We introduce a deep reinforcement learning method that learns to control articulated humanoid bodies imitate given target motions closely when simulated in physics simulator. The motion, which may not have been seen by the agent and can be noisy, is supplied at runtime. Our recover balance from moderate external disturbances keep imitating motion. When subjected large cause fall down, our character get up track trained mocap clips CMU motion capture database number of other publicly...

We present a new Laplacian solver for minimal surfaces---surfaces having mean curvature of zero everywhere except at some fixed (Dirichlet) boundary conditions. Our solution has two main contributions: First, we provide robust rasterization technique to transform continuous values (diffusion curves) discrete domain. Second, define variable stencil size diffusion that solves the surface problem. prove converges right solution, and demonstrate it is least as fast commonly proposed multigrid...

Abstract In this paper we present dart throwing algorithms to generate maximal Poisson disk point sets directly on 3D surfaces. We optimize by efficiently excluding areas of the domain that are already covered existing darts. case triangle meshes, our algorithm shows dramatic speed improvement over comparable sampling methods. The simplicity basic naturally extends other surface types, including spheres, NURBS, subdivision surfaces, and implicits. further extend method handle variable...

Signed distance fields (SDFs) are a popular shape representation for collision detection. This is due to their query efficiency, and the ability provide robust inside/outside information. Although it straightforward test points interpenetration with an SDF, not clear how extend this continuous surfaces, such as triangle meshes. In paper, we propose per-element local optimization find closest between SDF isosurface mesh elements. allows us generate accurate contact sharp point-face pairs,...

In this paper we re-examine the idea that implicit integrators with large time steps offer best stability/performance trade-off for stiff systems. We make surprising observation performing a single step n constraint solver iterations is less effective than computing smaller steps, each iteration. Based on observation, our approach to split every visual into substeps of length Δt/n and perform iteration extended position-based dynamics (XPBD) in such substep. When compared traditional...

Abstract We present a rigid body simulation method that can resolve small temporal and spatial details by using quasi explicit integration scheme is unconditionally stable. Traditional simulators linearize constraints because they operate on the velocity level or solve equations of motion implicitly thereby freezing constraint directions for multiple iterations. Our always works with most recent directions. This allows us to trace high speed objects colliding against curved geometry, reduce...

Abstract We examine the relationship between primal, or force‐based, and dual, constraint‐based formulations of dynamics. Variational frameworks such as Projective Dynamics have proved popular for deformable simulation, however they not been adopted contact‐rich scenarios rigid body simulation. propose a new preconditioned frictional contact solver that is compatible with existing primal optimization methods, competitive complementarity‐based approaches. Our relaxed model generates improved...

Abstract Diffusion curves are a powerful vector graphic representation that stores an image as set of 2D Bezier with colors defined on either side. These diffused over the plane, resulting in smooth color regions well sharp boundaries. In this paper, we introduce new automatic diffusion curve coloring algorithm. We start by defining geometric heuristic for maximum density control points along curves. Following this, present algorithm to these so is close possible source least squares sense....

The current state of the art in real-time two-dimensional water wave simulation requires developers to choose between efficient Fourier-based methods, which lack interactions with moving obstacles, and finite-difference or finite element handle environmental but are significantly more expensive. This paper attempts bridge this long-standing gap complexity performance, by proposing a new method that can faithfully simulate obstacles real time while simultaneously preserving minute details...

We present an efficient wavefront tracking algorithm for animating bodies of water that interact with their environment. Our contributions include: a novel technique enables dispersion, refraction, reflection, and diffraction in the same simulation; unique multivalued function interpolation method our simulations to elegantly sidestep Nyquist limit; dispersion approximation efficiently amplifying number simulated waves by several orders magnitude; additional extensions allow time-dependent...

This paper presents a method for simulating water surface waves as displacement field on 2D domain. Our relies Lagrangian particles that carry packets of wave energy; each packet carries information about an entire group trains, opposed to only single crest. approach is unconditionally stable and can simulate high resolution geometric details. also straightforward interface artistic control, because it essentially particle system with intuitive parameters like wavelength amplitude....

