We’re excited to announce the official release of the 4D Face Model 2019 (2019.1). The main updates to the preview version include: A more expressive identity model, built from more scans of a wider range of ethnicities, and now 36 expressions/action units, with improved quality. Below video cycles through the action units as blendshapes:

We now make the model available with both the expressions as PCA model, as well as blendshapes. The PCA expression model is useful for tasks and algorithms that were made to work with traditional 3D morphable models, as modelling expressions with PCA is quite a common technique in the 3D morphable model community. Blendshapes (or morph targets) on the other hand are commonly used in the world of computer graphics and animation, and have the distinct advantage that each blendshape corresponds to a specific expression or facial action unit.

The identity model is now built from a wider range of ethnicities, and we’ve incorporated data from a diverse population, covering the Caucasian, Asian, African, Indian and Middle-Eastern regions, among others. Below a few pictures of various identity and expression variations. Some images are shown with a per-vertex colour, just for visualisation purposes.

As in the preview version, the 4DFM comes in two different resolution (subdivision) levels, and with an uv embedding.

We are very excited to offer a 3D morphable face model with market-leading identity variation and expressions, covering the full head. With 36 expressions/action units, optimised topology, two resolution levels and uv embedding, we offer the best out-of-the-box experience possible, to drive tomorrow’s applications in augmented or virtual reality (AR/VR), games, entertainment, or health care.

The 4DFM 2019 is available now. Get in touch with us via email or contact form for more details, pricing and licensing, and check out the 4DFM product page. We’d love to hear from you.