GPU curvature estimation on deformable meshes

Abstract

Figure 1: On the left is one frame of an elephant animation visualizing estimated curvature. Blue indicates concave areas, red indicates convex areas, and green indicates saddle-shape areas. The middle is using surface curvature to approximate ambient occlusion. On the right is an orange volume with ∼1.5 billion vertices with Lambertian shading and ambient occlusion which we estimate curvature on in 39.3 ms. AbstractSurface curvature is used in a number of areas in computer graphics, including texture synthesis and shape representation, mesh simplification, surface modeling, and non-photorealistic line drawing. Most real-time applications must estimate curvature on a triangular mesh. This estimation has been limited to CPU algorithms, forcing object geometry to reside in main memory. However, as more computational work is done directly on the GPU, it is increasingly common for object geometry to exist only in GPU memory. Examples include vertex skinned animations and isosurfaces from GPU-based surface reconstruction algorithms.For static models, curvature can be pre-computed and CPU algorithms are a reasonable choice. For deforming models where the geometry only resides on the GPU, transferring the deformed mesh back to the CPU limits performance. We introduce a GPU algorithm for estimating curvature in real-time on arbitrary triangular meshes. We demonstrate our algorithm with curvature-based NPR feature lines and a curvature-based approximation for ambient occlusion. We show curvature computation on volumetric datasets with a GPU isosurface extraction algorithm and vertex-skinned animations. Our curvature estimation is up to ∼18x faster than a multithreaded CPU benchmark.

Curvature-Approximated Estimation of Real-Time Ambient Occlusion

“…It has already been shown that screen space estimation is easy to implement in computer games on current graphics hardware. Wesley et al (Griffin et al, 2011) introduced real time curvature estimation in the world space. They employed a programmable geometry shader to detect the normal curvature by analyzing the normal and position vectors of neighboring polygons.…”

Joint pose and principal curvature refinement using quadrics

“…This idea was extended in [20], where Guillaume et al apply this work to mesh based segmentation. However, these previous approaches either only applied to a single model [21] [2], or rely on the alignment of multiple scan-matches to highly accurate in order to build a single shape model for curvature estimation. Additionally, these previous approaches often assume relatively low-noise in the model, which is not the case when using low-cost depth sensors.…”

A Parametric Space Approach to the Computation of Multi-scale Geometric Features

“…Most feature computation methods belongs to this category and thus operate on the CPU. GPU-based methods, have been proposed for the computation of specific features, like curvature (Griffin et al, 2011), but these methods do not generalize to the sampling of arbitrary neighborhoods. Screen space methods sample the geometric information of a mesh from a 2D pixel buffer, where each pixel encodes the projected surface position of the mesh from a specific point of view.…”

“…massively parallel stream processors, like commodity GPUs, at an arbitrary neighborhood scale. For these reasons, real-time computation is often limited to meshes with relatively low geometric complexity and 1-ring vertex neighborhoods (Griffin et al, 2011).…”