We present a framework for the simulation of rigid and deformable bodies in presence contact friction. Our method is based on non-smooth Newton iteration that solves underlying nonlinear complementarity problems (NCPs) directly. This approach allows us to support dynamics models, including hyperelastic articulated mechanisms, coupled through smooth isotropic friction model. The fixed-point nature our means it requires only solution symmetric linear system as building block. propose new...

Shadow mapping suffers from spatial aliasing (visible as blocky shadows) well temporal flickering). Several methods have already been proposed for reducing such artifacts, but so far none is able to provide satisfying results in real time.This paper extends shadow by reusing information of previously rasterized images, stored efficiently a so-called history buffer. This buffer updated every frame and then used the calculation. In combination with special confidence-based method update (based...

We present a method designed to address some limitations of typical route map displays driving directions. The main goal our system is generate printable version that shows the overview and detail views within single, consistent visual frame. Our proposed visualization provides more intuitive spatial context than simple list turns. novel multifocus technique achieve this goal, where foci are defined by points interest (POI) along route. A lens encapsulates POI at finer geospatial scale...

Abstract We introduce a triangle mesh based convolutional neural network. The proposed network structure can be used for problems where input and/or output are defined on manifold with or without boundary. demonstrate its applications in cloth upsampling, adding back details to Principal Component Analysis (PCA) compressed cloth, regressing clothing deformation from character poses, and hand skin bones' joint angles. data training this work generated high resolution extended position...

Capturing real-world objects with laser-scanning technology has become an everyday task. Recently, the acquisition of dynamic scenes at interactive frame rates feasible. A high-quality visualization resulting point cloud stream would require a per-frame reconstruction object surfaces. Unfortunately, computations are still too time-consuming to be applied interactively. In this paper we present local surface and technique that provides feedback for reasonably sized clouds, while achieving...

We present a method to learn and propagate shape placements in 2D polygonal scenes from few examples provided by user. The placement of is modeled as an oriented bounding box. Simple geometric relationships between this box nearby scene polygons define feature set for the placement. sets all example are then used probabilistic model over possible scenes. With model, we can generate new with similar any given scene. introduce extensions that enable propagation generation shapes 3D scenes,...

Shell mapping is a technique to represent three-dimensional surface details. This achieved by extruding the triangles of an existing mesh along their normals, and 3D function (e.g., texture) into resulting prisms. Unfortunately, such nonlinear. Previous approaches perform piece-wise linear approximation subdividing prisms tetrahedrons. However, often leads severe artifacts. In this paper we present correct (i.e., smooth) that does not rely on decomposition We efficient GPU ray casting...

Diffusion curve images (DCI) provide a powerful tool for efficient 2D image generation, storage and manipulation. A DCI consist of curves with colors defined on either side. By diffusing these over the image, final result includes sharp boundaries along smoothly shaded regions between them. This paper extends application diffusion to render high quality surface details 3D objects. The first extension is view dependent warping technique that dynamically reallocates texture space so object...

In this paper, we introduce a novel scene representation for the visualization of large-scale point clouds accompanied by set high-resolution photographs. Many real-world applications deal with very densely sampled point-cloud data, which are augmented photographs that often reveal lighting variations and inaccuracies in registration. Consequently, high-quality captured i.e., both together, is challenging time-consuming task. We propose two-phase approach, first (preprocessing) phase...

This paper presents a new approach to generate textured depth meshes (TDMs), an impostor-based scene representation that can be used accelerate the rendering of static polygonal models. The TDMs are precalculated for fixed viewing region (view cell).The relies on layered produce voxel-based representation. Secondary, highly complex polygon mesh is constructed covers all voxels. Afterwards, this simplified using special error metric ensure voxels stay covered. Finally, remaining polygons...

Abstract This paper generalizes the well‐known Diffusion Curves Images (DCI), which are composed of a set Bezier curves with colors specified on either side. These diffused as Laplace functions over image domain, results in smooth color gradients interrupted by curves. Our new formulation allows for more control away from boundary, providing similar expressive power recent Bilaplace models without introducing associated issues and computational costs. The model is based special function...