Incremental loading of terrain textures

Abstract

Graphic Processing Units (GPUs) are getting more powerful and are currently capable of displaying millions of polygons at a steady frame rate. The challenge is however, to display unique and interesting data while navigating in a 3D world. This task consists of both selecting which data to display, as well as storing all the required data.This thesis focuses on efficient methods for displaying portions of large terrain textures, by loading the textures incrementally, and thereby avoid some of the storage limitations of current graphics hardware and increase the detail level of the terrain rendered. These methods include algorithms for selecting the parts of the terrain to display, loading the data in time when it's needed and present the data in an appropriate quality while maintaining performance. How the camera behaves plays a key role in these algorithms.The process of displaying large terrains is divided into a multi step process, where coarser algorithms detect the active region of the terrain, then more accurate algorithms calculate which portions to load, and then the most precise algorithms aim to only display the necessary data. Camera movement is also taken into account in this process, and the algorithms attempt to fetch the necessary data in time before it's needed, which is a novel approach. The proposed implementation is compared to related approaches such as Virtual Texturing, a comparable methods which rely on feedback when required data is missing during rendering.