Researcher: Prof Alan Johnston
Research Institution: University College London
Project Overview:
Prof Alan Johnston from University College London is seeking industry partners to help support the development of his work on facial motion analysis to enable the product to be commercialised.
Product:
• A method to describe the variation of the face either within or across individuals;
• The capacity to map expressions between faces, allowing photorealistic performance driven animation;
• The capacity to drive and animate a photorealistic character;
• The capacity to map expression or identity between views of the face; and
• The analysis, characterization and generation of types of faces for example faces rated high on dominance, friendliness or trustworthiness.
Market:
• Performance driven animation and character generation in video games and film;
• Casting – enables users to identify what a character or personality characteristic might look like in the public imagination;
• Transformation – enables users to estimate what a face would look like under changes in viewpoint, age or lighting;
• Occlusion – enables users to estimate a whole face from a partly occluded face; and
• Facial motion signatures.
Benefits:
• A novel high fidelity optic flow technique is used for registration, permitting changes in identity or expression to be encoded with high spatial and temporal resolution allowing, for example, the encoding, analysis and generation of nuanced expressions;
• Markerless registration is used meaning it does not require special systems for motion capture, interventions or the physical presence of subjects; and
• The output (animation, transformation, facial type) is photorealistic so it looks real rather than computer generated.
Partners:
Industry partners are sought to provide financial assistance to help support the development of work on facial motion analysis to enable the product to be commercialised.
Contact Details:
Email Prof Alan Johnston at a.johnston@ucl.ac.uk
