A vertex-first parametric algorithm for polyhedron blending
Introduction
Blending is widely used in geometric modeling for manufacturing, strength, aesthetic and usage purposes. A lot of blending is applied to smooth the vertices and edges of a polyhedral model.
Early in the late 1970s, Doo and Sabin [1] (also Catmull and Clark at the same time [2]) tackled this problem. Being the classical papers on surface subdivision [3], [4], both focused on the polyhedron blending. Their solutions were expressed in recursive formulas for subdivision, rather than blending surface equations. Another numerical method is described by a rolling ball: Choi and Ju [5] put forth the constant-radius rolling ball method, whereas Chuang and Hwang [6] presented an approach of varying the radius of rolling ball (VRRB). Farouki and Sverrisson [7] blended the free-form parametric surfaces with prescribed-precision approximation of constant-radius. Later, researchers aimed to find the exact analytical solutions. According to the blending surface equation forms, the relevant publications can be classified into implicit and parametric. In implicit forms, Braid described various blending for the boundaries of models [8]. Hartmann carried out -blending for vertices by functional splines [9]. Mou et al. presented -blending for convex polyhedra by algebraic splines and provided a free parameter to control the shape of the blending surface [10]. Kosters made use of the quadratic surfaces to blend complex corners [11].
In general, parametric forms of surfaces are more convenient to use. For example, it is trivial to determine points, curves and trimmed patches on a parametric surface. In parametric forms, Szilvasi–Nagy presented a flexible local blending operation for polyhedra [12]. Varady and Rockwood produced vertex blending with elegant setbacks [13]. However, both algorithms blended the edges before the vertex. As a result, smoothly connecting the vertex blending surface with the edge blending surfaces as well as the primary planar faces was quite complicated, since a vertex collects several edges. Great efforts have been made to attain -continuity. At last, they often turned to splitting the vertex blending surface.
Obviously, adopting an inversed, vertex-first strategy will greatly facilitate the matching. In particular, Hartmann already provided an efficient way [14] to blend the edges if the vertices have been blended. When , number of edges meeting at a vertex, equals 3, 4 or more, TB (triangular Bézier) surfaces, TPB (tensor product Bézier) surfaces and regular -patches would be the most promising candidates for the respective vertex blending surfaces. Nevertheless, there still exist large gaps between the candidates and the goal surfaces.
To bridge the gaps, after investigating the schematic control points related to the cross boundary derivatives, by designating the vertex, some edge points and some face points to be the proper control points, we were pleased to turn out a required vertex blending surface, keeping -continuity with the primary planar faces. The correspondence of the control points in schema to the positional control points (not always one to one but sometimes several to one) is the crux of our approach (look ahead at Fig. 6, Fig. 8, Fig. 10, Fig. 11).
In what follows, Section 2 reviews some useful properties of Bézier surfaces and regular -patches. The latter cross boundary derivatives are derived in Section 3 as a preparation for later sections. In Section 4, we construct the vertex blending surface and discuss the influence of their control points on the blending. In Section 5, the edge blending surfaces are generated as the convex combination of two reparameterized local base patches on the constructed vertex blending surfaces. The influence of the trimmed patches on the edge blending surface is also explained. Section 6 gives two practical examples–a die and a monument–to demonstrate the application. Finally, Section 7 concludes the paper.
Section snippets
Preliminaries
To construct the vertex blending surfaces of or 4, we make use of the TB or TPB surfaces. Let us recall some important results of them:
Cross boundary derivatives of -patches
Let us compute the first-order cross boundary derivatives of an -sided regular -patch of depth .
As shown in Fig. 3, is a straight line not parallel to the boundary line in the domain polygon , where and . Substituting into Eq. (9) results in
Actually, it expresses a curve on the -patch. Let , . Note that . Evaluated at , the first-order derivative of
Vertex blending
From now on, if an index ranges from 1 to , then should be replaced by 1 and 0 by .
Problem statement. Let be an -edge vertex with edges . Construct a vertex blending surface of degree in contact with , simultaneously, where denotes a planar face passing through , , .
