Laurent Taillefer worked for more than 10 years in visual effects. He worked in France especially at Mikros Image then went to Montreal. He worked for several studios such as New Breed VFX, Vision Globale or Mokko Studio and participated in projects like PIRANHA 3D, MIRROR MIRROR or ASTERIX AND OBELIX: GOD SAVE BRITANNIA.
What is your background?
I have been a CG Supervisor in various Montreal companies for the last 4 years.
Prior to this I was working as an all-rounder and VFX artist, as well as a TD, first in France then in Canada. I’ve been in the field of VFX for 14 years now, and I studied film and cinematography in France
How did Mokko Studio get involved on this show?
We were referred to Louis Morin, VFX Supervisor, to work on the creatures of the show as they knew that we had worked at designing and building the creatures on RIDDICK.
How was the collaboration with director Christophe Gans?
It went very smoothly. We would have bi-weekly reviews with the VFX Supervisor and the director when needed, so feedback was frequent and consistent.
Although Christophe has a clear vision of his film, he is very open to proposals and collaboration. In this regards, we were able to undertake a lot of creative work, both artistically and technically. He was good at providing references, and quite explicit when it came to leading us through design questions. Mokko had worked with him in the past, so there was a strong trust relation, which prevailed throughout the film.
What was his approach about the visual effects?
He is of the generation of filmmakers who understand and integrate visual effects in their creative process, so he is very rational and efficient about it. Sequences were designed with post-production in mind, and he is open to build upon the interaction with our team.
How did you work with Production VFX Supervisor Louis Morin?
Collaboration was very smooth, from the briefs and breakdowns to reviews and approvals. The process was very straight forward, we would get briefs, references for assets and environments, approval on our concept work, on-set surveys, and work through shots.
We mainly would exchange through the bi-weekly review and approval sessions.
Louis also handled all exchanges between various vendors working on shared shots.
What have you done on this show?
The deer hunting sequence, the park and vine tree growing, the cloud vortex, the attacking vines, the transformation and the magical forest opening.
How did you approach the creature work on this show and especially the deer?
It was a standard creature workflow – building upon client-approved references both for morphology and shading, then rigging and animation tests.
The main challenges were the golden look, that needed to be both magical yet photoreal; and rigging that needed to be as lightweight as possible.
Can you explain in details about the deer creation?
We did the look development using Renderman’s Prman, so we had to tweak our in-house fur shader to allow enough control to achieve a realistic golden look, especially for close-ups. Getting the golden look approved was tedious, since the references of golden, brushed animals such as parade horses, tend to look very cg-like naturally, which was something we needed to avoid.
How did you manage its rigging and lighting?
Rigging was also very straight-forward. We used the multi-resolution character pipeline we developed for previous shows, so getting performance and controllability was not an issue. Secondary animation and dynamics were built using simple tools such as pose-space deformers, instead of a fully muscle and skin system. This allowed us to tailor and refine our character effects based on shot needs, while keeping a high performance. Previz of the framing of the deer in the shot made this possible, by narrowing the list of poses and actions used.
To achieve a full control on the fur look, we used a point-based lighting rig specific to the deer, so it was lit in a pass separate from the environment. We would render interactions (shadows, bounces, etc) in another pass. All passes setup were automated, so this was not an overhead during production.
Can you tell us more about the animation for the deer?
The director was very open to suggestions from our Creative Team, lead by Alain Lachance, VFX Supervisor. The rig was fast and flexible enough, as was our hi-res dynamic preview setup, so turnaround was relatively short, allowing for us to experiment a lot.
How did you approach the dramatic sequence of the deer death?
Since the environment was being reconstructed entirely, including the lighting direction, we were able to orient the dramatic escalation freely. We were able to stage the park environment to place the deer in the most relevant positions and scales, even if it meant breaking spatial continuity. This allowed us to get the most out of our animation and acting, and build the drama as the sequence went. From layout on, we would edit the entire sequence upon each new version, so evaluating it as a whole was also instrumental.
Can you tell us more about the impressive long slow-motion shot after his death?
Aside from the amount of rotoscoping needed to isolate the soldiers, we were able to play with all the elements quite freely. We picked up from the editorial mockup, and iteratively refined the choreography with the VFX supervisor and director.
We developed the rose tree asset on other shots – which were locked in time and frame, so when the layout of this shot was getting close to final, there was little overhead to integrate this master element. Once all timings were locked, we could final the plate restoration and interaction vfx.
How did you created the killer vines?
Once the designs were approved, we modeled and shaded in a standard way, making great use of displacement for detail.
The rig was snake-like, so the animators had a set of paths they animated, then animated and twisted the vines along, with a final layer of controls to move the vine around the path.
Can you tell us more about their animation?
The director was very open minded about their behavior, so we worked by proposals and iterations. Our previews were well received, so we were able to operate quite freely with a good deal of trust from the client.
We did stress tests, to define how many vines we could reasonably render per shot (quite a lot!), and from there on, we gave the animation supervisor the ability to populate shots, so the number of vines could be adjusted to meet the shot’s narrative.
The key element in these animations was the crowd-like behavior, and how they would react and pose as a whole, taking specific silhouettes (such as a grasping hand) on key frames, to suggest the evil powers at work and their intentions.
Can you explain in details about your work on the environments?
Since we use prman, we rely heavily on our rib-archive pipeline. Once the main layout has been set, with all cameras tracked and proxies in place, we turn all hero models into ribs, which are updated upon each model, texture and shader release. This allows us to keep our master environments (the park, the forest) in synch at all times. All ribs have a proxy geometry, allowing it to be visible and manipulatable simply, as well as to be rendered for animation previews.
We pushed this method to a higher level of detail for the park’s bushes, going from very high-resolution items (modeled down the leafs) to background resolutions.
Since we instance the ribs, the system allows for heavy geometry rendering.
We automated, upon the release of the assets, the baking of all possible data (occlusion, color bleeding, etc), so it didn’t need to be recalculated upon rendering the shots, even if the items were moved.
For the forest, we also used Maya particle systems to carry the brambles. The base assets for vegetation were modeled with Speed Tree.
What was the real size of the sets for your sequences?
Since we had creatures and dynamics, we opted for a relatively small scale (1unit per cm). The park existed in real scale, although we moved the camera from shot to shot to extend the runway and increase the hunt’s drama.
The forest was broken into 3 stages: the clearing with the opening fir trees; the run through the forest, and the underground passage. The forest itself was a static set through which the camera would run, with trees placed differently on each shot.
What do you keep from this experience?
The interaction with the VFX supervisor and the director were amongst the most constructive I’ve experienced.
How long have you worked on this film?
A rough 8 months.
How many shots have you done?
142 shots.
What was the size of your team?
50.
What is your next project?
Shhh… (laughs).
The team is working really hard on a few blockbusters right now.
What are the four movies that gave you the passion for cinema?
Seeing THE RETURN OF THE JEDI and TRON as a kid got me into FX, and David Lynch’s ERASERHEAD / Wong-Kar Wai’s CHUNKING EXPRESS into cinema.
A big thanks for your time.
// WANT TO KNOW MORE?
– Mokko Studio: Official website of Mokko Studio.
© Vincent Frei – The Art of VFX – 2014
