The Art of VFX

REAL STEEL: Erik Nash – VFX Supervisor – Digital Domain

the area — Mon, 02 Apr 2012 20:00:00 UTC

What is your background?
I started my career as a visual effects camera assistant on STAR TREK: THE MOTION PICTURE and spent eight seasons as the lead motion control cameraman on the STAR TREK TV series: THE NEXT GENERATION and DEEP SPACE NINE. I joined Digital Domain and worked as a visual effects director of photography on APOLLO 13, TITANIC and a few other features, then began working as a visual effects supervisor – which is what I do today. I have been fortunate to supervise shows such as I, ROBOT, PIRATES OF THE CARIBBEAN: AT WORLD’S END and most recently, REAL STEEL. Having started my career behind the camera it was particularly thrilling for me to be able to work in the virtual production realm where it’s now possible to direct CG characters as if they are actually present on set – just like you’d shoot a scene with human actors.

How did Digital Domain get involved on this show?
My own experience and thinking for this project sort of began back in 2008, when I had the chance to get an in-depth look at the virtual production process on AVATAR. When I read the script for REAL STEEL a year later, I saw this material as a perfect fit for a similar virtual production paradigm. I didn’t know it at the time, but one of AVATAR’s producers – Josh McLaglen – was also attached to REAL STEEL and had the exact same thoughts for this project. The topic came up in an early meeting and, as I made the case for virtual production I could see Josh nodding in agreement. Dreamworks awarded Digital Domain the visual effects contract and we launched into a six-month preproduction phase.

How was the collaboration with director Shawn Levy?
Shawn was very open and collaborative. We had a really terrific working relationship.

What was his approach to the visual effects?
His sole interest in visual effects was in how it could help him tell the story. He wholeheartedly trusted the VFX team to help him fulfill his vision.

Can you tell us about your collaboration with Legacy Effects?
We had a great working relationship with Legacy Effects. Their supervisor John Rosengrant and I were in complete agreement regarding how the work would be best broken up. They built three practical robots – Ambush, Noisy Boy and Atom – that were used extensively throughout production for shots requiring human contact and upper body animation. We worked with Legacy and with the Real Steel production art department to finalize robot designs and mechanics that could be applied to practical robots and CG models alike, so we could create a seamless transition between the two methodologies as needed on screen.

Can you explain to us in detail the creation of the different robots?
The practical Legacy ‘bots were invaluable for lighting and texture data, as they provided a tangible point of reference for digital characters that needed to be indistinguishable from the real thing. Digital Domain modeled, textured, and rigged eight unique, hero robots for the fight sequences: Ambush, Noisy Boy, Midas, Atom, Metro, Blacktop, Twin Cities, and Zeus, in addition to numerous background robots that appear throughout the film.

How did you rig them?
Early conversations between our rigging team and the Giant Studios team enabled us to get the skeletons and naming conventions worked out. Defining our nomenclature and hierarchies in advance allowed us to streamline the handoff. After principal photography, we would get a live-action plate from production and a Maya file from Giant, and we could go straight into lighting and comp, without having to worry about tracking or layout or any sort of file translation. Because our rigs and their rigs were exactly the same, there was no retargeting needed on the back end.

What were your references for their different animations?
Each robot had a different Mocap performer assigned to it. This gave each robot a distinct personality and way of moving. Each performer had a very clear understanding of the physicality of their particular robot counterpart because they were familiar with the concept art and could see their own performance applied to the Giant Studios version their robot as they were performing on the Mocap stage.

How was the presence of the robots simulated on set?
Once we had all of our fights mapped out in the virtual realm during pre-production, we brought our motion capture system to Detroit and installed it in each of our fight locations. The on-set mocap system allowed us to replay the pre-captured robot action as a live composite through the motion picture camera with complete spatial accuracy. Through the cameraman’s eyepiece and on the director’s monitor, CG boxing robots were visible in the real world set. The actors could then see playback of the recorded take with visible robots in the shot. This was a groundbreaking use of virtual production technology – it had never before been used to this extent on a film that featured so many real-world locations.

Did you create some previs to help the shooting?
For the Metal Valley stunt sequence, Previs Supervisor Casey Schatz modeled all of the camera cranes, lighting rigs, rain bars and stunt rigging that would have to be fit into a very confined area around the set. This allowed all of the department heads to see how all of this huge equipment had to interrelate. We used the previs to figure out how we were going to set up everything on location. When we actually got there, everything went in exactly as we’d planned, and we were able to shoot that very complicated setup in three short nights.

Digital Domain worked on the TRANSFORMERS trilogy. How did this experience served for this show?
The biggest help our TRANSFORMERS experience gave us was organizational and managerial. The greatest similarities between the projects were the quantity and complexity of the work. Stylistically, the projects were very different.

How did you made such great realistic renders for the robots?
DD’s “Light Kit” system, which was originally developed for BENJAMIN BUTTON, produces near-final lighting renders right out of the box using high dynamic range (HDR) images captured on set during principal photography. Once the compositing, integration and lighting teams properly format the HDRs, the resulting 360° digital worldview of each location is ingested by our software pipeline for lighting CG characters within that environment. Typical HDRs are used as an infinite ray dome; on REAL STEEL we projected the HDRs onto 3D geometry and actually traced it in 3D space. Thereby the lighting is much more physically accurate. As our CG character moved through the set, it would get all of the influence of the HDR in the proper three-dimensional space.

Can you tell us more about your work on the futuristic city?
There really wasn’t any “futurization” done thru VFX. All of the cityscapes were achieved in-camera on location in Detroit, Michigan.

How did you create the huge arena for the final fight?
Production Designer Tom Meyer and his art department delivered concept art and a 3d model of a fictional 20,0000-seat sports arena. Our environments team led by Geoff Baumann and Justin van der Lek modified the design to fit the footprint of Cobo Arena in Detroit where the sequence was shot. Then the entire interior was model in detail, textured and lit to match the practical, partial set. We then populated then arena using a crowd system developed specifically for REAL STEEL.

Did you develop specific tools for this show?
The use of virtual production technology – a process we tailored specifically for this film – was the biggest innovation. Virtual production had never before been used this extensively for a location-based feature.
Also, our digital effects supervisor Swen Gillberg and our environments team devised a new crowd technique dubbed “Swen’s Kids.” Eighty-five extras were replicated photographically at very high resolution to achieve crowds as large as 20,000 spectators in the arena fight sequences. The process started with individual extras against green screen who would go through a scripted series of movements – seated, standing, clapping, arms raised, etc. – that were captured by three digital cameras positioned at different angles. The three cameras ultimately captured over 80 terabytes of HD footage. A software tool was written within Nuke to manage the footage and allow for easy replication of crowds depending on the needs of each shot, including factors such as crowd size, proximity to camera, enthusiasm level and camera angle, among others. This system was used to produce anywhere from a handful of background spectators to a crowd of thousands, all generated through an easily managed system that required fewer computing resources than a full 3D solution.

Was Digital Domain Vancouver involved on this show?
100% of Digital Domain’s work on REAL STEEL was done in Venice, CA

What was the biggest challenge on this project and how did you achieve it?
The biggest challenge was figuring out a way to create eight-foot tall working robots that could do battle in a boxing ring and look entirely believable doing so. Our virtual production process was a key front-end solution to meeting that challenge.

