The "Previz" project at ENS Louis-Lumière School

By Christian Guillon

[ English ] [ français ]

Cinema schools from around the globe have their own annual congress, the CILECT. This year, the congress was held at Newport Beach, from 13-16 October 2014, just an hour south of Los Angeles, hosted by Chapman University, and the theme of the conference was “Previsualization”.

What we today call “previsualization” is an evolution of storyboarding, which went through the intermediate stage of “animatics” and which now takes the form of more-or-less sophisticated 3D animations. Some previsualizations can be for the entire movie and be very complete.

In Los Angeles, some studios, like Third Floor or Halon, have become specialized in this area. Besides classic 3D animation, they use all of the most modern tools, like Simulcam, or Performance Capture, on either traditional 3D software, like Maya, or on real-time rendering engines like “Unity”, which are usually used for video games. They develop close ties with directors well before production begins, which can sometimes disturb the work of the creative team (set designers or cinematographers- who find that there have already been decisions made as to certain elements and that have already been visualized on screen.

This was the subject of this week’s debate during the Congress at the very “WASPy” Hyatt Hotel of Newport Beach : "Previz or not previz". The attendees were, as usual, divided between the Ancients and the Moderns. Some intend to resist a practice that eliminates factors of randomness from production, such as improvisation, sudden inspiration, creative synergy of the shooting team, and once again predict the death of cinema.
Others see previsualization as a wonderful tool for internal communication and shooting preparation for the team, but a tool that must be managed in order to avoid it getting out of hand. Very paradoxically, the Europeans weren’t the most reticent ones to this practice, but rather the Americans, despite the fact that Hollywood has witnessed the generalization of this practice over the last few years.

It seems that American cinema schools are “more catholic than the Pope” in their defence of auteur films, which they venerate via the French New Wave, central to their pedagogy. We noticed that academies that are less immersed in “dominant cinema” and much further away from the Hollywood model, in Portugal or in Sweden, for example, have no compunctions about trying out highly original and pertinent pedagogical experiments using 3D previsualization, despite often being limited by very European budgets. On behalf of the Ecole Nationale Supérieure Louis-Lumière, I presented the R&D project “PREVIZ” that I am spearheading at the Ecole, in partnership with a number of manufacturers (Technicolor, Loumasystems, Solidanim, Polymorph) and three university laboratories (INSA, GipsaLab, LIRIS).

We are working on the following step, called “PREVIZ ON SET” : the live, real-time mixture of images being filmed using real actors and real sets with digital images being generated by computer (virtual characters and sets). This practice is becoming generalized in the production of films with special effects, as a director’s aid. It allows the director to see in real time what he used to be able to see only in postproduction : the final composition of all of the elements of the image. But it especially allows the director to act on these elements, to direct them together, in one single stroke.

The same thing goes for technicians, cameramen, cinematographers, set designers, etc. Tools that enable virtual sets to be mixed with live actors filmed against a green backdrop in real time already exist, including in France (e.g. our friends at Studio 20 who presented at the Micro Salon and who are being hosted by Transpmedia).
To go a little bit further, the “Previz” project puts emphasis on (virtual and real) character interaction and on related research projects, such as those on “depth maps”, being developed by Technicolor, which might allow us to eventually get rid of the green backdrops, or the work being done on fusing together various existing methods of real time camera tracking, being carried out by Loumasystems, in order to integrate the data supplied by the encoders installed on their cranes.

I presented the “making of” the first experiments carried out at the Ecole last July, which were still rudimentary. You can already see the students at the controls of the Louma following two little pre-animated virtual cartoonlike characters with the camera as they move about the set constructed for the occasion and interact with two flesh and blood actors.
Following with the viewfinder characters that no one can see on set is an out-of-the-ordinary experience. This type of tool will return freedom and spontaneity to the direction of artificially created scenes. When this type of technology evolves to encompass the cinematographer’s work, he will be able to adjust the virtual light on the set and the real light on the actors, in real time during shooting.

This was far from being the case in July, and Michel Abramowicz, AFC, who we have gotten involved in this adventure, was probably a little bit frustrated by what he saw. He was full of new ideas, but certainly disappointed not to yet be able to see them come to fruition, as are we all. In the “making of”, we can also see the first real-life tests of the depth maps as carried out by Technicolor. In these specific lighting conditions specially adapted to their as-of-yet fragile technology, one can now isolate a real character filmed without a green backdrop and precisely locate him in a complex 3D environment with all the distance information that allows one to place objects and set elements in front of or behind him without masks or masking objects. “Work in progress”, to be continued…

(The image above shows students on Set 1 of the Ecole with the Louma 2 – in the Portfolio below, three screen shots from the “making of”)