The application uses machine learning and computer vision technology to automate the process.
Faceware Technologies, the developer of facial capture and animation tools, has introduced Faceware Portal, which utilizes a combination of machine learning and Faceware’s own computer vision technology to create a technique called neural net processing. This development tackles one of the most time-consuming elements of the facial animation workflow: manual facial tracking.
The process is fast and simple. Users upload their captured video to Faceware Portal, set the processing options, and click Process to create automated animation data. After downloading the animation, users next load any character in any engine, and then load the animation data from the Portal. As a result, each frame of the performance is transferred to the character.
According to Faceware, neural net processing automatically captures facial movements with frame accuracy in a matter of seconds. Faceware Portal then outputs the results in FWR for use in Faceware’s software or as JSON files for use in third-party applications such as Epic’s Unreal Engine, Autodesk’s Maya, Unity’s software, and others. With the new JSON direct data format, users are able to bypass the retargeting process and proceed from video directly to animation.
Faceware Portal initially is being offered to Faceware’s Enterprise customers. Pricing has not yet been released.