Jens Fursund

With the rising interest in personalized VR and gaming experiences comes the need to create high quality 3D avatars that are both low-cost and variegated. Due to this, building dynamic avatars from a single unconstrained input image is becoming a popular application. While previous techniques that attempt this require multiple input images or rely on transferring dynamic facial appearance from a source actor, we are able to do so using only one 2D input image without any form of transfer from a… Vis mere

With the rising interest in personalized VR and gaming experiences comes the need to create high quality 3D avatars that are both low-cost and variegated. Due to this, building dynamic avatars from a single unconstrained input image is becoming a popular application. While previous techniques that attempt this require multiple input images or rely on transferring dynamic facial appearance from a source actor, we are able to do so using only one 2D input image without any form of transfer from a source image. We achieve this using a new conditional Generative Adversarial Network design that allows fine-scale manipulation of any facial input image into a new expression while preserving its identity. Our photoreal avatar GAN (paGAN) can also synthesize the unseen mouth interior and control the eye-gaze direction of the output, as well as produce the final image from a novel viewpoint. The method is even capable of generating fully-controllable temporally stable video sequences, despite not using temporal information during training. After training, we can use our network to produce dynamic image-based avatars that are controllable on mobile devices in real time. To do this, we compute a fixed set of output images that correspond to key blendshapes, from which we extract textures in UV space. Using a subject's expression blendshapes at run-time, we can linearly blend these key textures together to achieve the desired appearance. Furthermore, we can use the mouth interior and eye textures produced by our network to synthesize on-the-fly avatar animations for those regions. Our work produces state-of-the-art quality image and video synthesis, and is the first to our knowledge that is able to generate a dynamically textured avatar with a mouth interior, all from a single image.

https://www.youtube.com/watch?v=_JMx5VN1eeM&feature=youtu.be
https://www.fxguide.com/featured/a-i-at-siggraph-part-2-pinscreen-at-real-time-live/ Vis mindre

We present a fully automatic framework that digitizes a complete 3D head with hair from a single unconstrained image. While the generated face is a high-quality textured mesh, we propose a versatile and efficient polygonal strips (polystrips) representation for the hair, compatible with existing real-time game engines.

The age of social media and immersive technologies has created a growing need for processing detailed visual representations of ourselves as virtual and augmented reality is growing into the next generation platform for online communication, connecting hundreds of millions of users. A realistic simulation of our presence in a mixed reality environment is unthinkable without a compelling and directable 3D digitization of ourselves. With the wide availability of mobile cameras and internet… Vis mere

The age of social media and immersive technologies has created a growing need for processing detailed visual representations of ourselves as virtual and augmented reality is growing into the next generation platform for online communication, connecting hundreds of millions of users. A realistic simulation of our presence in a mixed reality environment is unthinkable without a compelling and directable 3D digitization of ourselves. With the wide availability of mobile cameras and internet images, we introduce a technology that can build a realistic 3D avatar from a single photograph. This textured 3D face model includes hair and can be instantly animated by anyone in real-time through natural facial performances captured from a regular RGB camera. Immediate applications include personalized gaming and VR-enabled social networks using automatically digitized 3D avatars, as well as mobile apps such as video messengers (e.g., Snapchat) with face-swapping capabilities. As opposed to existing solutions, our technology enables the automatic generation of a complete head model from a fully unconstrained image. Vis mindre

In this paper, we present a method for animating multiphase flow of immiscible fluids using unstructured moving meshes. Our underlying discretization is an unstructured tetrahedral mesh, the deformable simplicial complex (DSC), that moves with the flow in a Lagrangian manner. Mesh optimization operations improve element quality and avoid element inversion. In the context of multiphase flow, we guarantee that every element is occupied by a single fluid and, consequently, the interface between… Vis mere

In this paper, we present a method for animating multiphase flow of immiscible fluids using unstructured moving meshes. Our underlying discretization is an unstructured tetrahedral mesh, the deformable simplicial complex (DSC), that moves with the flow in a Lagrangian manner. Mesh optimization operations improve element quality and avoid element inversion. In the context of multiphase flow, we guarantee that every element is occupied by a single fluid and, consequently, the interface between fluids is represented by a set of faces in the simplicial complex. This approach ensures that the underlying discretization matches the physics and avoids the additional book-keeping required in grid-based methods where multiple fluids may occupy the same cell. Our Lagrangian approach naturally leads us to adopt a finite element approach to simulation, in contrast to the finite volume approaches adopted by a majority of fluid simulation techniques that use tetrahedral meshes. We characterize fluid simulation as an optimization problem allowing for full coupling of the pressure and velocity fields and the incorporation of a second-order surface energy. We introduce a preconditioner based on the diagonal Schur complement and solve our optimization on the GPU. We provide the results of parameter studies as well as a performance analysis of our method. Vis mindre