This articel presents a new approach to interaction with 3D objects of virtual models in the context of digital libraries, which was developed as part of the author`s ongoing PhD research. Collaborative networked environments support users in their work with shared content. One way to provide the visual feedback a user needs to interact with 3D models is to distribute image streams of rendered objects in a client-server-environment, which is described here. Image streams are coded in an object-based way according to the MPEG-4 international standard. As a result, users in a heterogeneous and error-prone network environment can cooperatively visualize complex 3D models. The use of coded video streams for collaborative visualization offers flexible means of user interaction for digital library applications. Therefore, future work should aim at providing a cooperative 3D visual interface for heterogeneous documents in digital library systems.