Unity acquires Ziva Dynamics – fxguide 1

Unity acquires Ziva Dynamics – fxguide

Spread the love


Unity today announced that Ziva Dynamics has been acquired by Unity. As of today, the entire Ziva team has been brought into Unity.

On the heels of securing the teams and tools at Weta Digital, behind SpeedTree, SyncSketch, Pixyz, and RestAR, Unity has now acquired all of Ziva Dynamics. As part of the announcement, Unity highlighted Ziva ZRT and how the R&D team has been leveraging leading practices from machine learning, deep learning, and biomechanics into facial simulation.

Just in terms of the announcement, Unity seems to be presenting the deal as a digital human play, with a strong emphasis on the contribution Ziva will make to realistic digital humans in Unity, presumably in part a reaction to Epic’s MetaHuman success. Not to take anything away from Ziva – whose Ziva Dynamics is already a leading force in solving and understanding the problems around complex anatomical simulation and especially muscle-driven simulation. The Ziva team has deep expertise and understanding of this complex area coupled with real-time tools and some incredible core technology.

Emma above is powered by state-of-the-art machine learning and is running in real-time in Unity. Her model was trained with over 30 terabytes of unique 4D data using the ZRT Trainer (see below), which enables her to emote over 72,000 trained shapes and achieve entirely novel face poses. This tech base can be used for realistic animation with emotive performances or to power anything that deforms, in any size, and in real-time, on consumer-grade hardware. Unity will be deploying Ziva for humans, creatures, clothes, realistic or stylized, at varying levels of fidelity.

ZRT trainer above used Maya/Wrap3, but the company pointed out at launch that studios could also use Houdini or any other DCC.

Ziva Dynamics are experts in soft-tissue simulation technology which enables creators to generate characters with innate secondary dynamics. The company was founded in early 2015 by Academy Award-winning VFX pioneer James Jacobs, who was formerly at Weta Digital. Jacobs won the 2013 Scientific and Engineering Award Oscar for the development of the Tissue Physically-Based Character Simulation Framework. He is now CEO and co-founder of Ziva in Vancouver. He formed the company with Jernej Barbic, who is the CTO of Ziva. Barbic is an expert on finite element (FEM) analysis, which is the method they use to achieve the really nice volume-preserving, solid look and feel of the muscle system. In Ziva, every muscle is made as a tet mesh, and every muscle is attached to the other muscles and to the bones. You can paint the fiber field and then activate them any way an artist wants. It can be active or passive. Then, the muscles are solved with an implicit FEM solver. There is also an inbuilt cloth solver for fine quality skin detail. There is collision and self-collision resolution, non-linear materials (real tissue is nonlinear), and anisotropic materials. (See more about Ziva’s past work in previous ZIVA fxguide story ).

Ziva VFX, is used to digitally replicate and couple the physics and materiality of soft tissue, such as muscles, fat, and skin, enabling artists to create the most lifelike CGI characters. Ziva has successfully managed to distill these complex processes into friendly artist tools that keep building complexity to a minimum. The focus is on ease of use, non-destructible workflows, and derivative assets creation. With fully dynamic anatomy simulation, these virtual animals and humans move, stretch, and flex just as they would in real life, removing weeks of artistic manual labor from many character workflows.

Unity acquires Ziva Dynamics – fxguide 2
Ziva’s use in The Meg.

Ziva users include students, indies, small studios, and large VFX facilities. Ziva’s simulation data can be used to train creatures and characters of all kinds to perform in real-time environments using ZivaRT. ZivaRT was designed to accurately reconstruct the behavior of non-linear deformations within game engines. ZivaRT’s multilayer classical machine learning model respects the constraints of the training inputs while consistently hitting predictable run-times and lightweight memory allocation. ZivaRT provides for real-time deformation, even on consumer hardware (GPU or CPU). ZivaRT is already in use in AAA games, such as Ninja Theory’s Senua’s Saga: Hellblade II and movies and ZIVA core technology has been seen in TV shows such as Game of Thrones and in films like Godzilla vs. Kong or The Meg.

The core technology of ZIVA was developed as being engine agnostic, prior to the Unity deal for example, Emma was running in UE5(below) and UE4 (4k @ 60fps).


Source link

You may also like...

Leave a Reply

Your email address will not be published.