Christopher DeCoro

Real-time mesh simplification using the GPU

Recent advances in real-time rendering have allowed the GPU implementation of traditionally CPU-restricted algorithms, often with performance increases of an order of magnitude or greater. Such gains are achieved by leveraging the large-scale parallelism of the GPU towards applications that are well-suited for these streaming architectures. By contrast, mesh simplification has traditionally been viewed as a non-interactive process not readily amenable to GPU acceleration. We demonstrate how it becomes practical for real-time use through our method, and that the use of the GPU even for offline simplification leads to significant increases in performance. Our approach for mesh decimation adopts a vertex-clustering method to the GPU by taking advantage of a new addition to the rendering pipeline - the geometry shader stage. We present a novel general-purpose data structure designed for streaming architectures called the probabilistic octree, which allows for much of the flexibility of offline implementations, including sparse encoding and variable level-of-detail. We demonstrate successful use of this data structure in our GPU implementation of mesh simplification. We can generate adaptive levels of detail by applying non-linear warping functions to the cluster map in order to improve resulting simplification quality. Our GPU-accelerated approach enables simultaneous construction of multiple levels of detail and out-of-core simplification of extremely large polygonal meshes.

Pose-independent simplification of articulated meshes

Methods for triangle mesh decimation are common; however, most existing techniques operate only on static geometry. In this paper, we present a view- and pose-independent method for the automatic simplification of skeletally articulated meshes. Such meshes have associated kinematic skeletons that are used to control their deformation, with the position of each vertex influenced by a linear combination of bone transformations. Our method extends the commonly-used quadric error metric by incorporating knowledge of potential poses into a probability function. We minimize the average error of the deforming mesh over all possible configurations, weighted by the probability. This is possible by transforming the quadrics from each configuration into a common coordinate system. Our simplification algorithm runs as a preprocess, and the resulting meshes can be seamlessly integrated into existing systems. We demonstrate the effectiveness of this approach for generating highly-simplified models while preserving necessary detail in deforming regions near joints.

Hierarchical Shape Classification Using Bayesian Aggregation

In 3D shape classification scenarios with classes arranged in a hierarchy from most general to most specific, the use of an independent classifier for each class can produce predictions that are inconsistent with the parent-child relationships of the hierarchy. To be consistent, an example shape must not be assigned to a class unless it is also assigned to its parent class. This paper presents a Bayesian framework for combining multiple classifiers based on a class hierarchy. Given a set of independent classifiers for an arbitrary type of shape descriptor, we combine their possibly inconsistent predictions in our Bayesian framework to obtain the most probable consistent set of predictions. Such error correction is expected to improve accuracy on the overall classification by utilizing the structure of the hierarchy. Our experiments show that over the 170-class hierarchical Princeton shape benchmark using the spherical harmonic descriptor (SHD) our algorithm improves the classification accuracy of the majority of classes, in comparison to independent classifiers. Our method is also more effective than straightforward heuristics for correcting hierarchical inconsistencies

Real-Time Isosurface Extraction Using the GPU Programmable Geometry Pipeline

Figure 1. We show the result of extracting a series of highly detailed isosurfaces at interactive rates. Our system implements a hybrid cubes-tetrahedra method, which leverages the strengths of each as applicable to the unique architecture of the GPU. The left pair of images (wireframe and shaded, using a base cube grid of 64) show only an extracted isosurface, while the right pair displays an alternate isosurface overlayed with a volume rendering.

Advanced interactive medical visualization on the GPU

Interactive visual analysis of a patients anatomy by means of computer-generated 3D imagery is crucial for diagnosis, pre-operative planning, and surgical training. The task of visualization is no longer limited to producing images at interactive rates, but also includes the guided extraction of significant features to assist the user in the data exploration process. An effective visualization module has to perform a problem-specific abstraction of the dataset, leading to a more compact and hence more efficient visual representation. Moreover, many medical applications, such as surgical training simulators and pre-operative planning for plastic and reconstructive surgery, require the visualization of datasets that are dynamically modified or even generated by a physics-based simulation engine. In this paper we present a set of approaches that allow interactive exploration of medical datasets in real time. Our method combines direct volume rendering via ray-casting with a novel approach for isosurface extraction and re-use directly on graphics processing units (GPUs) in a single framework. The isosurface extraction technique takes advantage of the recently introduced Microsoft DirectX^(R)10 pipeline for dynamic surface extraction in real time using geometry shaders. This surface is constructed in polygonal form and can be directly used post-extraction for collision detection, rendering, and optimization. The resulting polygonal surface can also be analyzed for geometric properties, such as feature area, volume and size deviation, which is crucial for semi-automatic tumor analysis as used, for example, in colonoscopy. Additionally, we have developed a technique for real-time volume data analysis by providing an interactive user interface for designing material properties for organs in the scanned volume. Combining isosurface with direct volume rendering allows visualization of the surface properties as well as the context of tissues surrounding the region and gives better context for navigation. Our application can be used with CT and MRI scan data, or with a variety of other medical and scientific applications. The techniques we present are general and intuitive to implement and can be used for many other interactive environments and effects, separately or together.

Density-based Outlier Rejection in Monte Carlo Rendering

The problem of noise in Monte‐Carlo rendering arising from estimator variance is well‐known and well‐studied. In this work, we concentrate on identifying individual light paths as outliers that lead to significant spikes of noise and represent a challenge for existing filtering methods. Most noise‐reduction methods, such as importance sampling and stratification, attempt to generate samples that are expected a priori to have lower variance, but do not take into account actual sample values. While these methods are essential to decrease overall noise, we show that filtering samples a posteriori allows for greater reduction of spiked noise. In particular, given evaluated sample values, outliers can be identified and removed. Conforming with conventions in statistics, we emphasize that the term “outlier” should not be taken as synonymous with “incorrect”, but as referring to samples that distort the empirically‐observed distribution of energy relative to the true underlying distribution. By expressing a path distribution in joint image and color space, we show how outliers can be characterized by their density across the set of all nearby paths in this space. We show that removing these outliers leads to significant improvements in rendering quality.

Subtractive Shadows: A Flexible Framework for Shadow Level of Detail

We explore the implications of reversing the process of shadow computation for real-time applications that model complex reflectance and lighting (such as that specified by an environment map). Instead of adding illumination contributions at each pixel across various lights, we compute the complete, unshadowed local illumination at each pixel using approximations, then subtract the lighting contribution from each light for which the pixel is in shadow. This provides flexible level of detail for shadow computation in ways that standard additive shadows do not, such as permitting the use of fast methods for accurate direct illumination combined with a small number of shadow-casting lights, and allowing for downsampled shadows to reduce fill cost. This technique preserves that portion of the scene with the greatest visual importance—the direct illumination—and allows shadows to be presented with lower fidelity in exchange for improvements in speed. With subtractive shadows, we are able to interactively manipulate and render arbitrary BRDFs and environment maps applied to complex, dynamic scenes with shadows, achieving in real time effects that previously required offline computation or preprocessing.