A multimedia presentation model provides designers a tool to formally specify the temporal and spatial relationships of objects. The formality helps designers to communicate with others, to check the integrity of designs, and provides a chance to simulate the designs. Although much research has been devoted to this subject, to the best of our knowledge, no multimedia models are able to describe the spatial-temporal relations of moving objects that may refer to each other for computing displaying addresses. The addresses may be recomputed several times during the objects' lifetimes to reflect their movements. Without a formal model, designers are forced to specify the relationships in an ad hoc manner that causes misunderstanding and hampers integrity check. The check includes if an object gets its addresses in time from another object, if an object is displayed in the right places, etc. The difficulty of designing such a formal model lies in integrating temporal constraints of objects with a real-time address transferring mechanism. In this paper, we present an extended Petri-net model, which models concurrent relationships of objects with new places, transitions, and firing rules to transfer and transform addresses in real time. Its descriptive power and correctness is demonstrated by five patterns of multimedia presentations and a sample play scripts.