GauWN: Gaussian-smoothed Winding Number and Its Derivatives

Haoran Sun,Jingkai Wang,Hujun Bao,Jin Huang

DOI: https://doi.org/10.1145/3680528.3687569

2024-01-01

Abstract:For a fixed polygon, one can easily determine whether a point is inside or outside it using the winding number. However, deforming a given polygon based on a set of points with expected inside/outside labeling is much more difficult. It asks the winding number to be differentiable with respect to locations of the inside/outside test point and the polygon vertices. We propose a method to address this even for a possibly intersected 2D polygon through Gaussian kernel convolution. Our method can be applied to various problems such as resolving embedding issues (e.g., intersections), editing curves using an in-out brush, and offsetting curves with feature preservation. It may seem difficult to compute the value and derivatives of this smoothed winding number (GauWN) efficiently, but the cost is only 4 to 6 times that of the vanilla one. To achieve this efficiency, we employ two key strategies: 1) For value computation, we extend the divergence theorem to handle self-intersected cases and transform the convolution into a line integral that can be computed efficiently. 2) For derivatives, we utilize local decomposition to find a line integral form and leverage the radial symmetry and orthogonal separability of the Gaussian kernel. With this differentiable winding number, we can solve the aforementioned problems efficiently by formulating them to involve both the explicit boundary and its implicit field. Surprisingly, there is no need to create a background mesh despite the involvement of an implicit field, making our method easy to apply.