Was there a shot or a sequence that kept you up at night?
The scene with the bull at the beginning of the movie was the most difficult. Flesh and fur is much harder to achieve digitally than metal.

What do you keep from this experience?
REAL STEEL stands out for me as by far the most rewarding project I’ve been involved with. Working with Shawn was incredibly collaborative and fulfilling. The work that our team at DD produced is as close to seamless as I could ever hope. It is a movie that I will always be very proud of.

How long have you worked on this film?
We worked on REAL STEEL for about 19 months total – starting with early prep work in December of 2009, through to final delivery in June of 2011.

How many shots have you done?
Digital Domain produced 626 VFX shots for Real Steel.

What was the size of your team?
Digital Domain’s team reached a maximum of 250 people, including artists and production support.

What is your next project?
We’re currently looking at several projects, but don’t know what will be next.

What are the four movies that gave you the passion for cinema?
2001: A SPACE ODYSSEY, CHINATOWN, BRAZIL and THE BIG LEBOWSKI.

A big thanks for your time.

// WANT TO KNOW MORE?

- Digital Domain: Dedicated Page about REAL STEEL on Digital Domain website.

TOWER HEIST: Greg Liegey – VFX Supervisor – Method Studios

the area — Sun, 12 Feb 2012 19:00:00 UTC

How was your collaboration with director Brett Ratner and Production VFX Supervisor Mark Russell?
Mark has a deep understanding of the process which helped us zero in on hero looks very quickly after the previews were delivered. He worked hard to keep all the vendors in sync with each other and get us consistent in terms of the big picture. That allowed to get into the fine-tuning stage earlier in the schedule.

Brett is a force of nature – he works on an instinctual level. He doesn’t over-analyze, he reacts. When the rest of us diverged into talk about technical aspects of the project, Brett would make sure to bring it all back to the visuals. He helped us concentrate on the story-telling.

What have you done on this show?
Method worked on six sequences and two stand-alone shots within Arthur Shaw’s apartment. We also completed 10 shots of digital alchemy (gold enhancement) using 2D color treatments or complete 3D replacements for gold objects in the scenes.

What was the size of the greenscreen for the penthouse?
The greenscreens wrapped around three sides of the apartment set and ran about 70 feet from end to end. They were about 26 feet tall and set back 50 feet from the glass.

Did you need to do some extensive roto work?
There were some angles where the actors ended up in front of multiple layers of apartment windows. The windows in those shots acted like ND filters which made the greenscreen very difficult to key cleanly. So for those, our roto team of David Marte and Alejandro Monzon did lots of articulate roto including reconstructing hair detail to a very convincing degree.

Many shots also required roto to limit areas of the reflective surfaces such as the floor where the greenscreen would have keyed cleanly even though the surface was opaque. We roto’d those areas & used the key to comp in an appropriate reflection.

The gold shots depended on solid roto for their success. The color correction shots had the objects roto’d from the scene and the 3D gold shots required articulate roto for the people holding the objects.

Which elements and assets did you received from production?
We received all of our NYC background tiles – HDRI stills and moving video footage – from Production. Other CG assets like the Ferrari model and the CG NYC were shared amongst the vendors.

Can you tell us more about the creation of this huge and well known background?
In the earliest days of the project, Peter Marin, the Method VFX supervisor who started the job in New York, and two senior composite artists, Aleksander Djordjevic and David Piombino, teamed up to construct a NYC skyline “bubble” in Nuke. The bubble was a sphere mapped with a stitched panorama of NYC skyline stills. Our matte painter, Amy Paskow, cleaned up the stitches and enhanced sky detail for particular views. Also, the BG plates were shot with trees in full summer greenery, but since the movie takes place around Thanksgiving, we replaced the trees with more autumnal versions. In order to align the perspective from inside the virtual apartment, we used the Google maps satellite view of Manhattan to place the bubble in relation to the apartment geometry.

After our preview delivery in May, we wanted to go for a more precise skyline placement. CG artist, Justin Maynard lightened-up a CG model of NYC so that it could be loaded into Nuke as a placement guide for the compositors. The CG NYC allowed for an exact placement of the BG bubble to match the view from the real-life penthouse apartment at the southwest corner of Central Park.

The Time Warner towers are huge reflective buildings which dominate the southwest views from Shaw’s apartment. We felt they needed special treatment for more realistic reflections and parallax. The panoramic still images of those buildings had frozen reflections and, as stills, would have no parallax on moving camera shots. Since all the apartment shots were shot with a Steadycam, we knew that it would pay off to add live reflections and some levels of parallax. We modeled the Time Warner towers and projected cleaned-up (reflection-less) textures onto that geometry. Then we positioned cards at the surfaces of the Time Warner faces to reflect the bubble environment. The reflected bubble on standalone geometry created a great sense of movement and life for what would have otherwise been much more dull and static.

As the show progressed, we placed several additional front row buildings on cards to further enhance feelings of parallax.

Due to the numbers of windows shots, did you create some automatic setups and scripts?
We created and maintained the BG bubble for use in all apartment shots. Eventually, we devised individual panoramas for each of the different scenes – varying the skies & lighting of the city for variation to match the different times of day for each sequence.

Did you share assets with other vendors?
We did share with Gravity since some of our sequences were so similar. They gave us their Ferrari model so we could use it as the basis for exterior reflections which we used to replace set reflections on the live-action Ferrari in our shots.

Can you tell us more about the gold enhancements shots?
Talking about the gold shots will be a spoiler for anyone who hasn’t seen the movie. At the finale of the film, the defrauded workers each get a piece of Shaw’s solid gold Ferrari. The props used for those shots had various gold treatments – some of which photographed better than others. Once the filmmakers saw the footage in an early DI session, they realized that the gold props weren’t as convincingly gold as they could be.

Starting with the simplest-solution-is-best mindset, we attempted 2D color corrections contained by articulate roto. For certain objects, that method worked very well – we gave the dull props a much more lustrous look.

Other objects proved to be tougher cases and didn’t respond as well to the 2D route. On those shots, involving the steering wheel and the grille, we resorted to a CG replacement of the props. The actual props were delivered to us so we could model them photogrammetrically. Justin Maynard modeled the pieces and matchmoved them to the actors’ action in the shots. Jaemin Lee textured and lit the pieces to match the plates. Flame artist Chris Hunt composited the objects into place – painstakingly adjusting the lighting balances and textural feel to give the objects the same reality as the props in the footage – only more golden…

What was the biggest challenge on this project and how did you achieve it?
The biggest challenge on the show was keeping the fine edge detail of the FG plates while compositing them onto bright-to-blown-out backgrounds. We also had to balance the window reflections to always give a sense that the glass surface was present.

Andy Jones wrote a tool for Nuke which helped us get smooth, clean keys for the greenscreens and still maintain a high level of fidelity to the original plates for extra-fine edge detail. Retaining that detail made the difference in selling the shots.

