Abstract | ||
---|---|---|
Inspired by the constant density constraints in position-based fluids (PBF), we propose a novel blue noise sampling algorithm. We formulate and solve a set of sampling points' positional constraints to enforce constant density, which is similar to the PBF. Points converge to distribute evenly and stochastically in space. Fourier spectral analysis of the converged points' distribution shows that it has great blue noise spectrum. By adjusting a single parameter, our method can generate blue noise samplings from Capacity Constrained Voronoi Tessellations (CCVT) to Lloyd's relaxation, and it can also trade off the noise and aliasing of samplings. We utilize the grid-based signed distance field to represent sampling regions, which makes our method fit for general dimensions. Varying the gird size can adjust the number of points. We put points at sampling boundary to solve the points deficiency. Adaptive sampling is achieved with a sampling density function. Experimental results show our method is efficient, stable and controllable in both two-dimensional (2D) plane sampling, and adaptive sampling. Moreover, it is suitable for image stippling. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1145/2949035.2949055 | CGI (Short Papers) |
Field | DocType | Citations |
Slice sampling,Mathematical optimization,Colors of noise,Signed distance function,Adaptive sampling,Stippling,Aliasing,Sampling (statistics),Voronoi diagram,Mathematics | Conference | 0 |
PageRank | References | Authors |
0.34 | 5 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Xiuyu Zheng | 1 | 0 | 0.34 |
Junjun Si | 2 | 0 | 0.34 |
Shuaifu Dai | 3 | 269 | 11.66 |