Figure 1: Morphable crowd models synthesize virtual crowds of any size and any length from input crowd data. The synthesized crowds can be interpolated to produce a continuous span of intervening crowd styles. AbstractCrowd simulation has been an important research field due to its diverse range of applications that include film production, military simulation, and urban planning. A challenging problem is to provide simple yet effective control over captured and simulated crowds to synthesize intended group motions. We present a new method that blends existing crowd data to generate a new crowd animation. The new animation can include an arbitrary number of agents, extends for an arbitrary duration, and yields a naturallooking mixture of the input crowd data. The main benefit of this approach is to create new spatio-temporal crowd behavior in an intuitive and predictable manner. It is accomplished by introducing a morphable crowd model that allows us to encode the formations and individual trajectories in crowd data. Then, its original spatiotemporal behavior can be reconstructed and interpolated at an arbitrary scale using our morphable model.

The standard C/C++ implementation of a spatial partitioning data structure, such as octree and quadtree, is often inefficient in terms of storage requirements particularly when the memory overhead for maintaining parentto-child pointers is significant with respect to the amount of actual data in each tree node. In this work, we present a novel data structure that implements uniform spatial partitioning without storing explicit parent-tochild pointer links. Our linkless tree encodes the storage locations of subdivided nodes using perfect hashing while retaining important properties of uniform spatial partitioning trees, such as coarse-to-fine hierarchical representation, efficient storage usage, and efficient random accessibility. We demonstrate the performance of our linkless trees using image compression and path planning examples.

No abstract

Figure 1: Morphable crowd models synthesize virtual crowds of any size and any length from input crowd data. The synthesized crowds can be interpolated to produce a continuous span of intervening crowd styles. AbstractCrowd simulation has been an important research field due to its diverse range of applications that include film production, military simulation, and urban planning. A challenging problem is to provide simple yet effective control over captured and simulated crowds to synthesize intended group motions. We present a new method that blends existing crowd data to generate a new crowd animation. The new animation can include an arbitrary number of agents, extends for an arbitrary duration, and yields a naturallooking mixture of the input crowd data. The main benefit of this approach is to create new spatio-temporal crowd behavior in an intuitive and predictable manner. It is accomplished by introducing a morphable crowd model that allows us to encode the formations and individual trajectories in crowd data. Then, its original spatiotemporal behavior can be reconstructed and interpolated at an arbitrary scale using our morphable model.

We present a neural network learning approach for estimating a set of cloth simulation parameters from a static drape of a given fabric. We use a variant of Cusick's drape, which is used in the fashion textile industry to classify fabric according to mechanical properties. In order to produce a large enough set of reliable training data, we first randomly sample simulation parameters using a Gaussian mixture model that is fitted with 400 sets of primary simulation parameters derived from real fabrics. Then, we simulate our modified Cusick's drape for each sample parameter set. To learn the training data, we propose a two-stream fully connected neural network model. We prove the suitability of our neural network model through comparisons of the learning errors and accuracy with other similar neural network and linear regression models. Additionally, to demonstrate the practicality of our method, we reproduce the drape shapes of real fabrics using the simulation parameters estimated from the trained neural networks.