Joe D. Warren

The program dependence graph and its use in optimization

In this paper we present an intermediate program representation, called the program dependence graph (PDG), that makes explicit both the data and control dependences for each operation in a program. Data dependences have been used to represent only the relevant data flow relationships of a program. Control dependences are introduced to analogously represent only the essential control flow relationships of a program. Control dependences are derived from the usual control flow graph. Many traditional optimizations operate more efficiently on the PDG. Since dependences in the PDG connect computationally related parts of the program, a single walk of these dependences is sufficient to perform many optimizations. The PDG allows transformations such as vectorization, that previously required special treatment of control dependence, to be performed in a manner that is uniform for both control and data dependences. Program transformations that require interaction of the two dependence types can also be easily handled with our representation. As an example, an incremental approach to modifying data dependences resulting from branch deletion or loop unrolling is introduced. The PDG supports incremental optimization, permitting transformations to be triggered by one another and applied only to affected dependences.

Conversion of control dependence to data dependence

Program analysis methods, especially those which support automatic vectorization, are based on the concept of interstatement dependence where a dependence holds between two statements when one of the statements computes values needed by the other. Powerful program transformation systems that convert sequential programs to a form more suitable for vector or parallel machines have been developed using this concept [AllK 82, KKLW 80].The dependence analysis in these systems is based on data dependence. In the presence of complex control flow, data dependence is not sufficient to transform programs because of the introduction of control dependences. A control dependence exists between two statements when the execution of one statement can prevent the execution of the other. Control dependences do not fit conveniently into dependence-based program translators.One solution is to convert all control dependences to data dependences by eliminating goto statements and introducing logical variables to control the execution of statements in the program. In this scheme, action statements are converted to IF statements. The variables in the conditional expression of an IF statement can be viewed as inputs to the statement being controlled. The result is that control dependences between statements become explicit data dependences expressed through the definitions and uses of the controlling logical variables.This paper presents a method for systematically converting control dependences to data dependences in this fashion. The algorithms presented here have been implemented in PFC, an experimental vectorizer written at Rice University.

Mean value coordinates for closed triangular meshes

Constructing a function that interpolates a set of values defined at vertices of a mesh is a fundamental operation in computer graphics. Such an interpolant has many uses in applications such as shading, parameterization and deformation. For closed polygons, mean value coordinates have been proven to be an excellent method for constructing such an interpolant. In this paper, we generalize mean value coordinates from closed 2D polygons to closed triangular meshes. Given such a mesh P, we show that these coordinates are continuous everywhere and smooth on the interior of P. The coordinates are linear on the triangles of P and can reproduce linear functions on the interior of P. To illustrate their usefulness, we conclude by considering several interesting applications including constructing volumetric textures and surface deformation.

Image deformation using moving least squares

We provide an image deformation method based on Moving Least Squares using various classes of linear functions including affine, similarity and rigid transformations. These deformations are realistic and give the user the impression of manipulating real-world objects. We also allow the user to specify the deformations using either sets of points or line segments, the later useful for controlling curves and profiles present in the image. For each of these techniques, we provide simple closed-form solutions that yield fast deformations, which can be performed in real-time.

Barycentric coordinates for convex sets

In this paper we provide an extension of barycentric coordinates from simplices to arbitrary convex sets. Barycentric coordinates over convex 2D polygons have found numerous applications in various fields as they allow smooth interpolation of data located on vertices. However, no explicit formulation valid for arbitrary convex polytopes has been proposed to extend this interpolation in higher dimensions. Moreover, there has been no attempt to extend these functions into the continuous domain, where barycentric coordinates are related to Green’s functions and construct functions that satisfy a boundary value problem. First, we review the properties and construction of barycentric coordinates in the discrete domain for convex polytopes. Next, we show how these concepts extend into the continuous domain to yield barycentric coordinates for continuous functions. We then provide a proof that our functions satisfy all the desirable properties of barycentric coordinates in arbitrary dimensions. Finally, we provide an example of constructing such barycentric functions over regions bounded by parametric curves and show how they can be used to perform freeform deformations.

Barycentric Coordinates for Convex Polytopes

An extension of the standard barycentric coordinate functions for simplices to arbitrary convex polytopes is described. The key to this extension is the construction, for a given convex polytope, of a unique polynomial associated with that polytope. This polynomial, theadjoint of the polytope, generalizes a previous two-dimensional construction described by Wachspress. The barycentric coordinate functions for the polytope are rational combinations of adjoints of various dual cones associated with the polytope.

Blending algebraic surfaces

A new definition of geometric continuity for implicitly defined surfaces is introduced. Under this definition, it is shown that algebraic blending surfaces (surfaces that smoothly join two or more surfaces) have a very specific form. In particular, any polynomial whose zero set blends the zero sets of several other polynomials is always expressible as a simple combination of these polynomials. Using this result, new methods for blending several algebraic surfaces simultaneously are derived.

Dual marching cubes: primal contouring of dual grids

We present a method for contouring an implicit function using a grid topologically dual to structured grids such as octrees. By aligning the vertices of the dual grid with the features of the implicit function, we are able to reproduce thin features of the extracted surface without excessive subdivision required by methods such as marching cubes or dual contouring. Dual marching cubes produces a crack-free, adaptive polygonalization of the surface that reproduces sharp features. Our approach maintains the advantage of using structured grids for operations such as CSG while being able to conform to the relevant features of the implicit function yielding much sparser polygonalizations than has been possible using structured grids.

Smooth geometry images

Previous parametric representations of smooth genus-zero surfaces require a collection of abutting patches (e.g. plines, NURBS, recursively subdivided polygons). We introduce a simple construction for these surfaces using a single uniform bi-cubic B-spline. Due to its tensor-product structure, the spline control points are conveniently stored as a geometry image with simple boundary symmetries. The bicubic surface is evaluated using subdivision, and the regular structure of the geometry image makes this computation ideally suited for graphics hardware. Specifically, we let the fragment shader pipeline perform subdivision by applying a sequence of masks (splitting, averaging, limit, and tangent) uniformly to the geometry image. We then extend this scheme to provide smooth level-of-detail transitions from a subsampled base octahedron all the way to a finely subdivided, smooth model. Finally, we show how the framework easily supports scalar displacement mapping.

A subdivision scheme for surfaces of revolution

This paper describes a simple and efficient non-stationary subdivision scheme of order 4. This curve scheme unifies known subdivision rules for cubic B-splines, splines-in-tension and a certain class of trigonometric splines capable of reproducing circles. The curves generated by this unified subdivision scheme are C^2 splines whose segments are either polynomial, hyperbolic or trigonometric functions, depending on a single tension parameter. This curve scheme easily generalizes to a surface scheme over quadrilateral meshes. The authors hypothesize that this surface scheme produces limit surfaces that are C^2 continuous everywhere except at extraordinary vertices where the surfaces are C^1 continuous. In the particular case where the tension parameters are all set to 1, the scheme reproduces a variant of the Catmull-Clark subdivision scheme. As an application, this scheme is used to generate surfaces of revolution from a given profile curve.