Behind the scenes of The Garfield Movie with DNEG and Ziva

The Garfield Movie was produced by Alcon Entertainment and DNEG CEO Namit Malhotra’s production company, Prime Focus Limited. DNEG Animation was the exclusive animation partner on the movie.

Character performance played a major role in how the film was created, delivering a playful approach that still respected the original comic strips. This can be seen in the extreme poses and use of multi-limbed animations expressed by Garfield and the cast.

Importance was placed on seamlessly marrying the animation style and silhouette recognisable in the comic strips with updated design details, such as realistic fur and fluffy character textures. All these design aspects needed careful consideration during the rigging process, and Ziva helped the animation team deliver the amazing character performances seen in the final movie.

Ziva consists of a number of associated technologies, but the two main components used for The Garfield Movie were Ziva VFX and Ziva Real-Time (RT). These tools were specifically created to animate digital humans and creatures.

Ziva VFX is simulation software that mimics the natural movement of soft tissue, enabling the creation of realistic and stylised characters. On the other hand, Ziva RT utilises machine learning to simplify the process of setting up complex characters for real-time animation, while maintaining their intricate movements and dynamics.

When we’re working in feature animation, character deformations are typically crafted manually as opposed to being simulated using the more traditional visual effects methods. This is a meticulous process that involves much careful direction during character creation.

Using Ziva has enabled us to maintain the characters’ lifelike appearance and physical feel. In contrast, the character artists can focus on the art direction and stay true to the movie’s visual style. At the same time, it allows animators to work with the character in real-time, making the animation process more interactive and efficient.

During the production, our team used Ziva for the complex task of creating rigging for two specific characters: Vic and Roland. Roland in particular presented a unique challenge, as his character is modelled after the Michelin Man, but in the form of a large cat with significantly more wrinkles.

With soft bodies, particularly within Ziva VFX, the manipulation of mesh behavior can be finely tuned through material attributes. This stands in contrast to the conventional methods, such as skin clusters, where reliance on volume joints, blendShapes and other techniques is necessary to define and preserve the volume. The simulator handles the aspects that ground the character in some degree of physical realism, freeing up the riggers and animators
to focus on the emotions of the performance.

Using simulation as a technique for character portrayal has proved to be an effective workflow for creating a more genuine, realistic representation of a character. This method facilitates a more profound exploration of a character’s range of attributes, emotions and responses, which ultimately creates a more captivating and relatable portrayal.

Our approach to the workflow in Ziva differs slightly from the standard practice for digital character work on the VFX side, as we apply simulation techniques to stylised characters. We put the emphasis on maintaining simple forms with distinct silhouettes and clean lines.

Rather than deferring simulation to the final stages of the pipeline, we integrate it at the outset during the asset-building phase. During this stage, we construct 3D models to outline the rigid bone structure and the tissue regions of the body.

At DNEG, we’re excited about harnessing the full potential of Ziva technology. We’re currently in the process of combining Ziva, its tools, and libraries with our existing character effects workflows to set a new standard for DNEG’s creature and character creation work.
Ziva has been working on several groundbreaking innovations that utilise machine learning and GPU accelerations, which have not been available for public release. DNEG will leverage these innovations soon on upcoming productions.

I believe we’re at an exciting juncture where machine learning (ML) and artificial intelligence (AI) are poised to revolutionise various aspects of the animation industry. The extent to which artists and studios embrace these technologies will be pivotal in determining their widespread adoption.

ML and AI technologies have been making significant inroads in rendering and compositing, yielding notable improvements in workflow efficiency and rendering times. Over the next two to five years, machine learning will play an increasingly integral role in the rigging process, not only on the deformer but also in kinematics, rigging specific workflows, and plenty more. This development holds immense promise for artists to enhance the rigging process and push the boundaries of what’s achievable in animation.

Since the very beginning, Ziva has always strived for the best balance between accuracy, quality and speed of character effects simulations. The team has been dedicated to delivering solvers that are physically accurate, anatomically correct, and very fast, which sets them apart from most other available options.

There are a few commercially available software applications that provide partial solutions, while some of the larger VFX studios lean on their in-house, proprietary options that aren’t available to the public. Ziva is the only product of its kind that addresses creature and character effects holistically. Going forward, Ziva will continue to innovate as a proprietary solution under DNEG’s banner.