Digital video communication has evolved tremendously in the past few years, experiencing significant advances in compression and transmission techniques. To quantify the performance of a video system, it is important to measure the quality of the video. Since humans are the ultimate receivers of a video signal, quality metrics must take into account the properties of the human visual system. So far, most of the metrics that have been proposed require access to the original video, what makes them unsuitable for real-time applications. We investigate how to estimate video quality in real-time applications using no-reference and reduced reference metrics. For this, we study the visibility, annoyance, and relative importance of different types of artifacts and how they combine to produce annoyance. The work uses synthetic artifacts that are simpler, purer, and easier to describe, allowing a high degree of control with respect to the amplitude, distribution, and mixture of different types of artifacts. We present metrics for estimating the strength of four types of artifacts. The outputs of the best artifact metrics are used to build a combination model for overall annoyance.