SwiftAvatar: Efficient Auto-Creation of Parameterized Stylized Character on Arbitrary Avatar Engines

Parameterized Generation of Avatar Face Dataset

Virtual communities such as Second Life and Entropia Universe are quickly becoming the next frontier of cybercrime. With GDPs of virtual economies of persistent worlds approaching billions of dollars it is necessary to develop tools for protection of virtual environments similar to those utilized to secure real world’s infrastructure, such as biometrics security systems. The first step in the d...

Learning Stylized Character Expressions from Humans

Modeling Stylized Character Expressions via Deep Learning

We propose DeepExpr, a novel expression transfer customized expression transfer approach from human to multiple stylized characters. We first train two Convolutional Neural Networks to recognize the expression of humans and stylized characters independently. Then we utilize a transfer learning technique to learn the mapping from humans to characters to create a shared embedding feature space. T...

Stylized and Performative Gaze for Character Animation

Existing models of gaze motion for character animation simulate human movements, incorporating anatomical, neurophysiological, and functional constraints. While these models enable the synthesis of humanlike gaze motion, they only do so in characters that conform to human anatomical proportions, causing undesirable artifacts such as cross-eyedness in characters with non-human or exaggerated hum...

Avatar Creation using Automatic Face Recognition

In the multimedia context, an avatar is the visual representation of the self in a virtual world. It is desirable to incorporate personal information, such as an image of face, about the user into the avatar. To this end we have developed an algorithm which can automatically extract a face from an image, modify it, characterize it in terms of high-level properties, and apply it to the creation ...