Publications

We present a new method for approximating subdivision surfaces with hardware accelerated parametric patches. Our method improves the memory bandwidth requirements for patch control points, translating into superior performance compared to existing methods. Our input is general, allowing for meshes that contain both quadrilateral and triangular faces in the input control mesh, as well as control meshes with boundary. We present two implementations of our scheme designed to run on Direct3D 11 class hardware equipped with a tessellator unit.

The goal of this course is to familiarize attendees with the practical aspects of subdivision surfaces for which we introduce substitutes for increased efficiency in real-time applications. The course starts by highlighting the properties that make SubD modeling attractive and introduces some recent techniques to capture these properties by alternative surface representations with a smaller foot-print. We list and compare the new surface representations and focus on their implementation on current and next-generation GPUs. Among the advantages and disadvantages of each approach, we address crucial practical issues, such as watertight evaluation, creases and corners, and seamless displacement mapping. Finally and most importantly, Valve and Industrial Light Magic will present a few breathtaking practical examples and demonstrate how these advanced techniques have been adopted into their gaming and movie production pipelines.

For use in real-time applications, we present a fast algorithm for converting a quad mesh to a smooth, piecewise polynomial surface on the Graphics Processing Unit (GPU). The surface has well-defined normals everywhere and closely mimics the shape of CatmullClark subdivision surfaces. It consists of bicubic splines wherever possible, and a new class of patchesc-patcheswhere a vertex has a valence different from 4. The algorithm fits well into parallel streams so that meshes with 12,000 input quads, of which 60% have one or more non-4-valent vertices, are converted, evaluated and rendered with 9 x 9 resolution per quad at 50 frames per second. The GPU computations are ordered so that evaluation avoids pixel dropout.

Polyhedral meshes consisting of triangles, quads, and pentagons and polar configurations cover all major sampling and modeling scenarios. We give an algorithm for efficient local, parallel conversion of such meshes to an everywhere smooth surface consisting of low-degree polynomial pieces. Quadrilateral facets with 4-valent vertices are ‘regular’ and are mapped to bi-cubic patches so that adjacent bi-cubics join C² as for cubic tensor-product splines. The algorithm can be implemented in the vertex and geometry shaders of the GPU pipeline and does not use the fragment shader. Its implementation in DirectX 10 achieves conversion plus rendering at 659 frames per second with 42.5 million triangles per second on input of a model of 1,300 facets of which 60% are not regular. The work was supported in part by NSF grant CCF-0728797

We present a fast algorithm for converting quad meshes on the GPU to smooth surfaces. Meshes with 12,000 input quads, of which 60% have one or more non-4-valent vertices, are converted, evaluated and rendered with 9×9 resolution per quad at 50 frames per second. The conversion reproduces bi-cubic splines wherever possible and closely mimics the shape of the Catmull-Clark subdivision surface by c-patches where a vertex has a valence different from 4. The smooth surface is piecewise polynomial and has well-defined normals everywhere. The evaluation avoids pixel dropout. The work was supported in part by NSF grant CCF-0728797.

A well-documented problem of Catmull and Clark subdivision surfaces is that, in the neighborhood of extraordinary points, the curvature is unbounded and fluctuates. In fact, since one of the eigenvalues that determines elliptic shape is too small, the limit surface can have a saddle point when the designer's input mesh suggests a convex shape. Here, we replace, near the extraordinary point, Catmull–Clark subdivision by another set of rules based on refining each bi-cubic B-spline into nine. This provides many localized degrees of freedom for special rules that need not reach out to neighbor vertices in order to tune the behavior. In this paper, we provide a strategy for setting such degrees of freedom and exhibit tuned ternary quad subdivision that yields surfaces with bounded curvature, nonnegative weights and full contribution of elliptic and hyperbolic shape components.

Good surgical training depends greatly on case experiences that have been difficult to model in software since current training technology does not provide the flexibility to teach and practice uncommon procedures, or to adjust a training scenario on the fly. The TIPS kit aims to overcome these limitations. To the expert, it presents visual and haptic tools that make illustrating procedures easy and can model unusual anatomic variations. For a non-specialist, it provides a locally customized learning environment and repeated practice in a safe environment. We used the toolkit to illustrate removal of the adrenal gland.

Systems biology research demands the ability to construct custom-designed bio-entities from heterogeneous data sources. EDIT is an ongoing project aimed at building a system to allow bio-researchers with minimum computer proficiency to define the structure of a desired bio-entity, and identify the data sources for its data fields. Then, the system automatically extracts and output the data in the predefined structure and format. The system consists of a host site, which maintains and updates the configuration files specified for source databases, and user sites, which perform data extraction and integration by using the configuration files. An event-trigger-rule server is used to support data refreshing and to notify biologists about changes made to source databases

Motivation: In the post-genomic era, biologists interested in systems biology often need to import data from public databases and construct their own systemspecific or subject-oriented databases to support their complex analysis and knowledge discovery. To facilitate the analysis and data processing, customized and centralized databases are often created by extracting and integrating heterogeneous data retrieved from public databases. A generalized methodology for accessing, extracting, transforming and integrating the heterogeneous data is needed. Results: This paper presents a new data integration approach named JXP4BIGI (Java XML Page for Biological Information Gathering and Integration). The approach provides a system-independent framework, which generalizes and streamlines the steps of accessing, extracting, transforming and integrating the data retrieved from heterogeneous data sources to build a customized data warehouse. It allows the data integrator of a biological database to define the desired bio-entities in XML templates (or Java XML pages), and use embedded extended SQL statements to extract structured, semistructured and unstructured data from public databases. By running the templates in the JXP4BIGI framework and using a number of generalized wrappers, the required data from public databases can be efficiently extracted and integrated to construct the bio-entities in the XML format without having to hard-code the extraction logics for different data sources. The constructed XML bio-entities can then be imported into either a relational database system or a native XML database system to build a biological data warehouse.