This is an early paper on visual programming for 3D graphics. The motivation for the work is to take the idea of a developer's toolkit and expose it to the end users in a visual way. This allows the user to move beyond the dimensions of the application developer's vision of the program's capabilities. They make a great analogy of their system to a sandwich: most applications are finished sandwiches, their application is the means to create a sandwich. The application defined is very similar to many other dataflow like applications developed in this period.
A useful component that they include in their model is the ability to record the output of the user interacting with the system. Isadora has a similar mechanism that is very useful for creating visual compositions. While useful, the mechanism is quite good enough to be used as a sequencing mechanism because in situations where a high degree of accuracy and timing is required, the user is unable to manage modification of all the parameters necessary at once.
Two other notable components are the "watcher" and the "mixer". The watcher allows the user to peak into what is happening in the connection network and is very useful for debugging. In Isadora, this feature is built in to the interface which is even more useful. The mixer allows for the interpolation between any two outputs of like or compatible type. This is useful for constructing transitions between scenes and is an often overlooked or too difficult to manage in an improvisational setting.
An interesting ontology is used in relation to objects and the data structures passed between them. The objects that do some work are called verbs, and the data types are nouns. An interesting observation is that the system is as good as the verbs provided. The beautiful images rendered by the system depend upon techniques developed by the verb makers as well as the content provided by the users. (Another note: The number of types of nouns depends upon what subsets of verbs are capable of handling as input and output.)