Key words

MPC, Moving, Picture, Godzilla, Maleficent, Disney, Framestore, Gravity, Cuarón, United, Passions, football, Pixomondo, Game, Thrones, GTO, Daenerys Targaryen, dragon, MUTO 

Pixomondo on the Throne (Game of)

Pixomondo is an international network of VFX suppliers, with entities in Stuttgart, Frankfurt am Main, Munich, Los Angeles, Baton Rouge, Beijing, Shanghai and Toronto. The Frankfurt studio has been working on visual effects for the Game of Thrones series since season 2.

In this heroic fantasy universe Daenerys Targaryen, a central character, adopts first one then several dragons. "These creatures played parts in three out of ten episodes, in sequences lasting from two to five minutes apiece, for a total of thirty shots", Chris Stenner told us. Differences do exist between visual effects for TV and the cinema, although the Pixomondo animation supervisor sees none concerning creativity. "The only noteworthy differences are the very exacting deadlines which the channels put us under." The result of such tight scheduling meant that the work to be done from division to division tended to pile up.

Concerning concept/design, a dragon was already on screen in season 1, before Pixomondo came in on the series. "The designs had been done by the Blue Bolt studio, so we started from those and modified them, especially since the dragons are getting older and heavier, and their skins equally display these changes."

Pixomondo sought out photographic references for reptiles and batrachians to examine "this kind of oily, scaly look". The 3D modelling was done on ZBrush by an artist who was not only in charge of the rig but also of the animation, to keep a handle on constraints. When the first dragon is just a "baby", it often rides around on the actress’ shoulder. The tracking department thus had to model this part of her body, integrate the 3D model, animate it on the video in keeping with her bodily sway, and then implement the match-moving.

In season 3, the dragon is no longer alone… and all of them have grown considerably. "It was a definitive challenge since they were more on the screen and more visible, so interactions had to be handled, no longer simply with the actors, but also among themselves," Chris Stenner notes. An added difficulty: they’re able to spew fire. "A flush of red light had to appear on their throats to indicate the rising flames." Since the creatures had also learned to fly, a number of tests were done to estimate the bearing capacity for the unfurled wings. The simulation experts needed to completely redo the skin set-up between the wing bones, since the latter had to navigate air streams and not appear too stiff.

The CGEV (re)plays the World Cup

United Passions is a feature film by Frédéric Auburtin which relates the history of the World Football Cup through several past FIFA presidents. Since this film covers scores of years, the production (Leuviah Films and Thelma Films) contacted CGEV (Compagnie Générale des Effets Visuels) to re-create three of the most mythical World Cup stadiums: Centenario in Uruguay (1930), Maracanã in Brazil where the 1950 final was played, and Azteca in Mexico City, location of the 1970 final.

"Today it’s impossible to film in these arenas which do still exist but where the environment and infrastructures have radically changed", Chervin Shafaghi explains. Extensive research was done to uncover as much photography or film footage as possible. Although the live-action footage was shot at Charléty stadium in Paris, a major portion of the stadium scenes were done in CGI.

"The Charléty stadium has a 20,000 capacity, whereas in 1950 Maracanã could accommodate up to 250,000. We therefore had to work on both full CG background extensions and multiple crowd generation." Although the Azteca is concerned by only two or three shots in the film, for the Centenario sequences it required creation of a foreground entirely in Maya-generated CGI, to then be rendered with VRay. The Charléty stadium was extensively filmed so as to yield HDRI images which "allowed us to then rebuild it in CGI, integrate real actors in it, and the crowds in the different stadiums."

Another concern about crowds was the historical period. "We couldn’t just apply generic models, since fashion had changed since then." CGEV decided to use the presence of bit players to set up a photo shoot of each one of them, from different angles, and to do a scan through Kinect. "The only thing we wanted here was to recover the camera positioning and therefore the position of the model in space," Chervin Shafaghi tells us. The Kinect-sourced scans were then re-projected onto basic models made with ZBrush. "The characters’ modelling and texturing only took two days, giving us five male and five female models."

The crowd simulation tool (by the Chinese editor Miarmy) integrated into Maya generated fifty animation clips containing different behaviour patterns. "We thereby exported 500 characters into the software, with random scripts to help mix the movements, clothing and clothes colours."

Weightless Framestore

For Gravity, Alfonso Cuarón’s feature film, Framestore produced all the visual effects for this film which, in the words of Nicolas Scapel, "has no possible real backgrounds, has long sequences staged in a weightless atmosphere and has an affirmed intent to be photorealistic."

This meant that a collaborative process – called the Digital Production Hub – needed to be established very far upstream to link the director, the producer, the art department and the scriptwriters. The film shoot took place in the Shepperton studios near London, whereas pre-production and post-production were done in Soho.

After testing numerous rigs and harnesses for the actors, Framestore decided on robots from KUKA, working with the American company Bot & Dolly (cf. "Emerging Tools" conference).

The previz and production planning were done conjointly with The Third Floor, an American company specialised in pre-visualisation which set up offices in London. "Thanks to the previz, Alfonso Cuarón was able to make his movie iteratively, with the teams in charge of rigging and animation", Nicolas Scapel explains. A phase of ‘techviz’ – aka technical pre-visualisation or tech layout – was required to program the robots’ rigs.

One of the most significant innovations specifically developed for this project was the Lightbox. This is a kind of cube composed of 80 LED panels that emit uniform and constant lighting. Digital images of the environment in which the actors – Sandra Bullock and George Clooney – were to play were projected onto these panels. The latter were fastened onto a specific rig which moved them in a given direction, in order to simplify the acting. The robots, controlled by motion control, then filmed their performance in perfect synchronicity with the projected images.

As Nicolas Scalpel’s said: "If it was impossible have the actors turn with the light, it was the light that needed to turn around them." The same was true for the backgrounds, which changed in relation to the sequences, whereas the actors were nearly motionless.

Overall, the manpower on this film involved 440 people, of whom 280 throughout the entire course of the three-year production.

Maleficent MPC

MPC does not wish their presentation content to be transcribed in our summaries.

MPC re-creates Godzilla and the MUTO

MPC does not wish their presentation content to be transcribed in our summaries.

Drafted by Stéphane Malagnac, Prop'Ose, France

Translated by Sheila Adrian

The Annecy 2014 Conferences Summaries are produced with the support of:

Conferences organised by CITIA

under the editorial direction of René Broca and Christian Jacquemart

Contact: christellerony@citia.org