Ziva Face Trainer

From hyper realistic to stylized, for games, animation, or VFX, Ziva Face Trainer helps 3D character artists to achieve complex and believable facial animations, without compromise.

From mesh to rigged and real-time

Creating characters that perform complex and believable expressions can be time-consuming and expensive. Ziva Face Trainer accelerates the workflow by providing an accessible way to automatically generate film-quality facial rigging for your characters, ready for animation in Maya and Unity.

Robust facial controls

Ziva Faces are high-performance, fully rigged puppets that can express over 72,000 training shapes, as well as novel face poses.

Christopher Chinea, Virtual Production / R&D Supervisor, Pixomondo

"Ziva Face Trainer is a unique workflow that boosts our creative process, allowing us to augment and spend time where it counts for fidelity, allowing us to provide higher quality base work as well as more detailed photoreal content."

Christopher Chinea, Virtual Production / R&D Supervisor, Pixomondo

Real-time faces trained in the cloud

Train your character mesh to produce VFX-quality deformation in as little as one hour. Face Trainer saves months of rigging work by leveraging a library of over 45TB of 4D data and proprietary machine learning algorithms.

Join beta
Quality without compromise

Scale model performance to memory budget needs. Ziva Faces are only 30MB at runtime, and run at real-time rates in game engines (3ms/frame on a single CPU thread).

For any style of face

Ziva Face Trainer works on any human-like face, from hyperrealistic to fantastical or stylized.

Deploy as a Unity asset

Open your Ziva Face directly in Unity with the Unity Ziva RT Player. Download the free plugin to deploy and see your deformations in a game-engine environment.

Unity Wētā Tools

Spend more time in the creative flow with a complete toolchain for 3D creation, simulation, and rendering – for any size of project. Discover the latest artistry tools for environments, character, virtual production, and compositing.

Frequently asked questions

What software do I need to create a face mesh?

Firstly, you will need to register for an account with Ziva (it’s free!). Here you can access Ziva Face Trainer. Then, you can use any 3D Software such as Maya, Houdini, and 3ds Max to create a face mesh, as long as that software creates polygons and can be converted into an FBX file. This can then be run through Face Trainer after undergoing the Face Preparation stage (using these instructions).

How can I deploy and play with my trained Ziva Face?

In order to manipulate your trained Ziva Face in real-time, you will need to download the free Ziva RT Player plugin for Unity (supporting version 2021.3 and above, MacOS not currently supported). This will allow you to deploy and see your deformations in a game-engine environment. There are also Ziva RT Player plugins for Unreal Engine and Maya (supporting 2019, 2020, and 2022 versions).

Is Ziva Face Trainer an app or a service?

Ziva Face Trainer is a cloud-based application that enables 3D generalists to automatically generate a VFX-quality facial rig for any characters – from photo realistic to stylized, for games, animation, or visual effects.

Where can I access my previews or trained Ziva Faces?

You can access your previews in the Preview tab and your Ziva Faces in the Library tab of the Ziva Face Trainer web app (desktop devices only):

https://store.zivadynamics.com/facetrainer

Also, once a preview or training session is completed, you will receive a notification via email, from where you can access the Preview or Library tabs.

What happens if I don’t like the preview of my Ziva Face?

You don't have to commit to a training session. Previews are always free so you can make changes to your face mesh and re-upload for another Preview. Every Preview takes about 15 minutes to process and you'll receive an email notification once it's ready to view.

Can I use my own topology on the Ziva Face Trainer?

We are working on enabling users to use Ziva Face Trainer while maintaining their own topology, however this is not available at the moment.

我们使用 Cookie 来确保为您提供网站的最佳体验。有关更多信息,请访问我们的 Cookie 政策页面

明白了