OUR TECH
Through cutting-edge technology and constant R&D,
EVR STUDIO strives to create believable digital AI humans
that people can emotionally connect with.
Digital Human 3D Scan
A facial scan system is the first solution towards creating the perfect digital human in a real-time engine. We built our own scan system that enables us to scan a person’s face at the highest resolution and quality, estimated at 100 times more precise compared to photogrammetry. We can extract physically based textures such as diffuse, specular, and normal maps from a subject.
- Micro Geometry : skin detail is scanned at 20 micrometer precision.
- Physically Based Texture : by extracting physically perfect textures, we can create skin in its exact condition.
- Dynamic Skin System : Every details from stretches
to wrinkles and blood flow under the skin, we can
reflect life-like details to each and every facial
expression in real-time.
Facial Capture
EVR Studio’s custom Facial Rig is based on Facial Action
Coding System and created to match ergonomic muscle
movements. This allows us to manifest realistic changes
made in facial muscles, and skin surface from stretches to
wrinkles made during various facial expressions.
We have built a proprietary technology based on facial muscle
structure to create millions of natural looking facial
expressions for our digital humans.
Facial data, audio, and video is stored and automatically
synchronized on our database server, which allow
us to apply and check results in real-time.
Fine wrinkles and changes made to the skin across different
facial expressions are implemented in real-time.
VR Mobile Streaming
Through our solution, VR content requiring heavy computing power can be streamed to less expensive mobile VR HMDs and all-in-one VR HMDs from the PC without the need of an expensive high-end VR HMD with minimal latency.
- Stream high-quality (1440p@60fps) bi-directional content from PC to target HMD at under 60ms.
- Asynchronous time warping can further reduce streaming experience to latency under 30ms.