Efficient Path Rendering For Low-End GPUs

Abstract

Modern discrete graphics cards are powerful, and succeed in massively rendering paths for varying domains, one being path-based layouts for automated guided vehicles (AGVs). However, integrated GPUs (iGPUs) are not as powerful as discrete graphics cards, and therefore there is a need for algorithms to accelerate the process of massively rendering paths on iGPUs. This thesis presents a method to massively render paths, more specifically B-spline curves, efficiently and scalably on iGPUs. Instead of rasterizing the polylines, we determine which pixels should be colored based on their intersections with line segments sampled from B-spline curves. We show that our method can massively render paths efficiently and scalably, but only for static layouts. To extend our method to dynamic scenarios, we suggest using a linear quad tree or AABB tree instead of a uniform grid.