Was there a shot or a sequence that prevented you from sleep?
Aside from the number of shots & the time we had available (i.e., the standard problems!), David Piombino had a tricky shot featuring the tail-light of the Ferrari. The anisotropic reflections in the glass had to have the exact right refractions of the skyline bubble in order to be convincing…it was touch and go…but we all had faith that he could pull it off.

What do you keep from this experience?
This job was my first job at Method Studios after the merger with my former company CIS Hollywood. I was away from my home base in LA and working with a group at Method NY whom I had never met before. Luckily, in addition to being warm and welcoming, they are an excellent team of resourceful artists. We got the job done and had some fun along the way.

How long have you worked on this film?
Method Studios started work on the film in January 2011 and completed work in early October.

How many shots have you done?
Method Studios worked on 160+ shots, of which 138 are in the final edit.

What was the size of your team?
We had five artists working on CG and matchmove; five composite artists for the main show (and four-to-five more during crunch times); two roto artists; two paint artists; one matte painter and a production team of three.

What is your next project?
I moved directly onto Garry Marshall’s NEW YEAR’S EVE which is delivering now for an early December release.

A big thanks for your time.

// WANT TO KNOW MORE?

- Method Studios: Official website of Method Studios.

THE THING: Jesper Kjolsrud – VFX Supervisor – Image Engine

the area — Sat, 11 Feb 2012 19:00:00 UTC

What is your background?

I got in to computer graphics through a university course in the north of Sweden. It was unique for using industry standard hardware and software, something that was very hard to get experience with at the time unless you already worked in a facility. That led to an animation job in Gothenburg and after a few months a position at MPC in London through one of the guest lectures from the course, Paul Franklin. One of the first jobs I worked on was THE BORROWERS, which was supervised by Peter Chiang and when he a year and a half later approached some of us to set up a new facility to do his latest job PITCH BLACK I moved on to what became Double Negative. I stayed there for over 10 years until my family and I felt like a break from London, which ultimately led me to Image Engine in Vancouver.

How did Image Engine get involved on this show?

We got a call from the director, Matthijs (Van Heijningen), and the VFX producer Petra (Holtorf) towards the end of 2009. They had seen and liked our work on DISTRICT 9 and thought we’d be a good fit for the job.

What was the Director’s approach to visual effects?

Matthijs is a very experienced commercials director who’s done big projects with complicated VFX so having that knowledge we tried to fit the job around him. In commercials you tend to work with artists directly so we tried to do the same without bothering Matthijs with the processes and pipelines of a project of this scale.

What was the real size of the exterior set of the base?

The base itself was built as a 1:1 scale set based on the production design with only certain areas like the back of the building missing as it was never featured. It was built in a quarry in Toronto against a slope that would fit in to the geometry of the mountain location in BC.

What was your feeling to give your contribution to the great and scary creatures that Rob Bottin and John Carpenter have created?

One of my favorite parts in Carpenter’s version is when the severed head grows legs. I for one wanted to see more of that. And the gruesome double-headed monster the Americans find. The way it was lit and filmed worked great but it was dead. I would have loved to have seen it move. With the techniques of the early 80′s that was extremely difficult to pull off but that’s something we can do quite well today.

Was there a shot or a sequence that prevented you from sleep?

If there’s one thing I’ve learned so far, you can’t let work interfere with sleep. Sleep is far too nice for that.

What do you keep from this experience?

Although the project was a huge challenge I’m very pleased how well it turned out. There are always curve balls being thrown at you. All shows have their challenges but now I’m more confident than ever that we can handle it.

How long have you worked on this film?

I started prepping just before Christmas 2009 and we finished the show at the end of May this year so about a year and a half.

How many shots have you done?

We ended up doing around 550 shots.

What was the size of your team?

I think we peaked at around 100 people.

What is your next project?

I’m currently working on another Universal project called R.I.P.D. It’s another creature show so in a sense we’re carrying on with what we were doing on ‘The Thing’.

// Neil Eskuri – Digital Effects Supervisor

How was the collaboration with director Matthijs van Heijningen Jr.?

Collaboration with the Director was pretty good. Matthijs had a lot of great ideas and disgusting images that he would send us for the look of the different creatures. Their maw, or mouth, what he thought the feet and toe nails might be. The gruesome inside tentacles from swarming snakes and worms. He sent close-up shots of insect feet and bird feet along with nature footage of Cuttlefish and squid hunting their prey. He often said he wanted the creatures to be ‘horrifically beautiful’.

How did you recreate the helicopter?

There was a ‘shell’ of a helicopter on set so we had good reference images from those shots. The only time we used the actual set copter was when it was on the ground and the interiors shots. Whenever it was flying, it was our model.

By photographing the actual model, we used those images to build and texture the CG model. Then the lighting was generated from HDRI images taken on set.

The look of the rotors went through several variations. Because of the rotational speed of the rotors and the shutter speed of the camera, the look can be different. Finding that right mix with motion blur took a lot of trial and error.

Can you tell us more about the impressive transformations shots and especially the one with the woman?

Juliette’s transformation was always a question mark because, although we knew what she was going to look like once she was fully transformed, how she got there was unknown. We had our concept artist, Juliana Kolakis, go through several different ideas and stages of how her chest would change. Matthijs constantly asked what we would see, how much of her human-ness remained? How much alien is pushing through the skin?

A ‘beat sheet’ was created based on the cut, which would explain what would be seen in each shot. This changed as the cut changed. Matthijs then suggested that the alien should already be pushing through the chest before her sweater falls off.

Since tentacles are a big part of the alien look, this gave us the idea of ball of snakes trying to push through the skin, which is what you see in the first shot. Then as the teeth of the alien mouth rip the skin, we see the inside tentacles push through while Juliette’s head is forced towards her back. Her proportions constantly made it difficult for animation since we now had this huge mouth and claw on her thin legs, how was she going to move and look powerful? Like most things, a lot of trial and error and versions were made to bring her to life.

Can you tell us more about the huge ice cavern and the spaceship?

The Spaceship was to be 1000 ft. in diameter and over 60 ft. high. A good portion was buried and covered in ice and snow so you wouldn’t see it all, but still have an idea of how huge it was. After we got the original model, certain design elements had to change once shots were being laid out to convey the vast size of the ship and the cave and still keep the framing Matthijs wanted for the shots.

Since the actors were shot on small mock-ups of the ship, when the designs began to change we quickly knew that very little, if any, of the set pieces could be utilized in the shots. We only kept the very edge of the ship and the hatch from the original photography.

Like most elements, the Ice Cave went through a lot of different versions. There were several meetings to discuss how long the cave had been there, the evolution of the cave over the centuries, if ice columns had been created from the ceiling to the ship, and a schematic was created to show the types of ice and snow in different areas of the cave. Again, armed with hundreds of images of different types of ice caverns, several concept art pieces were created to give the sense of a vast expansive ice cave and sub-caves housing the alien spaceship. It was probably the most difficult look to achieve.

Was there a shot or a sequence that prevented you from sleep?

They all kept me awake at night. Often, I’d wake up at 2 in the morning with some part of the project on my mind. Each had their own trials and tribulations, but the louvre sequence to blowing up the alien was the part that kept me awake the most.

