The performance of video-based scene analysis algorithms often suffers because of the inability to effectively acquire features on the targets. In this paper, we propose a distributed approach for dynamically controlling the pan, tilt, zoom (PTZ) parameters of a PTZ camera network so as to maximize system performance, through opportunistic acquisition of high quality images. The cameras gain utility by achieving the tracking specification and through high resolution feature acquisition. High resolution imagery comes at a higher risk of losing the target in a dynamic environment due to the corresponding decrease in the field of view (FOV). This optimization will determine not only how the cameras are controlled, but also when to obtain high quality images. The target state estimates, upon which the control algorithm is dependent, are obtained through a distributed tracking algorithm. Our approach is developed within a Bayesian framework to appropriately trade-off value (target tracking accuracy and image quality) versus risk (probability of losing track of a target). This article presents the theoretical solution along with simulation and experimental results on a real camera network.