Remark 1 Control points are the root of TB, TPB and -patches. As a matter of fact, their illustration may have two versions: one is irrelevant to their coordinates and the other
Hartmann method
Hartmann put forward a method for generating a parametric blending surface [14]: the blending surface is simply a convex combination of two reparameterized patches on the base surfaces. Song and Wang extended this method to generate a parametric blending surface based on arbitrary reparameterized partial patches [17]. A brief description of the above method follows:
- (1)
Define a blending function satisfying Hartmann condition, i.e., , its th-order derivatives
Practical examples
Example 5 Common dice for gambling are mainly cubes. A casino needs regular dodecahedron dice for new games. To avoid burr and wear, the vertices and edges should be rounded off. Since each vertex collects 3 edges, quartic TB surfaces are employed to blend the vertices -continuously (Fig. 18(a)). For the edge blending, we take and in Eq. (29). Then it becomes , . The reparameterized local base patches are indicated in Fig. 18(a) as well. The resulting edge blending for a
Conclusion
This paper proposes a new algorithm for polyhedron blending, characterized by:
- •
It yields exact parametric equations of the blending surfaces.
- •
To get rid of matching complication, the blending order is rationalized; edges follow the vertex.
- •
After properly placing the control points on the vertex, the edges and the primary planar faces, the blending process is simply generating an ordinary TB, TPB, or -patch, followed by Hartmann “blending by blending”. Hence it is easy to grasp and use.
- •
It can
Acknowledgments
The authors thank the editor and reviewers sincerely for their pertinent comments. This paper belongs to a project, for which we are applying for funding from National Natural Science Foundation of China.
Pei Zhou received his BS degree in Mechanical and Electrical Engineering in 2006 from the Central South University, Changsha, China. He is now a Ph.D. student at Shanghai Jiao Tong University. As an excellent graduate recommended by the former, he was admitted to the latter without examination. Being engaged in computer aided geometric design, he is focusing on blending at present.
References (18)
- et al.
Behaviour of recursive division surfaces near extraordinary points
Computer-Aided Design
(1978) - et al.
Recursively generated B-spline surfaces on arbitrary topological meshes
Computer-Aided Design
(1978) Subdivision surfaces for CAD—An overview
Computer-Aided Design
(2005)- et al.
Constant-radius blending in surface modeling
Computer-Aided Design
(1989) - et al.
Approximation of rolling-ball blends for free-form parametric surfaces
Computer-Aided Design
(1996) Non-local blending of boundary models
Computer-Aided Design
(1997)Implicit -blending of vertices
Computer Aided Geometric Design
(2001)- et al.
Simultaneous blending of convex polyhedra by algebraic splines
Computer-Aided Design
(2007) Flexible rounding operation for polyhedra
Computer-Aided Design
(1991)
Cited by (4)
Blending Polyhedral Edge Clusters
2019, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)A unified approach to blending of constant and varying parametric surfaces with curvature continuity
2018, ACM International Conference Proceeding SeriesBlending using ODE swept surfaces with shape control and C<sup>1</sup> continuity
2014, Visual ComputerFilling free-form n-sided holes toward blending and decoration
2013, Proceedings - 2013 7th International Conference on Image and Graphics, ICIG 2013
Pei Zhou received his BS degree in Mechanical and Electrical Engineering in 2006 from the Central South University, Changsha, China. He is now a Ph.D. student at Shanghai Jiao Tong University. As an excellent graduate recommended by the former, he was admitted to the latter without examination. Being engaged in computer aided geometric design, he is focusing on blending at present.
Wen-Han Qian is currently a Ph.D. advisor with the Robotics Institute, Shanghai Jiao Tong University. In the joint entrance examination of Peking, Tsinghua, and Nankai universities in 1949, he ranked first in China by total marks. After graduation from Jiao Tong University in 1952, he taught various courses in mechanical engineering, mathematics, and computer science at his Alma Mater, becoming a professor in 1985. Since 1992 he has enjoyed a special monthly subsidy from the State Council of the PRC for his “outstanding contributions to the development of China’s high education”.