Measuring and Enhancing the Legibility of GPU-rendered Text

Abstract

Whereas the rendering of tiny typefaces in 2D applications has been perfected over decades, the legibility of text in 3D visualizations has rarely been addressed. This affects road signs, meters, screens, and books in virtual reality applications ranging from car driving simulators to digital lecture halls. To improve text rendering and display, we devised a lightweight psychovisual test to measure one prominent aspect of legibility. We subjected promising GPU-based methods for crisp minification to this test. It turned out that legibility may be improved without undue costs. Not all techniques whose results look appealing on first sight will, however, enhance legibility.