Those sequences came at the end of production with very few weeks left. We still didn’t have the final designs for the ice cave or the spaceship and the Sanders alien was going through a complete overhaul. We knew it was going to be a tough finish, but with the talent of the crew and the flexibility of the Image Engine pipeline, we were able to deliver.

// Fred Chapman – Lead Character Technical Director

Can you tell us more about the design of the creatures?

The initial creature designs were provided to us from production. These designs go through several stages of approval before they reach us but it is still possible for us to suggest tweaks.

Most creature designers work mainly on aesthetics, they want to design something that looks cool and original, however what works for a static design won’t always work when it starts moving. We need to imagine what’s going on under the skin, where are the muscles, bones, tendons, how are the joints structured to give the range of motion required and also convey sufficient strength. In order to convey a sense of realism in a creature it needs to appear to interact with real world forces so each body part needs to appear to be able to withstand the stresses it would be under were it really moving in that environment. For example we requested bulkier muscles in the front limbs and stronger looking shoulder joints for the Edvard-Adam Thing so it would more able to support the mass of the torso, trying to keep the essence of the design and add just enough reality to make it work in the scene.

Once we’ve modeled, rigged and animated a creatures then the clients will get to see it move for the first time, that’s when the next round of design tweak requests start, these requested changes can sometimes go right through to the week of final delivery. At least one of the creatures was totally unrecognizable by the end of the show compared to what they had in mind on the day they shot the sequence.

What was your feeling to give your contribution to the great and scary creatures that Rob Bottin and John Carpenter have created?

The creature effects in the 1982 film were amazing for their time; I’ve always been a huge fan. It’s such a great honor and challenge to be involved in creating the modern equivalent and I’m really proud of what we achieved. Audiences are much less forgiving now than they were in the 80s so the approach used then would simply not have worked this time around. That said, we were very conscious of remaining faithful to the style and character of the creatures in John Carpenters film so I really hope that shows in our work.

Clients no longer have to accept that whatever they get in camera on the day of the shoot is the final version they’re forced to use in the film. They now have the option to come to us afterwards and describe what they really want or how they’d like to take it further. As long as the time and resources are available we can keep working on it until we achieve the result they’re happy with. That’s why in the final release of the film there are almost no practical creatures visible. I take it as a back-handed complement when I read reviews saying the CG creatures were not as good as the practical ones because I know the ones they thought were practical were ours too.

Can you tell us more about their creation and the challenges you have to achieve with them?

There are a number of huge challenges we faced on this show. The visceral, organic nature of the creatures is always a difficult look to re-create. From a rigging point of view we work from the inside out, even though you never see it the viewer has to get the sense that there is a complex internal anatomy to the creature. We don’t have the time or resources to try and recreate all of that underlying anatomy so much of what we do is about understanding the complexity and trying to mimic it as efficiently and simply as we can.

The rigging for those creatures should have been difficult. Can you tell us more about it?

For me the trick to rigging for VFX is about keeping the node graph as clean as possible. However good your initial planning, things will always change so you need to keep the rigs adaptable right through the show. There are 3 main ways we achieved this, by making use of custom nodes to keep the number of nodes and connections low, using a modular rigging system for consistency across each asset and not requiring too much from any single rig.

Most things we need to do in rigging can be achieved using existing nodes and a bit of ingenuity in how to use them. To create complex behavior often requires the layering up of one bit of rigging on top of another to create the final solution. This is both inefficient computationally and messy when you need to debug something not working right or make requested changes. In these cases I like to think about the data flow, what are the inputs an animator needs to have control over and what are the final outcomes that the rig needs to achieve. Then I work out the most efficient calculation to get from A to B and we create a single node to process this calculation.

Our modular rigging system, named “riglets” allows us to position pivots and controls in a template and connect up a rig from its component parts, a torso, clavicle, arm, hand, etc. As soon as we have a stable solution for how we want a body part to behave, we make it a riglet and the riggers should never have to think about that again, freeing them up to spend more time on making the deformations look better. It also allows us to mix and match body parts, so we can put a tentacle coming out of a chest or an alien foot on the end of an arm, etc.

A general rule I like to work to is to build rigs which are capable of doing around 70-80% of what the asset needs to do. It’s very common when you start building a rig to think about the extremes of what will be required and to try and to factor everything into a single solution. This can result in an overly complex and slow rig. I prefer to look at what are the core things a rig has to do in ever shot and build the rig to do that. If in a single shot it needs to do something different then we treat that as a special case and may build a custom rig asset or just fix it in the shot.

The most critical thing about any rig is how the animator feels about using it. Ultimately rigs are just tools for the animators which puts the rigger in the role of servicing the requirements of the animators. We need to strike the right balance between speed and detail, sometimes having different levels of detail of the same rig for use at different stages in the shot development. Its a constant challenge to stay one step ahead of what they may require and ensure that they are able to achieve what they need to as smoothly and efficiently as we can.

Can you tell us more about the impressive transformations shots and especially the one with the woman?

There is no set process to achieving the transformations, each one was handled as a one-off and in most cases required a number of different effects to be layered up to create the final shot. It was also an iterative process in narrowing down what the clients wanted. Initially we’re given quite a vague concept like “these two heads merge together” or “the skin on her chest rips open to reveal a giant mouth”. Exactly how we achieve that depends on the very specific details. Does it happen fast or slow? Does it start at the top and work down, or in the middle and work out? What happens to the extra skin? Does it shrink back, fall off, dissolve away? Until we’ve answered all these questions we don’t know for certain which parts we’ll do in rigging, animation, cloth, fx or comp. These transformation shots are often the most collaborative between the different disciplines, each having to adapt to what works best for the others.

How did you create the terrifying arm that attack one of the character?

The arm creatures were actually some of the easiest rigs to build and maintain. They’re good examples of why a modular rigging system was a good approach for this show. Through the show the redesigns meant the number of legs changed, human fingers were merged together and the mouth areas changed significantly. Each time we could easily go back to our modular template, make some adjustments and rebuild. One of the reasons these rigs were so simple was due to the proprietary multi-bone ik and spline ik nodes we created. Compared to standard out-of-the-box solutions these nodes have so many more features for controlling the exact behavior of each bone and shaping of the leg. For example in a segmented 5-bone leg the animators can control the weighting of each bone to adjust how much automatic movement goes into each bone with additional animated override. It also took care of IK/FK blending. All of the control values feed into a single node which outputs transform values directly into a single joint chain. This single proprietary node gives complete control and flexibility with the absolute minimum number of nodes and connections. The same is true for our IK spline tool, which gives us a huge amount of extra control over twisting, aiming and scaling of joints along the chain. We had a multi-bone setup for each leg and the claws around the mouth and our ik splines for the spine, mouth sphincters and muscles. That’s pretty much the whole rig.

What are your software and pipeline at Image Engine?

For rigging we use Maya as our main package with a significant amount of proprietary software on top. That’s everything from our cross-platform asset management tool to our custom constraint node.

What do you keep from this experience?

That I’m so lucky to be working on awesome films, for an awesome company, with some awesome colleagues. Image Engine is small enough that we can really be responsive to a client’s needs, we can make bold decisions and be really creative and efficient. At the same time we’re big enough that we have some world-class talent, especially in the RnD team who work closely with the artists to create tools and pipelines that enable us to create top quality work.

