Tau Films and Ziva Dynamics on Building Creatures from the Inside Out

Tau films ziva Primal

The new adventure-horror movie ‘Primal’ begins by taking viewers into the jungle, where Frank Walsh hunts and traps wild creatures for zoos and collectors. This time, he manages to capture a rare white jaguar. After nearly losing his life in a fight with the creature, his plans to safely sail home are interrupted when a political assassin, held on board the same ship, escapes and releases the dozens of animals Frank is bringing back with him – including the jaguar. Chaos and extreme danger follow.

“We needed the audience to feel as threatened by the animals as they were by the political assassin, so the CG work had to be undetectable. If they didn’t feel the realism of the jaguar’s presence in that opening scene, everything else in the film would fall apart,” said Walt Jones, VFX supervisor for Tau Films, whose team worked on this film.

The first step was to determine the physical stature of this creature – too small and it wouldn’t present enough of a threat and if it were too big, it would risk looking too extreme for audiences to accept. The jaguar asset had to be completely grounded in reality. To do this the Tau Films team aligned the structure of the cat to realistic secondary dynamics – natural tissue flexing and jiggle - created in Ziva Dynamics software.

Tau films ziva Primal4

Anatomy Transfer

To get the process underway before the team started on look development, Tau Films used the Ziva Anatomy Transfer tool to create its base model. They transferred the anatomical geometry from one of Ziva’s character rigs, a lion already prepared for simulation, to their white leopard creature model. Used with Ziva's transfer deformers, the process warped and re-shaped the lion geometry to fit the mesh of the leopard.

Python scripts are used to wrap common zBuilder functionality and transfer the Ziva rigs from parts of a source creature onto a new creature. zBuilder is a python module for character authoring that proceduralises the process of character creation in Ziva. Once zBuilder captures a rig’s characteristics, they can then be used to generate various other characters, simplifying collaboration and making it easier for a single artist to support a large number of assets.

Tau films ziva Primal6

Tau Films then designed two separate versions of the jaguar – one regular, another blood-stained – to give them more flexibility during the fight scenes. As art direction evolved, the jaguars could be adjusted quickly, bringing subtle variations to the shots as the cat moved from treetops to the cargo ship and other sets.

Automating Artists

The next step was getting the jaguar to move realistically. Designed by nature for dense rain forests, jaguars are both muscular and graceful, able to run, jump and climb with ease. Since Ziva VFX is physically based, Tau Films could simulate naturally the way muscles, skin and fat interact with each other as the animal moves, often automating the process for artists. As they worked, automated muscle excitation was applied directly into the model.

Tau films ziva Primal5

Applied alongside the existing animation, Ziva has Line of Action functionality that calculates the distance between two points on a model to identify when and how far muscle excitations should happen. If a point is placed on either side of a joint, when the joint bends and the two points move closer together, the muscle will fire as a result, according to how close together the points move.

James Jacobs, co-founder and CEO of Zive Dynamics has some interesting ideas about how and why photoreal animals are appearing more frequently in television and low-budget productions like ‘Primal’, not created at large well-known VFX facilities with access to scaled up teams and resources but at smaller, dispersed studios like Tau Films. “The automation that physics-based simulations allows has made a huge difference to achieving realism in low-budget projects,” he said.

“Similar to the way raytracing has bought photoreal rendering to smaller studios, real-world physics in software means that as much as 80% of the manual work of secondary dynamics can be automated. It gives materials and elements to animators, riggers and modellers that are much closer to real ones, and can be manipulated in a similar way. Stories can be told that might otherwise never be realised and, far from replacing artists, this kind of automation means that animators and other artists have a better opportunity to really be artistic.”  

Tau films ziva Primal3

Elastic Solids

Ziva makes use of elastic solids, which the developers call ‘tissues’ – tissues are three-dimensional solid deformable objects that return to their original shape. Tissues can be used to simulate biological solids like muscles, fat or skin and also be used for non-biological materials like gelatin, rubber, foam blocks and steel beams. In Ziva, elastic materials are modelled with the finite element method (FEM) and, as explained above, can be made to contract and fire automatically as the creature performs and changes in the rig are detected.

Automation ultimately proved important at Tau, as the team needed to create spider monkeys, tapirs and other animals as well as the jaguar. A multi-character pipeline was developed and spread over the studio’s five locations, with automated stages built-in to reduce manual labour. Because of the extra compute requirements, Tau wanted to use automation wherever possible, not only to gain time for iteration, but also to allow their small team to complete more shots.

Tau films ziva Primal2

Adopting Ziva's method also helped Tau Films save time by leaving behind their previous approach, conventional cloth-based skin simulations that required much more manual labour and iterations per shot.

Walt said, “We were able to simulate across locations, fully synced, while we built animals that we could blend invisibly into the world of the film. It was exciting.” Because they had a solid, realistic foundation of physics plus anatomy, they could take an automated approach more confidently. They had a robust pipeline that gave expected results from a given set of inputs, and was tractable and controllable.  www.zivadynamics.com