Computer Animation and Virtual Worlds

This paper presents an approach for video event inference from dozens of actions performed by multiple players. The various configurations of the events are represented by an And-Or graph. The bottom-up and top-down processes are used for inferring the events from the multi-player actions. We encode all the six conditions of temporal relations and four of role relations for each pair of nodes, which makes it possible to represent the relationship between the actions with the same semantic label.

A growth volume was introduced to model numerous trees for a virtual real-time ecosystem. To increase the efficiency of the system and to allow natural tree generation, an instancing-based branch control procedure was added for rendering. Experiments were performed to verify whether various and natural trees could be generated easily on the basis of the suggested method.

We propose a state-aware synthesis framework for video-driven facial animation. Our method extracts feature points to track facial actions and analyzes face texture to obtain a high-level understanding of expression states. Leveraging these two information, our method synthesizes expressive and emotion-correct facial animation for avatars.

We propose a new video composition method by 3D mean-value coordinate. Our method can maintain the temporal and spatial consistency of a composite video and is able to achieve high performance blending with less computation.

We describe a method that successfully generates mosaic animations from videos by applying a temporally and spatially coherent tile-arrangement technique. We arrange tiles on the basis of the feature lines extracted from video input. We then animate the tiles along the motion of the video, add and delete tiles to preserve the tile density, and smooth tile color via frames. Finally, we propose indices to evaluate the temporal coherence of video-based animations, and show that our animations are temporally coherent.

Cage-based inverse kinematics can naturally pose a wide range of target models, including models that consist of hundreds of components, by manipulating the handle points placed on the model surface. The proposed technique based on pose space interpolation allows us to deform the model locally, to place handles during run-time and to incorporate a linear rotation-invariant mesh representation that accommodates examples with extreme poses.

We propose data-driven race path synthesis that provides controllability for racer animations. We introduce formation field, a data structure that samples regions in formation space containing formations of exciting and realistic race paths, generated using a set of race paths. The synthesized race path has its quality guaranteed as it synthesized from existing paths with good quality guaranteed. This method is suitable for both real-time applications such as horse racing games.

Detail-aware spatial deformation transfer can transfer fine-scale deformations, for example, skirt motions and muscle bulging, to a multi-component model. Self-collisions are handled by approximating spatial relationships between surfaces, from the result of space deformation.

In this paper, we present a new technique to address two major limitations of the widely adopted linear blend shape technique, which are rotational deformations and deformations because of self collision or scene interaction, by introducing physical-based simulation based on mass-spring systems.

We present a new approach to interactive resequencing of captured motion data that allows the user to progressively extend animations such that each extension can best fit user requirements with respect to space, time, and pose. Our implementation on the iPad demonstrates that the user can rapidly build up his or her animation sequences as intended with the use of only tap and swipe touch gestures.

An easily parallelizable approach to real-time cloth animation in interactive application. The animation method is based on the harmonic oscillation model, and the realistic reflectance on woven surface is efficiently generated by rotating microfacet distribution function.

This paper presents a novel approach for simulating the dynamic coupling between a tree and raindrops based on physical deformation and fluid simulation. By the approach, tree animation in the rain can be simulated in a two-resolution way: branch motion and leaf motion. Interaction coupling liquid motion-based 2.5D shallow water equations on the hydrophilic surface with a flexible petiole is well implemented by a special design for various phenomena simulations.

A new thermal fluid model based on hierarchical lattice is proposed to process the solid–liquid interaction and the phase transition in the solid–liquid interface combined with free surface tracking. With the use of hybrid interaction with multidistribution functions, different realistic multiple fluids phenomena are rendered with different lattice sizes.

We present a new hierarchical fracture simulation method. Our fracture model is broken into small pieces in a gradual way by a given hierarchy. One can control the whole process of fracture modelling and simulation in a more direct way. We also introduce a simple and fast fracture criterion capturing the motion difference for a node in which no matrix solves is needed.

This paper presents a method for simulating the motion of water drops on a surface in real time.We describe the dynamics of a drop moving on the surface, and then we present our simulation model. We use a geometry-based representation of a drop. Each drop is modeled by a deformable 3D mesh. Our method is fast and robust and yields realistic results when applied to treat condensation on a surface or human sweating in real time.

We present a unified structure for the simulation of crowds in complex environments. We propose a topological multiresolution model, allowing efficient proximity querying and compatible with real-time rendering and hierarchical path planning. A fine exploitation of the multiscale aspect of the underlying model allows to achieve the same efficiency as the fastest existing methods. The generality of the approach allows the simulation to be executed on any two manifold, and the unified approach eases the handling of dynamic environments.

