as a regression of the technology that we have today. When you think about it typically these fuzzy video interference type displays, especially, the holograms and they wanted to have things perceived to be pristine and more clean and perfect than what we have these days. The restaurant is this wild consumer overload of information and the animated floor was for fun, the interacting fishes with the people who walk around on it. The terminals the folks use which we see Chang operate are called orisons; they’re floating discs that appear to materialize and can be manipulated around and switched off. We designed them to be opaque not transparent which is the typical way of representing graphic displays. We decided that we didn’t want to do that just to do something different plus the idea of the different shaped frame. A rectangular frame is so standard that we wanted to try something different and we did.” The futuristic technology needed to be relatable for the audience. “It is a fine line. You want people to understand the core nature of what is happening so they don’t get thrown out of the story as a whole but you want something in there that raises people’s curiosity. You want them to participate in the experience of watching the film. It’s not the sort of movie you sit back and have it tell you everything that is going on. You have to participate. There is a lot going on to pay attention to; I for one think that makes it a thoroughly enjoyable experience.”
“We aimed to shoot as much as we could practically,’ remarks Dan Glass. “There is a lot of stuff which is more invisible effects whether adding blood or arrows. Then there are things more visible. The high tech ship that arrives and some of the devices that Meronym has healing her arm or scanning the mountain. When they are climbing it was fantastic we went to Mallorca we were there for five weeks. We were literally up mountains with Halle Berry and Tom Hanks for which you get fantastic production value and I’d like to joke there are shots in there that most people knowing that the Wachowskis were involved would think, ‘That’s got to be a green screen.’ But it wasn’t. We took them right up to the top of the mountain. There is a sequence when they’re climbing the vertical ridge which was a green screen area but we shot real plates of the same mountain that we were at so it would blend well with the material around it.” The dramatic moment with the broken dishes being suspended in mid-air was produced by utilizing mainly traditional methods. “They shot it with five big cameras and there were a couple of instances where we were layering up the big wide shot where we layered up elements because the actors were in frame. A lot of that was practical. We tried these smart approaches to things as much as possible.”
Great effort went to make sure that the visual effects did not overshadow the storytelling. “I’m always trying to make it feel contextual so if it doesn’t need to be showy it shouldn’t be,” states Dan Glass. “But things that are showy typically in this film are the things if they existed, if that communication centre existed on the top of that mountain, you would show it. You would shoot it with an aerial helicopter to show off the grandeur of the place. It just didn’t happen to exist.” The interior of the same building features rings that hover upwards. “What I thought was cool about it is that it’s so strange that you feel that it must be something to it that we don’t understand; it offers this mystery which is appropriate for some technology in the future. You have some license with futuristic technology to go beyond what your viewers can comprehend and that is almost as important because if they go, ‘It’s one of those.’ Then it is not adding some futuristic mystery.” Stéphane Ceretti remarks, “I don’t think it was ever an issue. The story [stories] were strong to begin with and the cast we had on board, there was no doubt our VFX would not be going in the way. We never saw our work as something that had to happen as an eye candy. Visual effects were used to create worlds that had to look believable and bring truth to the different periods we were showing in the movie. We do have some pure action scenes with a lot of VFX in the Sonmi story, but they worked as an ensemble with the story and were necessary to keep the story going in exciting ways.”
“We used most of the tricks that exist,” states Dan Glass. “We built the motion base ourselves for the skiff chase and programmed that. There are things that have been done but we were making slightly customized versions for this film.” Stéphane Ceretti adds, “We had custom made gimbals and I am sure some of our vendors invented a few tools to finesse their pipeline or help with some of the specific challenges we had. But I could not say we invented a ground-breaking technology. Sometimes it is not so much about technology but more about the way you use it.” Glass notes, “What
Cloud Atlas to me typifies in the same way as we did back on
The Tree of Life; it is not focusing on technology per say; it’s about focusing on story and how we design and craft technology to work for us within the context of telling a great tale and showing some great images. Sometimes the obsession with technology for its own sake and saying that’s the benchmark of which we are trying constantly push you can achieve fantastic things. But that’s when you can end up dominating the film and if that’s the purpose of the film that’s fine but in the case of
Cloud Atlas that wasn’t the goal. The goal was to free the directors to be able to tell the ambitious tale and do that in a way that is as invisible to them as possible.”