We developed some great tools and techniques on THE THING that we will definitely be using again but we’re not complacent, there’s always plenty of room for improvement.

A big thanks for your time.

// WANT TO KNOW MORE?

- Image Engine: Dedicated page about THE THING on Image Engine website.

IMMORTALS: Jay Randall – VFX Supervisor & Founder – BarXseven

the area — Tue, 24 Jan 2012 19:00:00 UTC

What is your background?
I have a degree in psychology from the University of Ottawa, then I went to Vancouver Film School. I launched BarXseven in 2005 after working as a character animator and FX TD on many film and TV projects such as THE FIFTH ELEMENT, PEARL HARBOR, GODZILLA, and TITANIC.

How did BarXseven got involved on this show?
Raymond Gieringer and I have worked together in the past. He was brought onto IMMORTALS by Relativity as VFX supervisor and then hired me to supervise the VFX for 2nd unit. BarXseven then became one of the main VFX vendors.

How was the collaboration with director Tarsem?
Tarsem has great ideas and is a very energetic person. It was a great experience to work with him and watch him create. When it comes to VFX, he is a veteren director and understands our needs and was very approachable and open to input.

How was the collaboration with Production VFX Supervisor Raymond Gieringer?
Raymond and I have worked on a few movies together and always collaborate very well. He really knows his stuff and has great creative vision. Raymond has put his trust in BarXseven on many occasions and we always make sure to repay him by giving him excellent work.

What did you do on this show?
We did over 100 shots but the ones that stand out occur when Aries comes down from Olympus and smashes heads. The heads explode in slow motion as Aries clobbers the bad guys with his war hammer. Another fun sequence of shots that we did occurs when Zeus lands on earth and punishes Aries by whipping him with a fire whip, also in slow motion.

How was this big impressive continuous shot filmed?
It looks continuous on the screen but it was not filmed continuously. It took 3 days to shoot the sequence in many small, carefully planned steps. We shot Aries first, hitting targets on green screen from different camera angles. We then switched to the phantom camera and shot the Heraklions at 500fps from the same angles.

Can you tell us how you approached this huge shot?
We spent a lot of time previzing the shots and figuring out the best approach. We broke the sequence into small details and shot each element separately. We had a very precise plan to create a smooth flowing sequence that looks continuous.

Did you create previs or an animatic for the choreography of the fight?
I worked with Jean Frenette the fight coordinator, along with stuntman Alain Moussi to choreograph the sequence. We motion captured Alain’s movements and used that as a basis for the previz. We built the entire sequence in Maya before we shot it.

How did you manage the super slow-motion aspect of the shot?
We shot Aries with a Genesis camera at 48fps and then shot the Heraklions at 500fps on a Phantom camera. The special effects guys built us some dummy heads filled hamburger and fake blood that we exploded while filming with the phantom for reference. We simulated the heads exploding based on that footage.

Did you create digi-doubles especially the soldier who crosses the entire room in the air?
No digi-doubles were used. The soldier that flew through the air was stuntman Max Savarias rigged up with wires. He is a big guy and he had to fly all the way across the room and hit a specific target, so it was a challenging shot.

How did you create the heads exploding?
We spent a lot of time studying slow motion liquid explosions. We shot some reference of heads exploding on set and also watched every clip we could find on youtube.

We tracked the stuntmen in each shot and put cg helmets on them. We then used those helmets as CG fluid containers for the simulations. We used Maya to create the exploding helmets and also for brain and bone chunks. We used Realflow for the viscous liquid and then topped it off with spray that was created in Houdini. It was a tricky thing to get all the simulation to work together. Steve Elphick was the main guy for the simulations. He did the Realfow and Maya work. Mike Lyndon did the Houdini particles, Charles LeGuen did the lighting and rendering and Rob Rossello comped the shots in Nuke. There is one guy that gets split down the middle. Matt Evans did that guy.

What was the biggest challenge on this project and how did you achieve it?
Once we had the simulations working and we were happy with our look we had to get them to render. The scenes and particle caches were so huge that it was taking forever to render all the layers, when it rendered at all. Charles LeGuen worked tirelessly to optimize the scenes and get them to render without losing any quality.

Was there a shot or a sequence that prevented you from sleep?
All of it was quite challenging and there was always something that needed sorting but the shots that were the most difficult were the fire whip shots. Mitch Deoudes did the fire using Maya liquids. There is one shot that was about 700 frames long and shot at 500fps. To get the fire to look great throughout the entire shot was very difficult. Mitch worked through the entire Christmas break, even on Christmas day on that one. I think he wanted to go in on New Years day as well but his activities the night before made that impossible.

How many shots did you do?
BarXseven did a little over 100 shots.

What is your next project?
We are currently working on Tarsem’s Snow White movie, MIRROR, MIRROR.

A big thanks for your time.

// WANT TO KNOW MORE?

-BarXseven: Dedicated Page about IMMORTALS on BarXseven website.

SPACE BATTLESHIP YAMATO: Takashi Yamazaki – Director & VFX Supervisor – Shirogumi

the area — Sun, 18 Dec 2011 19:00:00 UTC

What is your background?
I have been strongly interested in Monster films since childhood, especially, backstage work had drawn my attention, and I had kept thinking how I wish I could make Monster movies when I am older. When I was in a junior high school, I watched two revolutionary films, STAR WARS and CLOSE ENCOUNTERS OF THE THIRD KIND, and my inclinations toward SFX films got stronger. I may add that there are so many people who have been influenced by these two fantastic movies, and I think that many people who are involved in VFX and CG are close to my age.
A few years later, I watched the making of STAR WARS EPISODE V: THE EMPIRE STRIKES BACK on television and made up my mind to have a job like this, so I started to widen my knowledge and to improve my skills by obtaining technical books.

In 1986, I entered Shirogumi, a Japanese VFX production company, as a miniature maker. Even though I had worked as a VFX designer on many Films, Commercials, and Event footages, I realized that there was almost no opportunity in Japan to find a VFX job that I wanted to do when I was younger. Then, I decided to create new demand by myself.
I wrote a script with a lot of aliens, spaceships, robots, and time traveling, which is called JUVENILE. This script caught the attention of Japanese production company, Robot, and I made my debut as a film director. After some films, I won several prizes with ALWAYS: SUNSET ON THIRD STREET.

Can you tell us your approach on this space opera movie?
At first, screen writer Shimako Sato participated in this project. I have worked with her so many times, I told her that I would like to help this film in any position because of I was a big fan of the original animation of SPACE BATTLESHIP YAMATO. It will be my pleasure if they use my ideas in any scene.

One day, I got a call that the production crew was looking for a director and they offered me through my experiences of film JUVENILE, RETURNER and DEMON WARRIOR 3.

To be honest, I hesitated to take this position as a director because I was not ready to direct such a dream project, however, I realized that it was a chance to make my childhood dream come true, so I undertook the position with honor.

How did you create the shot that start from the eye of Yuki Mori and stops in a middle of a huge space battle?
Shot of Yuki Mori’s eye was shoot by macro lens with DSLR Canon 5D Mark II. This scene starts from close-up of her face to full body, I took a track-back shot by crane camera with a six axes motion ride with actor. Six axes ride makes reality in shot. We switched CGI actors and aircrafts when the crane stops in the final position.

