Contents reuse is a key element for studios competitiveness.  For each shot, each prop, one can either start from scratch or reuse data that already existe somewhere to save time. Very often, these data are stored in a data bank, some sort of vault that contains the studio know-how. In order to do this, it is important to be able to find data very fast, and therefore to define good search criteria. Otherwise, the expected gain would become a waste of time loading inadequate data.

Assigning tags to objects is the classical mechanism, and this is our starting point. But we quickly face the problem of organizing these tags in categories, within an ontology, and that this structure may evolve with time. Moreover, when assigning tags, an artist does not always know if these tags will be the ones used later as search criteria (and studies prove this).

It is therefore necessary to add a mechanism to retrieve data according to their contents (color, texture, shape, behaviors, …etc). Hence the need to analyze the contents and associate appropriate descriptors. This is where the ISO / MPEG-7 norm is going.

Besides, we continue research in the direction of broadening the semantic recovery from content, a dedicated ontology and the unification of descriptors (RDF, microformats, …etc), to build better tools .