Publications by authors named "Ping-Man Lam"
Article Synopsis
- Cube mapping is popular in graphics applications because it benefits from hardware support, but it doesn't evenly sample spherical surfaces, leading to unevenness.
- Recent developments, like isocube mapping, improve this by ensuring even sampling but introduce issues with nonrectilinear texels that can hurt filtering quality.
- The newly proposed unicube mapping combines the strengths of both cube and isocube mapping by using a simple modification in the lookup process, resulting in better real-time texture filtering and easier dynamic environment mapping.
All-frequency lighting with multiscale spherical radial basis functions.
IEEE Trans Vis Comput Graph
March 2010
Article Synopsis
- This paper introduces a new method called multiscale spherical radial basis function (MSRBF) representation for simulating all-frequency lighting, which can cover both distant and local light sources.
- The method organizes spherical radial basis functions (SRBFs) into multiple levels based on their size, allowing larger functions to handle lower frequency lighting while smaller functions manage higher frequencies.
- The MSRBF approach enhances rendering efficiency, allows for scalable quality adjustments, and demonstrates practical use in real-time lighting and data compression.
Efficient relighting of RBF-based illumination adjustable images.
IEEE Trans Neural Netw
December 2009
Article Synopsis
- * It highlights the radial basis function (RBF) as a more efficient method than spherical harmonics for IAI representation, but notes that high-frequency lighting effects require numerous RBFs, which can slow down relighting.
- * The investigation proposes a partial reconstruction scheme that utilizes the locality of RBFs, resulting in significantly faster relighting speeds while maintaining similar levels of distortion compared to traditional RBF and spherical harmonic methods.
In image-based relighting, a pixel is associated with a number of sampled radiance values. This paper presents a two-level compression method. In the first level, the plenoptic property of a pixel is approximated by a spherical radial basis function (SRBF) network.
View Article and Find Full Text PDFSpherical harmonic (SH) basis functions have been widely used for representing spherical functions in modeling various illumination properties. They can compactly represent low-frequency spherical functions. However, when the unconstrained least square method is used for estimating the SH coefficients of a hemispherical function, the magnitude of these SH coefficients could be very large.
View Article and Find Full Text PDFThe global extended Kalman filtering (EKF) algorithm for recurrent neural networks (RNNs) is plagued by the drawback of high computational cost and storage requirement. In this paper, we present a local EKF training-pruning approach that can solve this problem. In particular, the by-products, obtained along with the local EKF training, can be utilized to measure the importance of the network weights.
View Article and Find Full Text PDF