How did you filmed the shots showing pilots in their cockpit?
We shot actors on a six axes motion ride set. Motion ride is a part of the set, we created full body by CGI. It was difficult to composite glass on cockpit by complex reflections, we shot with cockpit frames and replaced by CGI.

Did you create previz for the choreography of the space battle?
Pre-Visualization (previz) was a important for such as VFX film.
For actors, I let them imagine the background of the scene. Previz helps them recognize around the shot.
For a stage team, They need to design the stage size of the scene, miniature team also design and create props from camera position in previz.
For me, I need to find a rhythm and tempo for editing film.
For editing, editor needed previz to fill for un-finished VFX shot.

Therefore, I started to work on previz a few month before shooting, and the CG team also created previz for own VFX scenes.

Can you tell us more about the creation of the huge set extension? How did you create them?
It was created by an ordinary way. The camera tracked the minimum sets in front of a green screen backdrop with 2d3 Boujou match move software and we composited with the CGI set.
But there is a problem on the stage size. The actor couldn’t go through left side in the frame. We switched him with digital actor when passing the another actor on the middle.

How did you create the impressive shot in which the Yamato is going out the ground?
This scene is the prominent to describe film of SPACE BATTLESHIP YAMATO, I thought this is a very special. However, It was very hard task to for my team who did not have enough effect skills for braking huge objects, smokes. We started with a research for software for braking objects.
After the research, we decided to use RayFire plug-in. but RayFire plug-in is made for 3ds Max and our team was using Maya. We start to learn basic operation from 3ds Max team’s. We were able to use RayFire on 3ds Max exporting FBX from Maya. Finally, we could finish this sequence by the deadline. So, we used 3ds Max as an expensive plug-in for effects.

Can you explain to us the creation of the shockwave and explosion that destroy the enemy armada?
For the explosion of Gamilas carrier, I told my broad intention to Kei Yoneoka, the CG artist whoes specialty is in visual effect. The details of the intention was; “The Engine of Gamilas forms micro-black hole if it gets damaged. At the moment, everything around it is drawn into, but when it goes to the critical point, the engine of Gamilas triggers a large explosion.”
I left everything else to Yoneoka after I expressed the ideal implosion and the timing of release by using onomatopoeia effectively. For more specific information, please go to his website.

SPACE BATTLESHIP YAMATO mother ship destruction makingA from Kei Yoneoka on Vimeo.

Kei Yoneoka: “This shot can be divided mainly mother ship destruction by black hole and huge explosion. Most of them are made with 3dsMAX. I put keyframe animation on the sphere object to control of the timing that black hole begins to occur and speed of rotation. Then I used FumeFX’s velocity information which emit from sphere object to make particle an organic motion. After that I rendered those tens of millions particles by using Krakatoa. When I made a center core of black hole, I used Realflow with computing vorticity then I extracted that information and rendered by using Krakatoa. The destruction of mother ship was done by Thinking Particles.
The huge explosion after shrinking of the black hole was also put keyframe animation on the sphere object to control of the expanding timing. Then I used FumeFX’s velocity information which comes from sphere object to make particle an organic motion. I put a plant or microorganism’s picture to the tens of millions particles as a texture then rendered with Krakatoa. I sculpted emitter object having bumpy shape with modo in advance in order to make an organic style.”

How did you create all those pyro elements such as the missile trails, lasers and explosions?
We mostly used N particle of Maya for the pyro elements. I combined a lot of shooting footage on explosions.

What was the real size of the sequence with the heroes on a truck?
It was very tiny studio. There is a only truck in front of green screen. We shot a crane camera to make it looks like following a moving truck. And we covered CGI elements around truck.

How did you create all the huge environment around them?
We created the strange rocks using Pixologic Zbrush. We also put matte paintings and photos of miniature for background.

Through the scene, Actors performed arriving in the enemy’s territory and fight with Gamilas fighters, and arrived to innermost place and escape from there. But we shot only truck and some rocks on green screen, we had to create hole backgrounds digitally.

This whole scene was quite difficult challenge for us to create many kinds of backgrounds, but we are confident that the audiences did not realize how little the actual stage size. This is VFX.

Can you tell us the creation of the particles going off Yuki Mori?
We used Maya nCloth for them. We just processed the particle, but I was particular to blew jacket and helmet by the shock of the particles.

How did you create the beautiful final explosion?
The final explosion was also made by a CG artist, Kei Yoneoka whose speciality is in visual effect.
The concept for this scene was to re-create the sad, but beautiful atmosphere of what one of the original piece FAREWELL TO SPACE BATTLESHIP YAMATO depicts and to make it 2010 version. Yoneoka created mysterious and beautiful explosion, which deserved to cap this film, by incorporating marvelous gaseous nebula-like image in an aftereffect of the blast, which looks like supernovas.

SPACE BATTLESHIP YAMATO Final Explosion making from Kei Yoneoka on Vimeo.

Kei Yoneoka: “In this shot, I also put keyframe animation on the sphere object to control of the expanding timing. Then I used FumeFX’s velocity information which comes from sphere object to make particle an organic motion. For the emitter object of the element like a dark nebula, I used gray scale texture which I modified cloud or white cape pictures to make the particles having a natural distribution. Those tens of millions particles was rendered by Krakatoa.”

Did you create specific tools for this show?
No, we did not invent any tools. I focused on employing what we had efficiently.

What was the biggest challenge on this project?
The biggest challenge was that the Japanese described full-scale cosmic space in a science fiction film. Compared with Hollywood films, the budgets are small (I guess one-tenth or one-twentieth.)
And we did not have enough knowledge. I also never done before there are a lot of challenge.
We have been working hard to remind me that Japanese can produce cosmic space feature film.
I hope this experience helps me with next challenge.
The difference between Hollywood and us is not only the budget.
It will be experience, I believe.

Has there been a shot or a sequence that prevented you from sleeping?
We always prepare the knowledge and skills before start project. But this project was started without enough skills to describe such as special effects. It was unprecedented challenge to produce the significant shot of a theatrical released film as learning basic operation of unfamiliar software.

We spent a lot of time for this scene, so I felt really relieved when we finished the shot.

How long have you worked on this film?
We worked on stage for 3 months and 6 months for the digital work.

What are your softwares and pipeline from Shirogumi?
We used Maya as a main CG software, 3ds Max and RayFire for certain scenes. Nuke for compositing, ZBrush for special modeling, Photoshop for matte paintings and Thinking Particles for explosion effect. We used DPX from DI room of IMAGICA, the post production company.

How many shots have you made and what was the size of your team?
The total number of shots were 436 cut.
There were 34 original staffs, 21 people for support and 12 people for motion capture related work.

What do you keep from this experience?
I realized how difficult it was to produce cosmic space films in Japan.
However, I believe it will refine through this kind of project. So, I would like to keep challenging for the new things.