Groups of conversing characters can add an increased sense of realism to a virtual crowd. This paper discusses a set of perceptual experiments focussing on formations of such groups. Our results show that, although participants were sensitive to both distance and orientation changes between talkers and listeners in a virtual conversation, they were not as sensitive to anomalous gesturing behaviours across different distances.

We examine the decisions and physiological reactions of participants (heart rate and electrodermal activity) when engaged in a decision task (prisoner's dilemma game) with emotionally expressive agents. We observe that people's decisions are influenced by such emotional displays, but these influences differ depending on the extent to which these displays provoke an affective response. Specifically, we show that an individual difference predicts the extent to whether people will engage affectively or strategically with such agents, thereby better predicting their decisions.

This paper adopts Centroidal Voronoi Tessellation and incorporates environment semantics and agent characteristics to determine agent goal locations and allow agents to demonstrate preferences. This work leads to more reasonable simulations than current randomly distributed goals and leads to more consistent agent behaviors.

Virtual calligraphic carving aims to generate 3D virtual carving works from 2D calligraphic images. Our method is based on such heuristics: calligraphic carving work is assumed to be a depth map composed of smoothness scalar field and brush-pressure distribution, which ensures the smoothness of carving surface and interprets the strength of calligraphy, respectively. The experiments show that this method can generate virtual calligraphic carving works that are vivid and very similar to that of artists.

This paper proposes a new adaptive surface splatting method to reduce the number of splats by optimizing the size and shape of splats using geometric and color information. Using the optimized splats, we can improve the rendering quality, especially in the visually sensitive feature areas. Our adaptive surface splatting is very effective to render massive facial data on tablet PCs or smartphones.

We present a novel culling algorithm that uses deforming noncollinear filters to improve the performance of continuous collision detection. For vertex–face pair, in the orthogonal plane perpendicular to n_t, if vertex P_t is always on one side of edge a_tb_t and vertex c_t is always on the other side of a_tb_t, this pair should be culled out; for edge–edge pair, in the orthogonal plane perpendicular to n_t , if the two vertices of one edge are always on the same side of the other edge during the time interval, this pair should be culled out.

We present an efficient, user-guided volumetric approximation algorithm, specifically designed for physically based animation. Our approach first decomposes an input mesh object into parts in terms of swept sphere volumes using the simplified medial axis, and then each decomposed region is approximated with a swept sphere volume. These decomposed surface regions can be interactively refined further by splitting and/or merging using a sketch-based input.

On the basis of a finite element method framework, the dynamic withering procedure is computed with diffusion of moisture content and decay spread. The material properties are controlled by moisture content, and the shrinkage also depends on moisture content through the initial strain. The heterogeneous deformation is simulated realistically and efficiently with our approach.

This paper presents a feature-preserving surface reconstruction method from sparse noisy 3D measurements such as range scanning or passive multiview stereo. We rasterize input-oriented points into a probabilistic volume (base volume) and then create a guidance volume by Gaussian filtering. Both the base volume and the guidance volume are further filtered by regularized weighted least squares filter to detect and recover detail features. A global minimal surface is computed by graph cut and meshed as a geometric model.

We propose a new image-space technique to summarize a 3D animation into a single image using the depth information. We extract and compose important frames by constructing the gradient of a composite depth image and evaluating the amount of contribution of frames to the gradient. We provide a user with a control to specify the visual complexity of a single image summary by evaluating the amount of self-overlap.

The performance of a finite element-based simulation using explicit time integration schemes can be optimized by improving the initial tetrahedral meshes. We provide quality metrics related to the critical simulation time step.

We propose a new smoothing approach that uses spring systems to displace a node with respect to the positions that maximize the time step of the elements adjacent to that node.

The algorithm is combined with topological transformations. The simulation performance improves on average by 30%.

The use of linear interpolation to blend different terrain types with distinct features produces translucency artefacts that can detract from the realism of the scene. The approach presented in this paper addresses the feature agnosticism of linear blending and makes the distinction between features and non-features. Using the blend weights from Bloom's texture splatting, intermittent texture transitions are generated on the fly without artistic intervention. Furthermore, feature shapes are modified dynamically, thus further reducing repetition and adding authenticity.

Creating an attractive mesh animation is a laborious and time-consuming task. This paper presents a practical deformation transfer algorithm to make it easier. Various experimental results demonstrate the effectiveness and applicability of this paper.

We propose an integrated facial dynamics model addressing the animation of 3D humanoid faces in real time. The model is flexible enough to animate face meshes of various shape, connectivity, and scale. The model allows for driving virtual facial animation from videos of real actors performing face expressions.