TY - ADVS
T1 - HUMAIN
T2 - EKER
A2 - Maguire, Greg
PY - 2019/7/28
Y1 - 2019/7/28
N2 - Maguire’s design-led research developed within his spin out company, HUMAIN, has transformed facial animation to an extent that it has positive implications moving beyond the ‘uncanny valley’ in digital human realism. Ulster University established a collaborative research agreement (2017) with Dr Erika Rosenberg (Stanford University, now Chief Scientific Officer of HUMAIN) to develop the seminal work of Dr Paul Ekman’s Facial Action Coding System (FACS). EKER represents the bond between and continuation of Paul (EK)man and (E)rika (R)osenberg’s research. The background research (Ekman & Friesen, 1978) revised (Ekman, Friesen, & Hager, 2002) relates to cross cultural emotional facial expression. It was transformative in psychology and subsequently recognised as the ‘gold standard’ in animation sectors (Seymour, 2019). Maguire et al extends this work with new insights, through technological applied research. New databases were developed including a ‘4D Scanned Database of Posed Facial Muscle Movements’, which uniquely maps the animated transitional expressions hitherto undefined. Rosenberg is the world leading expert in FACS. Maguire et al captured the full FACS sequence of 44 Action Units (AU’s) using incredibly detailed 4D scans of micro expressions in motion. Performed and verified by Rosenberg, the world’s first and only authentic, clean data set of digital facial emotions has >50,000 connections. Original algorithms were designed to interrogate temporal 3D facial topology, affording precision analysis within a Machine Learning infrastructure. The algorithms recognise facial landmarks and their orientation for precision facial mapping. They convert high density data into simple spline models, for tracking providing a new edge flow process for animation preserving and controlling complex interspine curvatures. They capture nonlinearity and new datasets of rotational motions between each AU. Other algorithms permit cross cultural and gender interpretations of an authentic FACS data set and even interspecies applications towards humanisation for entertainment anthromorphisation. Rosenberg announced the collaboration with Maguire at SIGGRAPH (Seymour, 2019. 06:25) and demonstrated her performed AU’s driving another person’s digital face using the software (19:00 to 20:40).
AB - Maguire’s design-led research developed within his spin out company, HUMAIN, has transformed facial animation to an extent that it has positive implications moving beyond the ‘uncanny valley’ in digital human realism. Ulster University established a collaborative research agreement (2017) with Dr Erika Rosenberg (Stanford University, now Chief Scientific Officer of HUMAIN) to develop the seminal work of Dr Paul Ekman’s Facial Action Coding System (FACS). EKER represents the bond between and continuation of Paul (EK)man and (E)rika (R)osenberg’s research. The background research (Ekman & Friesen, 1978) revised (Ekman, Friesen, & Hager, 2002) relates to cross cultural emotional facial expression. It was transformative in psychology and subsequently recognised as the ‘gold standard’ in animation sectors (Seymour, 2019). Maguire et al extends this work with new insights, through technological applied research. New databases were developed including a ‘4D Scanned Database of Posed Facial Muscle Movements’, which uniquely maps the animated transitional expressions hitherto undefined. Rosenberg is the world leading expert in FACS. Maguire et al captured the full FACS sequence of 44 Action Units (AU’s) using incredibly detailed 4D scans of micro expressions in motion. Performed and verified by Rosenberg, the world’s first and only authentic, clean data set of digital facial emotions has >50,000 connections. Original algorithms were designed to interrogate temporal 3D facial topology, affording precision analysis within a Machine Learning infrastructure. The algorithms recognise facial landmarks and their orientation for precision facial mapping. They convert high density data into simple spline models, for tracking providing a new edge flow process for animation preserving and controlling complex interspine curvatures. They capture nonlinearity and new datasets of rotational motions between each AU. Other algorithms permit cross cultural and gender interpretations of an authentic FACS data set and even interspecies applications towards humanisation for entertainment anthromorphisation. Rosenberg announced the collaboration with Maguire at SIGGRAPH (Seymour, 2019. 06:25) and demonstrated her performed AU’s driving another person’s digital face using the software (19:00 to 20:40).
UR - https://humain-studios.com/
UR - https://pure.ulster.ac.uk/en/publications/humain-eker
M3 - Software
PB - HUMAIN
ER -