情報処理学会 第76回全国大会講演要旨

Shadow mapping has been established as the de facto standard technique to render shadows. However, because it relies on looking up values in a depth texture using the discrete z-buffer, it suffers from aliasing effects. Modern applications try to solve aliasing problems for the entire virtual camera frustum. In this paper we propose a technique to adapt shadow mapping to a user’s gaze. Because humans have a narrow field of view in which they can focus, we fit a shadow map to the area on screen that falls into the focus-able area, and thus make better use of the shadow map resolution.