Practice and Theory of Blendshape Facial Models

By John P. Lewis, Ken-ichi Anjyo, Taehyun Rhee, Mengjie Zhang, Frédéric H. Pighin, Zhigang Deng. Eurographics, 2014

321 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/987063e7717369a5156f9a8baadc3b6de1a83267

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Learning Pose-Aware Models for Pose-Invariant Face Recognition in the Wild (2019, methodology): “The ﬁxed 3D generic shape M was taken from a set of BlendShapes [44], selecting the one with neutral expression.”
Constraining dense hand surface tracking with elasticity (2020, background): “While blendshapes are well suited for animating faces [Lewis et al. 2014], even with rigid transforms factored off hands experience massive nonlinear local changes in shape, making direct applications of linear blendshapes difficult.”
Modeling Facial Geometry Using Compositional VAEs (2018, methodology): “They include Active Appearance Models (AAM) [7], blendshapes [23], principal components analysis (PCA) derived from a set of training shapes [22, 3], and multilinear models [37].”
3D Morphable Face Models—Past, Present, and Future (2019, unknown): “Note that the basis vectors of the expression space can be interpreted as a data-driven blendshape model, where the basis vectors are orthogonal and do not carry interpretable semantic meaning in general [Lewis et al. 2014a].”
Do We Really Need to Collect Millions of Faces for Effective Face Recognition? (2016, methodology): “1) we estimate facial expression by fitting a 3D expression Blendshape, similarly to [19].”
State of the Art on Monocular 3D Face Reconstruction, Tracking, and Applications (2018, methodology): “This problem can be alleviated by utilizing pairwise activation constraints of the form δ i δ j = 0 , i (cid:54) = j for some of the weights [LAR ∗ 14], which means that only one of the two blend-shapes should be activated.”
Reconstruction of Personalized 3D Face Rigs from Monocular Video (2016, background): “They commonly resort to facial expression control using a set of blendshapes that span intuitive atomic face expressions and are linearly combined to obtain a new pose [Lewis et al. 2014].”
Audio-driven facial animation by joint end-to-end learning of pose and emotion (2017, background): “Alternative encodings [Lewis et al. 2014] such as blend shapes or non-linear rigs can be introduced at later pipeline stages, if needed for compression, rendering, or editability.”