An important task in robot vision is that of determining the position, orientation and trajectory of a moving camera relative to an observed object or scene. Many such visual tracking algorithms have been proposed in the computer vision, artificial intelligence and robotics literature over the past 30 years. However, it is seldom possible to explicitly measure the accuracy of these algorithms, since the ground-truth camera positions and orientations at each frame in a video sequence are not available for comparison with the outputs of the proposed vision systems. A method is presented for generating real visual test data with complete underlying ground truth. The method enables the production of long video sequences, filmed along complicated six-degree-of-freedom trajectories, featuring a variety of objects and scenes, for which complete ground-truth data are known including the camera position and orientation at every image frame, intrinsic camera calibration data, a lens distortion model and models of the viewed objects. This work encounters a fundamental measurement problem—how to evaluate the accuracy of measured ground truth data, which is itself intended for validation of other estimated data. Several approaches for reasoning about these accuracies are described.