What is your next project?
One of the next project is called FRIENDS NAKI IN MONONOKE ISLAND.
It is a digital animation film based on an old children’s tale (or a fairy tale), which is created with miniature background and CGI characters technology.

http://www.friends-movie.jp/

The other project is called ALWAYS SUNSET ON THIRD STREET ’64.
This is the third series of ALWAYS. We shot with real 3D camera, so it is a true 3D film It is not 2D to 3D converting film. My idea is taking audiences to the three-dimensional world in 1960’s Tokyo.

http://www.always3.jp/

What are the 4 movies that gave you the passion of cinema?
CLOSE ENCOUNTERS OF THE THIRD KIND
STAR WARS series
SEVEN SAMURAI
IT’S A WONDERFUL LIFE

A big thanks for your time.

// WANT TO KNOW MORE?

- Shirogumi: Dedicated page about SPACE BATTLESHIP YAMATO on Shirogumi’s website.

// SPACE BATTLESHIP YAMATO – TRAILER

JOHNNY ENGLISH REBORN: Rob Duncan – VFX Supervisor – Framestore

the area — Tue, 29 Nov 2011 19:00:00 UTC

What is your background?
I did a graphic design degree at college, but fell into digital effects when it was still in its infancy in the UK. I started off working on a variety of television shows, commercials and music videos, and when the technology became available I moved into feature films.

How did Framestore get involved on this show?
We have a good relationship with Working Title and they couldn’t think of anyone better to hand the work to! This was never going to be a mega-budget production and, without trying to sound like a furniture salesman, they trusted us to deliver quality work at an acceptable price.

How was the collaboration with director Oliver Parker?
It was great – he was always very available and approachable, and took on board my suggestions where appropriate. I would extend that to the producers and also to Rowan Atkinson, who was involved throughout the process. My producer Richard Ollosson and I could have a very open and honest conversation with them about what would be possible within the budget constraints, and we would all move ahead on that basis with an agreed plan of action.

What was his approach towards visual effects?
I’m sure he would be the first to admit that VFX is not really his area of expertise, so he was happy to leave many of the creative decisions to me.
Because this was first and foremost a comedy film, it was important that the VFX supported the story, and didn’t overwhelm or distract from the comedy moments. When I spoke to him after we had finished, he pointed out that he could have easily gone down the cheesy route and not really care too much about the VFX, but decided that having good quality invisible effects would be a better choice, since it would help to reinforce the thriller aspects of the story, leaving the comedy to come from the performances. No-one wanted the laughs to come in the wrong places.

Can you tell us more about the Tibet environment? How did you create it?
I was sent solo to Argentina to obtain material that could be used for digi-matte paintings for the “Himalayas” scene. This was done with a local film crew while the main unit continued in the UK.
Over the course of one day’s shooting, we got some fantastic helicopter footage of the Andes for the establishing shot of the monastery, and when we landed we trekked off into the foothills to get some more controllable tiled plates that we would need for the ground level monastery panoramas. In post, we also had to adapt the original footage to make a winter scene, to sell the idea of the changing seasons.
The monastery courtyard you see in the film is a full-size set built at Ealing Studios which we had to top up only slightly, otherwise our task was to create the view beyond the walls.
There was another scene set in the Himalayas which we had almost finished the VFX for, but it got cut for editorial reasons.

What references and indications did you receive from the director?
Because the Ealing-for-Tibet scenes had been shot relatively early in the schedule, there was time to get a local Argentinian scout to offer up some options and send them to us. We were then able to go through and narrow them down to 2 or 3 possible locations before I set off for South America. The production then left it up to me to get the right stuff – we didn’t need anything too specific to fit the brief, as long as it could pass for the Himalayas and felt remote from civilisation, then everyone would happy.

About the London sequences, how did you create the different panoramas we can see outside the office of MI-7 boss?
The boss’s office was a real location in the City of London, meaning that some of the window views are real. Like a lot of VFX work in this film, we were merely extending, extrapolating or replicating what could be shot for real rather than creating from scratch – I prefer this approach wherever possible since the basis of the scene is real and therefore it is much harder to stray off course when adding in the digital work.
In this case, where we needed to invent the exterior view, it was created from tiled plates filmed from a different vantage point on the same building, from a window cleaner’s cradle suspended at the same height as the main unit footage. Unfortunately we were unable to shoot the plates on the same day, which meant that our lead compositor Bruce Nelson had to completely relight the material to make it match (sorry Bruce!).
The plates were then split up into different layers according to depth so we could create some parallax when the camera was translating inside the office. We built some very basic geometry of the interior so that we could be sure the various camera views had some consistency.
The reason we had to provide an exterior view at all was because we needed a false wall and window out of which the unsuspecting cat could be pushed, meaning it only had 1 metre to fall rather than 30 metres – it certainly cut down on the re-set time and the number of cats used.

What did you do on the roof chase sequence in Hong Kong?
Ultimately, very little. In pre-production this was potentially one of our most difficult sequences, since the buildings chosen to stage the action were due to be under refurbishment while we were shooting, meaning that they were going to be covered in bamboo scaffolding – although this would have been very photogenic, it would have rendered redundant the key action beats written in to the script.

At this time we were preparing to digitally reconstruct all the buildings in a two-block radius in order to remove the scaffolding. On the recce trip, I took thousands of digital stills of the buildings’ facades and other details, so we could use them as textures. However, agreement was reached with the local authorities to remove the bamboo where it would otherwise be in shot, so we were able to film the sequence unencumbered – I cite this as an example of solving a problem in pre-production rather than leaving it until post, when it would have been considerably more expensive to fix and, perhaps just as importantly, wouldn’t have contributed to the storytelling in any meaningful way.
Luckily, the bamboo which remained on the blindside of the building became a story point because it was used for the henchman’s final escape.

Can you tell us more about the helicopter sequence? Which is the balance between real stunts and CG?
When I originally read the script, it seemed to suggest that an extensive amount of CG helicopter shots would be needed to achieve the required action. However the production were nervous about committing to such a large VFX sequence (understandably since it would have eaten up the whole digital budget, and then some) and therefore found a largely practical solution by shooting most of the action on a test track closed to the public. This allowed the pilot to fly very low to the ground and close to cars driven by trained stunt people.

Although this reduced the real-to-CG balance considerably, it didn’t eliminate it entirely, so we still had to create a digital replica. Our CG supervisor John-Peter Li and our lead compositor Adrian Metzelaar did a brilliant job of matching it, using a bespoke side-by-side setup that I had shot in between main unit takes. This enabled us to make it completely indistinguishable from the real helicopter long before we knew which shots would be required for the film, and acted as a proof of concept should there be any scepticism about the believability of the digital vehicle. The shots in the film were approved by the client on version 1. Elsewhere, I was sent up in another helicopter to grab aerial plates for some of the action which takes place inside the cockpit – the actors were filmed safely on the ground against greenscreen first, so that we knew which angles and lens sizes to shoot when airborne.

How did you create the take off of the helicopter and the cutting of the trees?
The takeoff was performed by the real helicopter (this time on a private golf course which had its own helipad, so it was perfectly safe to do so). We had the digital helicopter on standby for this shot, but it was ultimately not required.
Apart from the fact that it would have been too dangerous to attempt, the stunt would have required a huge re-set time if we had tried to chop down the trees in situ. Instead, we shot a plate of the trees standing up to achieve the composition, and then they were laid flat on the ground for the pilot to determine the correct flightpath as he passed safely overhead, causing no destruction.