“The city of Neo Seoul, the Sonmi futuristic city was easily the biggest challenge because of the design scale of what you might have for an entire movie but that’s only a portion of an overall film,” states Dan Glass. “Trying to do that on a budget and yet bring something that was quite exciting visually. We looked a lot at references from
Blade Runner [1982] and the likes we wanted as grimy futuristic which is somewhat pessimistic future vision of things. Coming up with a look and making sure that felt consistent across the different scenes that we had, especially in areas where scenes for various reasons had to go to different vendors. It was getting that level of collaboration that we could work on things, share stuff, and bring it all up to the same level and look over the course of post-production. I love working that way. I’m collaborative. I personally chose Method Studios which is a big part of why they’re involved is there are people I trust and at the same time I’ve been working with other vendors for their specialities and tried to bring a great sense of collaboration across the project like this.” Stéphane Ceretti agrees with his Method Studios colleague that the most significant challenge was being creative despite of the financial restraints. “I would definitely say the budget. It was really difficult to get all these shots done for the money we had. Marc Kolbe, Dan and myself worked quite hard to work around the issue without compromising the quality. It meant we had to go to more vendors that we would have liked but that was the only way to make it happen.”
As to whether there was pressure for
Cloud Atlas to become a 3D movie, Dan Glass remarks, “There never was and I’m glad of that.” Stéphane Ceretti believes, “It’s a unique film. It was a challenge for us to do it but it was all worth it. This movie is very special, you might love it or hate it, but I really think everyone should see it to make their own opinion. This is a movie that touches the audience at a very personal level. One need to see it to make up his mind, don’t believe what other people tell you about it.” Glass agrees, “It was clearly a highly unusual film, highly ambitious and difficult to know how we would succeed but I love taking on and being involved with projects that you don’t quite know you are going to achieve it in the beginning. That’s part of my drive, my masochism I don’t know. It’s something that I find very enjoyable. It absolutely fell into that camp.” There are certain scenes that stand out. “The Sonmi Chase in Neo Seoul is quite amazing, but I like the car crash on the bridge,” says Ceretti. “We found a way to show it and have the audience experience it in a very unique way.” He explains, “The fact that we got Halle in the car on a rotating gimbal for real. We had to cut the car in half and do a lot of pre-shoot work to get everyone confident it would work. Uli Nefzer our SFX supervisor did a great job getting the car on a gimbal, the camera inside on a rail and synchronizing all the components together to match our previs.”
The original vision was maintained throughout the project. “It pretty much stayed the course,” states Dan Glass. One thing that’s fantastic about working with the Wachowskis, in particular, and Tom also, is that they are clear, write their own content and are visual; they typically know what they want and there is a lot of room for creative collaboration and experimentation along the way. They don’t veer dramatically at least story-wise or concept-wise from where they start. It’s a great experience working with them in that regard.” The British visual effects veteran attended the World Gala Presentation of
Cloud Atlas at the 2012 Toronto International Film Festival. “The Saturday night showing when we had that seven minute standing ovation that was exciting, especially, because knowing the directors well and what it took them on a personal level to get this film made against all odds; it was heart-warming to see that reception for them in that theatre.” Reflecting on his two and a half year involvement with the cinematic epic, Glass says, “It is the whole nature of the movie. It is about how everything fits and the tremendous variety that is contained in the film. We have so many one-off views and things that had to be designed that typically a large movie. You have to design a number of things but then you get to reuse them because you see them at different angles. We had so many unique things that we had to come up with and make work within the story. It’s the overall impression. The way it is put together that I most enjoy.”
Cloud Atlas production stills © 2012 Warner Bros. Entertainment Inc.
Cloud Atlas VFX images © 2012 Warner Bros. Entertainment Inc. All rights reserved. Images courtesy Warner Bros. Entertainment Inc, Method Studios. Trixter, RISE Visual Effects Studios, Scanline VFX and Exozet Effects.
Many thanks to Dan Glass, Stéphane Ceretti, Matt Dessero, Geoffrey Hancock, Alessandro Cioffi, Florian Gellinger, Angela Barson, Clark Parkhurst, Russell Earl, Falk Gärtner and Ismat Zaidi for taking the time to be interviewed.
Trevor Hogg is a freelance video editor and writer who currently resides in Canada.