Accurately Measuring the Satisfaction of Visual Properties in Virtual Camera Control

Declarative approaches to camera control model inputs as properties on the camera and then rely on constraint-based and/or optimization techniques to compute the camera parameters or paths that best satisfy those properties. To reach acceptable performances, such approaches often (if not always) compute properties satisfaction in an approximate way. Therefore, it is difficult to measure results in terms of accuracy, and also compare approaches that use different approximations. In this paper, we propose a simple language which can be used to express most of the properties proposed in the literature and whose semantics provide a way to accurately measure their satisfaction. The language can be used for several purposes, for example to measure how accurate a specific approach is and to compare two distinct approaches in terms of accuracy.