Reanimating Faces in Images and Video

By Volker Blanz, Curzio Basso, Tomaso A. Poggio, Thomas Vetter. Computer graphics forum (Print), 2003

452 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/ba81d93e6da193d458be0ae1bd896ba47afce4b5

How researchers see this paper

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

In Computer Vision and Pattern Recognition

Citations in the corpus

Real-time expression transfer for facial reenactment (2015, methodology): “One type of face replacement techniques uses a morphable 3D model as an underlying face representation that parameterizes identity, facial expressions, and other properties, such as visemes or face texture [Blanz and Vetter 1999; Blanz et al. 2003; Blanz et al. 2004; Vlasic et al. 2005].”
Self-Supervised Multi-level Face Model Learning for Monocular Reconstruction at Over 250 Hz (2017, methodology): “A similar model for facial animation is presented in [6].”
Face transfer with multilinear models (2005, unknown): “Previous work combines identity and expression spaces by copying deformations from one subject onto the geometry of other faces [DeCarlo and Metaxas 2000; Blanz et al. 2003; Chai et al. 2003].”
Robust semantic analysis by synthesis of 3D facial motion (2011, methodology): “The use of 3DMMs has been shown before to be effective for the analysis of images [5], video [3] and in reconstructing 3D scan data of neutral faces [4].”
Spacetime faces: high resolution capture for modeling and animation (2004, methodology): “Several previous authors explored performance-based methods for animating faces, using either video of an actor [ 12 ,20], or speech [21‐23] to drive the animation.”
A 3D Face Model for Pose and Illumination Invariant Face Recognition (2009, background): “They have become a well established technology able to perform various tasks, most important face recognition [8, 18, 11], but also face image analysis [7] (estimating the 3D shape from a single photograph), expression transfer between individuals [6, 17], animation of faces and whole bodies [6, 1],”
EMOCA: Emotion Driven Monocular Face Capture and Animation (2022, background): “Second, reconstruction metrics like landmark reprojection loss [8], photometric loss [10], face recognition loss [30], or multi-image consistency losses [73, 80], are either not affected by facial expressions, or require perfect image alignment to capture subtle cues.”
MoFA: Model-Based Deep Convolutional Face Autoencoder for Unsupervised Monocular Reconstruction (2017, background): “A closely related approach to AAMs is the 3D morphable model of faces (3DMM) [3], which has been used for learning facial animations from 3D scans [2].”
StyleRig: Rigging StyleGAN for 3D Control Over Portrait Images (2020, background): “These models provide artist-friendly control (often called a face rig), while navigating the various parameters of a morphable face model (3DMM) [3, 4].”
Example-based facial rigging (2010, methodology): “Automatic creation of facial models using (multi-) linear PCA models was proposed in [Blanz and Vetter 1999; Blanz et al. 2003; Vlasic et al. 2005], though the resulting linear blendshapes are not necessarily meaningful for facial animation control.”
Face recognition based on a 3D morphable model (2006, background): “In the Morphable Model, facial expressions can, for example, be modeled by recording 3D scans of a face at different expressions, establishing pointto-point correspondence to the neutral reference face of the Morphable Model, and adding the facial expressions as new vectors to the dataset [1].”
Reconstruction of Personalized 3D Face Rigs from Monocular Video (2016, methodology): “The facial anthropometry across people can also be modeled, for example, as a parametric PCA space learned from a database of laser scans [Blanz and Vetter 1999; Blanz et al. 2003].”
The Creation and Detection of Deepfakes (2020, background): “their desired impact. Before we review the subject, we note that expression reenactment has been around long before deepfakes were popularized. In 2003, researchers morphed models of 3D scanned heads [52]. In 2005, it was shown how this can be done without a 3D model [53]. Later, between 2015 and 20”
3D Morphable Face Models—Past, Present, and Future (2019, unknown): “First, off-line techniques have been proposed [Blanz et al. 2003; Bregler et al. 1997; Kemelmacher-Shlizerman et al. 2010; Li et al. 2014a, 2012; Theobald et al. 2009; Vlasic et al. 2005b].”
Neural Face Editing with Intrinsic Image Disentangling (2017, unknown): “This model and its variants have been used for a variety of tasks including relighting [34, 8], face attribute editing [7], expression editing [5, 22], authoring facial performances [32, 31], and aging [16].”
Dynamic Neural Radiance Fields for Monocular 4D Facial Avatar Reconstruction (2020, methodology): “Our method is built upon a low-dimensional morphable model [7, 41] which is a building block of numerous facial reconstruction and animation approaches [14, 15, 46, 45, 40, 41, 5, 6, 21, 39, 38].”