Captured performances are uploaded as digital video files via the Image Metrics FACEWARE processing portal; a secure, high-bandwidth cloud service that provides easy-to-use tools to process your video performances and deliver to you analyzed data of each performance as an Image Metrics Performance Data (IMPD) file. Artists will import these IMPD files into FACEWARE to drive poses they set on their rigs. In the next step in the workflow, they will define the relationship of actor to character—enabling rapid turnaround to meet the strict schedules of today's dynamic production environment.
Once uploaded to our processing portal, our patented analysis process is initiated, which leverages computer vision technology to study a performer's movement frame by frame. By initially pinpointing key features on the face for each frame, the entire face can be isolated. Proprietary algorithms utilize pre-established knowledge about the human face, such as distance from eyes to nose and nose to mouth, to build a complex movement data model of the performance. Algorithms also take into consideration detailed pixel information about a person's eye, teeth, tongue, skin, and jaw movements, allowing FACEWARE to build complex three dimensional data models specific to an individual and his or her performance.
The algorithms include features to compensate for camera movement or shake, lens distortion and sub-optimal exposure. Each performance is quality controlled, with results verified by our image analysis experts to ensure dependable data for use in the FACEWARE animation toolset.