Some weeks later, the trees were shot upright (one at a time) against bluescreen on the Pinewood backlot, with the ‘slicing’ created by placing explosive charges on the trunks. There was some discussion in the planning stages about using an industrial tree-logging machine, but the slicing would not have been instantaneous and the device itself would have interfered with a lot of the flying debris. In order to avoid obvious repetition, we blew up 10 trees, all at different perspectives to match the original plate layout. They were shot individually so that we could get perfect synchronisation with the helicopter’s path. Even this wasn’t chaotic enough, so we shot some additional debris which would help to fill up the frame with fine leafy material.

About the final sequence, can you tell us the creation of the building on the top of the mountain and his cable car?
This started as an Art Department concept drawing, which we then adapted/developed to make it fit on to the chosen location of l’Aiguille du Midi near Chamonix, France.
In the story, le Bastion is meant to be a Swiss government fortress, inaccessible except by the single cable car which has a key role at the end of the film.
The Production Designer Jim Clay was very keen for the building to have a strong minimalist look, so of course it was impossible to find anything which fit this brief on top of a mountain. Because l’Aiguille du Midi actually has an observatory on the summit (the highest in Europe) serviced by a cable car system, it proved to be a excellent aid in terms of composing the shots – we were mostly able to cover up the real buildings with our digital creation and then erase what little was left.
The full-size cable car (which was used to stage the fight sequence) could not be used in the big establishing shots, because of the camera moves involved, so we built a digital replica with a combination of lidar scans and photographic textures. Whenever we saw the cable car in a wide shot, we stripped out the actors from the real prop and placed them in the digital version.

Can you explain to us in detail how you create the huge environment?
The environment was always going to be a combination of real locations and invented ones, in order to facilitate the journey mapped out in the script. For instance, the summit existed so we didn’t need to create that (apart from the aforementioned le Bastion building), and when Johnny falls out of the cable car this was all shot on the slopes in Megeve, France and Surrey, England.
The major part of the digital build would be confined to the cable car journey, but that brought with it its own challenges since the total distance was quite substantial. During the sequence the camera would be looking in all directions, so we had an environment which potentially covered hundreds of square kilometers.

We knew it would be infeasible to build the whole environment, so we decided to take a modular approach whereby we built a smaller section of the mountain slope which could be bolted together in different configurations to prevent it looking like a repeat. This took care of the close- to mid-ground terrain and beyond that, where parallax was less of an issue, we were able to resort to layered-up digi-matte paintings.

Can you tell us how you get your references for this environment?
As explained before, the real l’Aiguille du Midi is a working observatory, open to the public. We discussed taking the main unit crew up there, but due to time and equipment constraints this was ruled out, so I went as a tourist along with my colleague Dominic Ridley. We spent two sessions (different times of day) capturing the environment as digital stills which could later be worked up into digi-matte paintings and also used as textures for the CG parts of the mountain.
Because we were at such a high vantage point we had spectacular and uninterrupted views of the French Alps which didn’t require much modification to work for the story.
This worked perfectly for tiled panoramas (because we were on solid ground it was completely controllable), but we still needed some moving aerial plates for the establishers of le Bastion, so it was back in the helicopter again.
I also used the opportunity to shoot travelling aerial reference for the cable car’s journey by flying up and down a generic slope on a nearby mountain range – I had mapped out in advance about 10 generic angles that I thought would be good for later reference. Although these passes would be unusable in the final film – it would have been impossible to matchmove them to work with the greenscreen fight footage – they proved invaluable for laying out the sequence and were used in the early temp screenings before any CG backgrounds were available.

How did you create the free fall shots?
In the usual fashion – stringing the actor up on wires, throwing him around, and blowing air at him to make it seem as if he is travelling very fast.
Although this would be a typical studio setup because of the rigging involved, I insisted that we shoot outside so that we got natural daylight, which is almost impossible to recreate on stage and which tends to blow the illusion straight away. Rowan was hanging from a crane arm, so rather than the rigging moving past the camera, the camera moved past the actor to get a sense of movement.

These greenscreen shots were intercut with a stuntman doing the skydive and parachute drop for real in the Alps, which was sometimes made to look higher and more consistent with our digital environment by replacing the live action backgrounds.

Can you tell us more about the big explosion of the cable car?
This shot was a real hybrid – there was the full-size cable car suspended on a crane so that we could get the villain’s performance from the right perspective; there was a third-scale practical miniature built by Mark Holt’s SFX crew rigged with explosives; and then there was the digital replica which was used up to and slightly beyond the explosion.

The miniature was blown up in the Paddock Tank area at Pinewood and was photographed at 125fps to achieve the right scale. Three miniatures had been built in case repeats were needed – we got the best action on take 2 and so used the third one for alternative angles.
Our lead compositor Mark Bakowski then added a CG missile and smoke trail (mixing in some real smoke), more explosion and debris elements from our general library, and then combined everything with a plate shot in the Alps, even making the branches react to the shockwave created by the blast.

What was the biggest challenge on this project and how did you achieve it?
The biggest challenge was definitely the cable car fight sequence, not only because of the scale of the environment that it took place in, but more specifically because of the sheer number of photorealistic CG trees needed to cover it. Our lead TD Dan Canfora took some off-the-shelf tree generating software and refined it, at the same time as developing a tool to quickly and easily populate a mountainside with low res instances for layout purposes, and then replacing them at render time with the full res versions. He had to ensure that we would be able to render thousands of trees, some very close to camera, and he came up with a number of tricks to accomplish this such as simplifying the geometry depending on distance to camera and/or splitting into layers, etc.
We were also aware that the edit was frequently changing, so we had to remain flexible in terms of background layout without foregoing a sense of continuity when the edit was finally locked – it meant that the majority of backgrounds were only rendered in the last 3 weeks of the schedule.

Was there a shot or a sequence that prevented you from sleep?
I slept like a baby – I would often wake up in the middle of the night, screaming. Actually, because we were in constant contact with the cutting room, we were able to anticipate problems before they took root and could come up with a plan to deal with them. I would like to think that as the VFX industry has matured somewhat over the years, it has become easier to deal with a changing and expanding workload during a (more-or-less) normal working day. Hopefully you won’t have to ask that question for much longer!

What do you keep from this experience?
That I really don’t like helicopters.

How long have you worked on this film?
About a year in total, if you include the pre-production period.

How many shots did you do?
We ended up with about 230 shots in the film, plus a few dozen omits. Baseblack, our friends over the road, also did about 60-70 shots.

What was the size of your team?
About 50 I think.

What is your next project?
I’m not at liberty to say.

What are the four movies that gave you the passion for cinema?
It would be unfair to try and narrow it down to only four, since there have been so many that have contributed in different ways. Having said that, STAR WARS hit me at just the right time in my youth and really opened my eyes to what was possible – most of us in the VFX industry owe a great deal to that film.

A big thanks for your time.

// WANT TO KNOW MORE?

- Framestore: Dedicated page about JOHNNY ENGLISH REBORN on Framestore website.