The quantitative evaluation of optical flow algorithms by Barron et al. (1994) led to significant advances in performance. The challenges for optical flow algorithms today go beyond the datasets and evaluation methods proposed in that paper. Instead, they center on problems associated with complex natural scenes, including nonrigid motion, real sensor noise, and motion discontinuities. We propose a new set of benchmarks and evaluation methods for the next generation of optical flow algorithms. To that end, we contribute four types of data to test different aspects of optical flow algorithms: (1) sequences with nonrigid motion where the ground-truth flow is determined by tracking hidden fluorescent texture, (2) realistic synthetic sequences, (3) high frame-rate video used to study interpolation error, and (4) modified stereo sequences of static scenes. In addition to the average angular error used by Barron et al., we compute the absolute flow endpoint error, measures for frame interpolation error, improved statistics, and results at motion discontinuities and in textureless regions. In October 2007, we published the performance of several well-known methods on a preliminary version of our data to establish the current state of the art. We also made the data freely available on the web at

We present a technique for capturing an actor’s live-action performance in such a way that the lighting and reflectance of the actor can be designed and modified in postproduction. Our approach is to illuminate the subject with a sequence of time-multiplexed basis lighting conditions, and to record these conditions with a highspeed video camera so that many conditions are recorded in the span of the desired output frame interval. We investigate several lighting bases for representing the sphere of incident illumination using a set of discrete LED light sources, and we estimate and compensate for subject motion using optical flow and image warping based on a set of tracking frames inserted into the lighting basis. To composite the illuminated performance into a new background, we include a time-multiplexed matte within the basis. We also show that the acquired data enables time-varying surface normals, albedo, and ambient occlusion to be estimated, which can be used to transform the actor’s reflectance to produce both subtle and stylistic effects.

When rendering only directly visible objects, ray tracing a few levels of specular reflection from large, lowcurvature surfaces, and ray tracing shadows from point-like light sources, the accessed geometry is coherent and a geometry cache performs well. But in many other cases, the accessed geometry is incoherent and a standard geometry cache performs poorly: ray tracing of specular reflection from highly curved surfaces, tracing rays that are many reflection levels deep, and distribution ray tracing for wide glossy reflection, global illumination, wide soft shadows, and ambient occlusion. Fortunately, less geometric accuracy is necessary in the incoherent cases. This observation can be formalized by looking at the ray differentials for different types of scattering: coherent rays have small differentials, while incoherent rays have large differentials. We utilize this observation to obtain efficient multiresolution caching of geometry and textures (including displacement maps) for classic and distribution ray tracing in complex scenes. We use an existing multiresolution caching scheme (originally developed for scanline rendering) for textures and displacement maps, and introduce a multiresolution geometry caching scheme for tessellated surfaces. The multiresolution geometry caching scheme makes it possible to efficiently render scenes that, if fully tessellated, would use 100 times more memory than the geometry cache size. 1.

Abstract not found

In this paper we present a novel abstraction layer above the well-known triangular 1-to-4 split subdivision based on hexagons. This approach allows adaptive bidirectional subdivision, that is refinement and coarsening of the mesh, without causing problems like irregular refinement or hanging nodes. Thus it is feasible to adapt the mesh resolution locally, according to geometric features, viewing parameters and error bounds. Our method does neither rely on time-consuming preprocessing nor hierarchical data structures, which yields to low memory consumption. We show, how our technique can be applied to terrain rendering of huge data sets, where we achieve interactive frame rates using hexagonal subdivision. 1

This paper presents a type of visibility data structure for accelerated ray tracing. The visibility field is constructed by choosing a regular point subdivision over a hemisphere to obtain a set of directions. Corresponding to each direction there is then a rectangular grid of parallel beams, with each beam referencing a set of identifiers corresponding to objects that intersect it. Objects lying along a beam are sorted using a 1D BSP along the beam direction. The beam corresponding to any ray can be looked up in small constant time and the set of objects corresponding to the beam can then be searched for intersection with the ray using an optimised traversal strategy. This approach trades off rendering speed for memory usage and pre-processing time. The data structure is also very suitable for hemisphere integration tasks due to its spherical nature and results for one such task-Ambient Occlusion- are also presented. Results for several scenes with various rendering methods are presented and compare favourably with a well established approach, the single-ray Coherent Ray Tracing approach of Wald and Slusallek et al. 1.

We present a method for automatically converting a digital 3D model into a multilayer model: a parallel stack of high-resolution 2D images embedded within a semi-transparent medium. Multilayer models can be produced quickly and cheaply and provide a strong sense of an object’s 3D shape and texture over a wide range of viewing directions. Our method is designed to minimize visible cracks and other artifacts that can arise when projecting an input model onto a small number of parallel planes, and avoid layer transitions that cut the model along important surface features. We demonstrate multilayer models fabricated with glass and acrylic tiles using commercially available printers.

Obscurances, from which ambient occlusion is a particular case, is a technology that produces natural-looking lighting effects in a faster way than global illumination. Its application in volume visualization is of special interest since it permits us to generate a high quality rendering at a low cost. In this paper, we propose an obscurancebased framework that allows us to obtain realistic and illustrative volume visualizations in an interactive manner. Obscurances can include color bleeding effects without additional cost. Moreover, we obtain a saliency map from the gradient of obscurances and we show its application to enhance volume visualization and to select the most salient views. Categories and Subject Descriptors (according to ACM CCS):

Monte Carlo ray tracing remains a simple and elegant method for generating robust shadows. This approach, however, is often hampered by the time needed to evaluate the numerous shadow ray queries required to generate a high-quality image. We propose the use of volumetric occluders stored within a kd-tree in order to accelerate shadow rays cast on a closed, watertight mesh. Intersection with a volumetric occluder is much cheaper than intersection with mesh geometry, although performing these intersections requires modification to the traversal order through the kd-tree. We propose two such modifications, both of which enable the use of volumetric occluders for cheap shadow ray termination. We also propose using a software-managed cache to store and reuse volumetric occluders for even earlier termination. Our approach provides a performance improvement of up to 2.0x for our test scenes while producing images identical to those produced by the unaccelerated baseline.

This technical memo describes a fast point-based method for computing diffuse global illumination (color bleeding). The computation is 4–10 times faster than ray tracing, uses less memory, has no noise, and its run-time does not increase due to displacementmapped surfaces, complex shaders, or many complex light sources. These properties make the method suitable for movie production. The input to the method is a point cloud (surfel) representation of the directly illuminated geometry in the scene. The surfels in the point cloud are clustered together in an octree, and the power from each cluster is approximated using spherical harmonics. To compute the indirect illumination at a receiving point, we add the light from all surfels using three degrees of accuracy: ray tracing, singledisk appoximation, and clustering. Huge point clouds are handled by reading the octree nodes and surfels on demand and caching them. Variations of the method efficiently compute area light illumination and soft shadows, final gathering for photon mapping, HDRI environment map illumination, multiple diffuse reflection bounces, ambient occlusion, and glossy reflection. The method has been used in production of more than a dozen feature films, for example for rendering Davy Jones and his crew in two of the “Pirates of the Caribbean ” movies.